DETAILED ACTION
This action is responsive to the application filed on 08/29/2026. Claims 16-24 and 26-32 are pending and have been examined. This action is Final.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Priority
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C.
120, 121, 365(c), or 386(c) is acknowledged.
Response to Arguments
Argument 1: The applicant argues that the amended independent claims are not directed to an abstract idea under the first prong of Alice/Mayo, relying mainly on PowerBlock Holdings, Inc. v. iFit, Inc., 146 F.4th 1366 (Fed. Cir. 2025). They say each claim is “limited to a specific implementation of a technological improvement.” For claim 16, the improvement is recommending a response chat message that matches the personality of the AI character the user selected, citing the “caring type” example at 93. For claim 21, it is training the conversation engine in the direction the creator wants, through a user interface that lets the creator modify answers, citing paragraph 110-111. For claim 28, it is creating an AI character whose speech tone and topics of interest match the user’s preferences, through a two-window interface with a character introduction window and a matter-of-interest window, citing 222. The applicant also argues that the newly added “graphically displaying and rendering” limitations (the virtual chatroom and chat message in claim 16, the AI character and editing interface in claim 21, and the two windows in claim 28) are not mental processes or methods of organizing human activity, citing 2019 Example 37. In the alternative, they argue these features give “significantly more” and integrate any abstract idea into a practical application by improving the AI conversation system, citing 2024 Example 47 (claim 3) and Example 48 (claim 2). The dependent claims are argued only as eligible for the same reasons as their parent claims.
Response to Argument 1: The examiner has considered the arguments made by the applicant above, however they are not persuasive. Unlike the claims in PowerBlock, which were directed to a specific improvement in the mechanical operation of a physical weight-selection apparatus, the amended claims recite selecting a character based on its personality and determining a response associated with that personality (claim 16), allowing a creator to modify answers and using the modified answers to train toward the creator’s preference (claim 21), and determining a speech tone and topics of interest from introduction contents and a tag topic (claim 28), each of which can be performed in the human mind or is a way of managing interpersonal communication, as set forth in the rejection above. The asserted benefits, namely responses that better suit a selected personality, a conversation engine trained toward a creator’s preferences, and characters that better match a user’s interests, are improvements to the content produced by the abstract idea itself rather than to the functioning of a computer or another technology (see MPEP 2106.05(a)(II)). The newly added “graphically displaying and rendering” limitations are additional elements evaluated under Step 2A Prong 2 and Step 2B, and unfortunately their presence does not remove the abstract idea recited by the remaining limitations, and they recite only a generic electronic device and graphical interface that display a chatroom, chat messages, and input windows, which amounts to mere instructions to apply the exception and insignificant extra-solution activity (see MPEP 2106.05(f) and (g)). Example 37 is not analogous because its eligible claim improved the graphical user interface itself by automatically arranging icons based on usage, whereas the instant claims merely use a graphical interface to present conversation content and receive input. Likewise, Example 47, claim 3, and Example 48, claim 2, recited specific technical operations that improved a technology, such as remediating detected malicious network traffic and separating mixed speech signals, whereas the instant claims recite training and response generation only as a result-oriented, high level of generality, which applies generic machine learning to a new data environment. See Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205 (Fed. Cir. 2025). Thus, the claims do not integrate the abstract idea into a practical application or amount to significantly more, and the dependent claims, which were not separately argued, remain rejected for the reasons set forth above.
Argument 2: The applicant argues that the amendments overcome the obviousness rejections because none of the cited references teach the newly added limitations, and offers little analysis beyond that. For claim 16, they note the Office conceded that Hancock and Schmitt do not teach determining response data from the recommendation data. They then argue that Kannan does not teach selecting the AI character based on its personality, associating the response with that personality, or displaying the response as a chat message in a virtual chatroom. Kannan is argued alone, without addressing Schmitt. For claim 21, they argue only against Jacob: they say Jacob does not teach a user interface for the creator to modify answer data, or training the engine in the direction the creator wants by including the modified answers. They do not address Xiaojiang, which the action already applied to the dependent claims. For claim 28, they argue that neither Schmitt nor Li teaches a displayed user interface with both a character introduction window and a matter-of-interest window. They also argue that neither reference determines speech tone and conversation contents from both the introduction contents and the tag topic, although the claim requires only “at least one of” them. Claim 25 is cancelled. The remaining dependent claims (17-20, 22-24, 26-27 and 29-30) and the new claims 31 and 32 are argued only through their parent independent claims, with no separate arguments.
Response to Argument 2: The examiner has considered the arguments made by the applicant above, however they are not persuasive, and to the extent they are directed to the newly added limitations they are addressed by the new grounds of rejection necessitated by amendment. Regarding claim 16, The applicant argues Kannan individually, but one cannot show nonobviousness by attacking references individually where the rejection is based on a combination of references (see In re Keller, 642 F.2d 413 (CCPA 1981); In re Merck & Co., 800 F.2d 1091 (Fed. Cir. 1986)). Schmitt displays the conversation in the chat view of its web app interface and generates text based on the chat history and the chatbot’s “personality,” and Foster renders available personas, receives a selection of a persona, and produces responses “consistent with the persona,” which teaches selecting the AI character based on its personality and associating the response data with that personality. Regarding claim 21, Applicant addresses only Reddy (US 20140032471A1, referred to in the prior action as “Jacob,” its second-named inventor), while Xiaojiang, already of record, teaches a chat log through which the merchant replaces an unsatisfying reply with a better reply, the chatbot thereafter answering customers with the merchant’s reply, and the merchant training the chatbot with those replies. Regarding claim 28, Schmitt’s interface includes a character view that shows the attributes the user defines, Reddy’s script editor includes a separate category list through which a character’s content is assigned to topics such as games and stories, and Li determines speaking style from persona background information. Furthermore, the applicant’s argument that the speech tone and conversation contents are determined from both the introduction contents and the tag topic is not commensurate with the scope of the claim, which requires only “at least one of” them. Amended claim 29, new claims 31 and 32, and the remaining dependent claims are not separately argued and are rejected as set forth above.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition
of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the
conditions and requirements of this title.
Claims 16-24 and 26-32 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 16,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
Step 2A Prong 1:
“maintaining a conversation session…between a user and at least one conversation partner” -- This limitation is directed to organizing and managing a communication interaction between a user and at least one conversation partner. Managing a conversation session between parties falls within certain methods of organizing human activity, such as managing personal behavior or relationships or interactions between people, including social activities and interpersonal communication. Therefore, this limitation recites a mental process.
“generating recommendation data for a response…in response to receiving conversation data from the at least one conversation partner, wherein the AI character is selected based on a personality of the AI character” -- This limitation is directed to determining or suggesting how to respond to conversational content using a character chosen based on its personality. Under the broadest reasonable interpretation, a person can read a received message, choose a persona or character based on the personality the person wants the reply to reflect, and consider what response should be provided in that character’s voice. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process. This limitation also manages interpersonal communication and therefore also falls within certain methods of organizing human activity.
“determining response data using the recommendation data, the response data being associated with the personality of the selected AI character” -- This limitation is directed to selecting a response from recommended responses where the response reflects the personality of the selected character. A person can review suggested replies and decide which reply to send, choosing wording that is consistent with a particular personality, such as a writer composing a reply in a character’s voice. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“An artificial intelligence (AI) conversation engine training method” -- The preamble merely identifies the technological environment in which the abstract idea is performed. The body of the claim does not recite any specific training operation, and identifying the method as an AI conversation engine training method amounts to no more than generally linking the use of the judicial exception to a particular technological environment or field of use, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(h)).
“graphically displaying and rendering, by an electronic device, a virtual chatroom on a graphical interface of the electronic device” -- This limitation is directed to displaying a chatroom on a generic electronic device and graphical interface. The electronic device and graphical interface are recited at a high level of generality and are used merely as tools to present the conversation in which the abstract idea is performed. The claim does not recite any improvement to the graphical interface itself, to the way the electronic device renders information, or to computer functionality. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component (see MPEP 2106.05(f)), and displaying information is insignificant extra-solution activity, which cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, presenting or displaying data on a generic computer interface is a well-understood, routine, and conventional activity (WURC) that does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“graphically displayed and rendered in the virtual chatroom” -- This limitation is directed to displaying the conversation session in the virtual chatroom. Displaying the conversation is the presentation of information on a generic graphical interface and does not recite an improvement to the interface or to computer functionality. Therefore, this limitation amounts to mere instructions to apply the judicial exception using a generic computer component (see MPEP 2106.05(f)) and insignificant extra-solution activity (see MPEP 2106.05(g)), which cannot integrate the judicial exception into a practical application. Furthermore, under Step 2B, displaying data on a generic computer interface is a well-understood, routine, and conventional activity (WURC) that does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“using an AI character selected by the user” -- This limitation is directed to using a selected AI character to generate recommendation data. The AI character is recited at a high level of generality and is used merely as a tool to perform the abstract idea of generating a recommended conversational response. The claim does not recite a particular AI model architecture, training technique, data structure, or improvement to computer functionality. Therefore, this element amounts to no more than mere instructions to apply the judicial exception using a generic computer component or generic computer environment, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“receiving conversation data from the at least one conversation partner” -- This limitation is directed to receiving data used by the method. Receiving conversation data is mere data gathering and is insignificant extra-solution activity. The limitation merely provides the input data used to perform the abstract idea of determining a recommended conversational response. Therefore, this limitation cannot integrate the exception into a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, sending/receiving data over a network is a well-understood, routine, and conventional activity (WURC), and it does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“the determined response data is graphically displayed and rendered in the virtual chatroom on the graphical interface of the electronic device presented in a form of a chat message.” -- This limitation is directed to outputting the result of the abstract idea by displaying the determined response as a chat message. Displaying the outcome of a mental determination on a generic graphical interface is insignificant post-solution activity (see MPEP 2106.05(g)), and the generic electronic device and graphical interface are used merely as tools to present the result, and thus it does not integrate to a practical application (see MPEP 2106.05(f)). Furthermore, under Step 2B, displaying data on a generic computer interface is a well-understood, routine, and conventional activity (WURC) that does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Therefore, claim 16 is non-patent eligible.
Regarding claim 17,
Step 1: The claim is directed to a method, which follows under the category of process. The claim satisfies Step 1.
Step 2A Prong 1:
“The Al conversation engine training method of claim 16, wherein the AI character is capable of participating in the conversation session as a conversation participant independent of the user.” - This limitation is directed to allowing the AI character to participate in the conversation session as an independent conversation participant. This limitation continues to organize and manage a communication interaction between conversation participants, including the user, conversation partner, and AI character. Therefore, this limitation is directed to certain methods of organizing human activity, including managing personal behavior or relationships or interactions between people, such as social activities and interpersonal communication.
There are no elements to be evaluated under Step 2A prong 2 and Step 2B.
Therefore, claim 17 is non-patent eligible.
Regarding claim 18,
Step 1: The claim is directed to a method, which follows under the category of process. The claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The AI conversation engine training method of claim 16, wherein the AI character does not participate in the conversation session as a conversation participant” -- The limitation recites that the AI character, first described in claim 16, will not participate in the conversation session as a participant. The limitation amounts to no more than mere further limiting to a field of use/environment, and thus does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(h)).
“and a conversation engine of the AI character is activated by the user and generates the recommendation data for the response.” -- The limitation recites that the conversation engine of the AI character will be activated by user input and then generate the data for the response. The limitation amounts to no more than mere instructions to apply onto a computer, and thus it does not integrate to a practical application, nor provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 18 is non-patent eligible.
Regarding claim 19:
Step 1: The claim is directed to a method, which follows under the category of process. The claim satisfies Step 1.
Step 2A Prong 1:
“generating the recommendation data for the response…in response to receiving the conversation data from the at least one conversation partner without the AI character” -- This limitation is directed to generating recommended response information based on received conversation data. Determining or suggesting a response based on conversation content can be performed mentally by a person reviewing the conversation and deciding what response should be recommended. Therefore, this limitation recites a mental process, including observation, evaluation, judgment, and opinion, under MPEP 2106.04(a)(2)(III). This limitation also manages interpersonal communication and therefore relates to certain methods of organizing human activity.
“the conversation engine is maintained by learning the conversation data through the conversation session” -- This limitation is directed to maintaining the conversation engine by learning conversation data through the conversation session. The limitation is recited at a high level of generality and focuses on the result of learning from conversation data, rather than a specific technological training or learning technique. To the extent the limitation uses conversation content to guide future responses, it reflects evaluating and applying conversational information, which can be characterized as a mental process.
Step 2A Prong 2 and Step 2B:
“using the conversation engine” -- This limitation merely recites use of a conversation engine as a tool to generate recommendation data. The claim does not recite a specific model architecture, neural network structure, training algorithm, parameter update rule, loss function, or technological improvement to the conversation engine. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component, which cannot integrate the exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05((f)).
“receiving the conversation data from the at least one conversation partner” -- This limitation is directed to collecting input information used by the method. Receiving conversation data is mere data gathering and is insignificant extra-solution activity. Therefore, this limitation cannot integrate the exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(g)). Furthermore, under Step 2B, sending/receiving over a network is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“the conversation engine is maintained by learning the conversation data through the conversation session” -- This limitation is recited functionally and at a high level of generality. The claim states the desired result of maintaining a conversation engine by learning conversation data, but does not recite how the learning is technically performed. The claim does not recite a specific learning method, training process, model update operation, technical data structure, or improvement to computer functionality. Therefore, this limitation does not integrate the exception into a practical application and does not provide significantly more than the abstract idea (see MPEP 2106.05(f)).
Therefore, claim 19 is non-patent eligible.
Regarding claim 20:
Step 1: The claim is directed to a method, which follows under the category of process. The claim satisfies Step 1.
Step 2A Prong 1:
“The AI conversation engine training method of claim 16, wherein the generating comprises generating the recommendation data based on an ON state or an OFF state of a response recommendation function selected by the user” -- This limitation is directed to generating recommendation data based on whether a user-selected response recommendation function is turned ON or OFF. This limitation manages whether response recommendations are provided during a conversation, which organizes and controls communication assistance in a conversation session. Therefore, this limitation is directed to certain methods of organizing human activity, including managing interpersonal communication and social interaction.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Therefore, claim 20 is non-patent eligible.
Regarding claim 21,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
Step 2A Prong 1:
“creating an AI character based on an input of a creator” -- This limitation is directed to creating or defining a virtual character/persona based on user-provided information. Under the broadest reasonable interpretation, this limitation involves organizing information about a character based on a creator’s input. Creating a persona based on user input is an activity that can be performed mentally or manually by a person selecting characteristics for a fictional or virtual character. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
“to allow the creator to modify at least one answer data;” -- This limitation is directed to the creator reviewing an answer and modifying it. A person can read an answer given in a conversation, judge that it is not the answer the person wants, and rewrite it, for example with pen and paper. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
“by including the modified answer data as part of the collected answer data.” -- This limitation is directed to deciding which answers make up the collected set of question and answer data, namely adding the creator’s modified answers to the collection. A person can compile a list of questions and answers and add corrected answers to that list. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“AI character” -- This limitation is directed to the virtual character used in the claimed method. The AI character is recited at a high level of generality and merely provides the environment or tool through which the abstract idea is carried out. The claim does not recite a particular AI model architecture, natural language processing technique, training pipeline, or improvement to computer functionality. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component or generic computer environment, which cannot integrate the exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“graphically displaying and rendering, by an electronic device, the AI character on a graphical interface of the electronic device;” -- This limitation is directed to displaying the AI character on a generic electronic device and graphical interface. The electronic device and graphical interface are recited at a high level of generality, and the claim does not recite any improvement to the graphical interface itself or to the way the electronic device renders information. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component (see MPEP 2106.05(f)), and displaying information is insignificant extra-solution activity, and it does not integrate to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, displaying data on a generic computer interface is a well-understood, routine, and conventional activity (WURC) that does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“collecting question and answer data of each of the AI character and followers that follow the AI character with respect to an initial question to generate collected question and answer data” -- This limitation also involves collecting information and data used by the method. Collecting Q/A data is mere data gathering and is insignificant extra-solution activity. The limitation merely provides the input information used to perform the abstract idea of training or guiding the AI character’s conversation engine. Therefore, this limitation cannot integrate the exception into a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, gathering Q/A to generate data is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“providing a user interface to the creator on the electronic device” -- This limitation is directed to providing a generic user interface through which the creator enters the modified answer. The user interface is recited at a high level of generality and is used merely as a tool to receive the result of the creator’s mental evaluation and editing of an answer. The claim does not recite any particular interface structure, improved interface control, or improvement to computer functionality. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
“transmitting the modified answer data to at least one follower;” -- This limitation is directed to sending the modified answer to a follower. Transmitting data is insignificant extra-solution activity because it merely outputs the result of the creator’s modification of the answer, and it does not integrate the judicial exception into a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, sending/receiving data over a network is a well-understood, routine, and conventional activity (WURC) that does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“training the conversation engine by applying the collected question and answer data to the conversation engine of the AI character, wherein the conversation engine is trained in a direction desired by the creator” -- This limitation is recited functionally and at a high level of generality. The claim states the desired result of training a conversation engine using collected Q/A data so that the engine is trained in a direction desired by the creator, but does not recite how the training is technically performed. The claim does not recite a specific model structure, parameter update operation, loss function, feature extraction process, embedding technique, or other technical improvement to AI or computer functionality. Training the engine in a direction desired by the creator reflects the content of the creator’s preferred answers rather than an improvement to the functioning of the conversation engine, the training process, or the computer itself (see MPEP 2106.05(a)). Applying generic machine learning training to a new set of data does not integrate a judicial exception into a practical application. See Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205 (Fed. Cir. 2025). Therefore, this limitation amounts to no more than mere instructions to apply the abstract idea using generic computer functionality, which cannot integrate the exception into a practical application or provide significantly more than the abstract idea (see MPEP 2106.05(f)).
Therefore, claim 21 is non-patent eligible.
Regarding claim 22:
Step 1: The claim is directed to a method, which is one of the four statutory categories. Therefore, the claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“providing a first answer of the AI character to the initial question to each of the followers that follow the AI character based on an answer input of the creator” - This limitation is directed to providing or sending conversational data to followers. Providing or transmitting answer data is insignificant extra-solution activity because it merely outputs or communicates gathered data in the claimed conversational process and does not integrate the judicial exception into a practical application (see MPEP 2106.05(g)). Additionally, under Step 2B, providing or sending data over a network using generic computer functionality is well-understood, routine, and conventional activity and does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“receiving a response to the first answer from each of the followers and providing a second answer of the AI character to the response received from each of the followers to each of the followers based on the answer input of the creator” - This limitation is directed to receiving response data and providing or sending additional answer data. Receiving and transmitting conversational data are data gathering and data output steps that are incidental to the claimed abstract idea. These steps amount to insignificant extra-solution activity and do not integrate the judicial exception into a practical application (see MPEP 2106.05(g)). Additionally, under Step 2B, receiving and sending data over a network using generic computer functionality is well-understood, routine, and conventional activity and does not amount to significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“collecting question and answer data in relation to the initial question by repeating, one or more times: the first answer of the AI character; the second answer of the AI character; and/or the response of each of the followers to the first answer of the AI character” - This limitation is directed to collecting question-and-answer data generated from repeated conversational exchanges. Collecting data for later use is mere data gathering and insignificant extra-solution activity that does not integrate the judicial exception into a practical application (see MPEP 2106.05((g)). Additionally, collecting and storing conversational data using generic computer functionality is well-understood, routine, and conventional activity and does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“first operation,” “second operation,” and “third operation” -- These limitations merely organize the data providing, receiving, and collecting steps into a sequence. The claim does not recite a technological improvement resulting from this ordering, nor does it recite an unconventional technical arrangement of computer components. Therefore, these limitations do not integrate the judicial exception into a practical application and do not provide significantly more than the judicial exception. (see MPEP 2106.05((f)).
Therefore, claim 22 is non-patent eligible.
Regarding claim 23:
Step 1: The claim is directed to a method, which is one of the four statutory categories. Therefore, the claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 21 , wherein the collecting comprises, when an answer conversation window is provided from the conversation engine of the AI character with respect to the initial question” -- This limitation merely identifies a generic user interface environment through which conversational data is provided. The claim does not recite an improved graphical user interface, improved interface control, improved display technique, or technical improvement to computer functionality. Therefore, this limitation amounts to no more than using a generic computer interface to apply the judicial exception, which cannot integrate the exception into a practical application and cannot provide significantly more than the judicial exception. (see MPEP 2106.05(f)).
“the answer conversation window is pushed for a certain period of time preset by the creator” -- This limitation merely defines a generic user interaction condition for triggering data collection. The limitation does not recite a specific technological improvement to the answer conversation window, the conversation engine, or computer functionality. Therefore, this limitation amounts to a generic control condition or field-of-use limitation that cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception. (see MPEP 2106.05(f) and MPEP 2106.05(h)).
“executing a function for collecting the question and answer data and collecting the question and answer data” - This limitation is directed to executing a function to collect Q/A data. Collecting question-and-answer data is mere data gathering and insignificant extra-solution activity because it merely obtains information used in the claimed conversational/training process, and it does not integrate to a practical application (see MPEP 2106.05(g)). Additionally, under Step 2B, executing a generic function to collect and store data using generic computer functionality is well-understood, routine, and conventional activity and does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
Therefore, claim 23 is non-patent eligible.
Regarding claim 24:
Step 1: The claim is directed to a method, which is one of the four statutory categories. Therefore, the claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“providing the second answer to each of the followers by inputting the second answer through the answer input of the creator” - This limitation is directed to providing or sending answer data to followers based on creator input. Providing or transmitting conversational data is insignificant extra-solution activity because it merely outputs data as part of the claimed conversational process and does not integrate the judicial exception into a practical application (see MPEP 2106.05(g)). Additionally, under Step 2B, providing or sending data over a network using generic computer functionality is well-understood, routine, and conventional activity and does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“answer input of the creator” - This limitation is directed to receiving creator-provided answer data. Receiving user input or gathered conversational data is mere data gathering and insignificant extra-solution activity. (see MPEP 2106.05(g)). Additionally, receiving data through a generic input interface is well-understood, routine, and conventional computer activity and does not provide significantly more than the judicial exception under Step 2B.
“provide a notification to the creator when the response to the first answer is received” - This limitation is directed to notifying the creator that response data was received. Providing a notification is a generic output or transmission of information and is insignificant extra-solution activity, which does not integrate the exception to a practical application (see MPEP 2106.05(g)). Additionally under Step 2B, sending notifications using generic computer/network functionality is well-understood, routine, and conventional activity and does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“to learn the second answer to the response received from each of the followers as an answer desired by the creator through the notification” - This limitation is recited functionally and at a high level of generality. The claim states the desired result of learning the creator’s desired answer, but does not recite how the learning is technically performed, such as through a particular model architecture, training rule, parameter update, loss function, feature representation, or natural language processing technique. Therefore, this limitation amounts to no more than mere instructions to apply the abstract idea using generic computer functionality, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 24 is non-patent eligible.
Regarding claim 26,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 21, further comprising: providing a conversation service between the AI character and the followers” -- This limitation is directed to providing a conversation service between the AI character and the followers. The limitation generically recites the exchange of conversation data between the AI character and the followers, which is insignificant extra-solution activity that cannot integrate the judicial exception into a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, providing/transmitting data between computers over a network is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“automatically collecting the question and answer data that is input from the creator through the AI character with respect to a question and a response of the follower” -- This limitation is directed to automatically collecting Q/A data. Collecting data is mere data gathering and insignificant extra-solution activity because it merely obtains information for later use in the claimed conversational/training process, and it cannot be integrated to a practical application (see MPEP 2106.05(g)). Additionally, under Step 2B, automatically collecting and storing user input data using generic computer functionality is well-understood, routine, and conventional activity and does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“in a state in which an automatic response function of the conversation service is turned OFF” -- This limitation merely defines a condition under which the Q/A data is collected. The ON/OFF state does not recite a specific technological improvement to the automatic response function, conversation service, or conversation engine. Therefore, this limitation amounts to a generic control condition or field-of-use limitation that cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f) and MPEP 2106.05(h)).
“to generate collected question and answer data” -- This limitation merely states the result of the data collection operation. Generating collected Q/A data from received user input does not improve computer functionality and is merely the result of generic data collection and storage. Therefore, this limitation does not integrate the exception into a practical application (see MPEP 2106.05(g)). Additionally, under Step 2B, generating Q/A gathered data is a well-understood, routine, and conventional activity (WURC) that cannot provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“applying the automatically collected question and answer data to the conversation engine” -- This limitation is directed to applying gathered data to a conversation engine. The claim does not recite a specific technical training mechanism, model architecture, parameter update rule, feature extraction process, embedding technique, loss function, or improvement to the conversation engine. Therefore, this limitation amounts to no more than mere instructions to apply collected data using a generic computer component, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 26 is non-patent eligible.
Regarding claim 27,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“The method of claim 26, wherein the automatic response function represents a function that allows the AI character to automatically converse with the follower in the conversation service based on the conversation engine being pretrained.” -- The limitation recites that the conversation engine’s auto response function will represent a function that allows the AI character to auto-converse with the follower in the conversation service based on the pretrained conversation engine. The limitation amounts to no more than mere further limiting to a field of use/environment, and does not integrate to a practical application nor provide significantly more than the judicial exception (see MPEP 2106.05(h)).
Therefore, claim 27 is non-patent eligible.
Regarding claim 28,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
Step 2A Prong 1:
“analyzing the introduction contents and determining a speech tone and conversation contents of interest based on at least one of the introduction contents and the at least one tag topic;” -- This limitation is directed to analyzing descriptive content and determining tone and topics of interest for a character based on the introduction contents and/or a chosen topic. A person can read introduction content and a list of topics, evaluate the wording, determine the speech tone, and identify topics or matters of interest to be assigned to a character. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
“creating the AI character to which the determined speech tone and conversation contents of interest are assigned.” -- This limitation is directed to creating or defining a virtual character/persona having assigned tone and topic attributes. Assigning speech tone and conversation contents of interest to a character is an organization of character/persona information that can be performed mentally or manually by a person deciding how a character should speak and what topics the character should discuss. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“graphically displaying and rendering, by an electronic device, a user interface comprising a character introduction window for” -- This limitation is directed to displaying a generic user interface having a window for entering character introduction contents. The electronic device, user interface, and window are recited at a high level of generality and are used merely as tools to receive the information used in the mental analysis. The claim does not recite any improvement to the user interface itself, such as an improved arrangement, interface control, or display technique, or any improvement to computer functionality. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component (see MPEP 2106.05(f)), and displaying an input window is insignificant extra-solution activity, and it cannot be integrated to a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, displaying data and receiving input through a generic computer interface is a well-understood, routine, and conventional activity (WURC) that does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“setting introduction contents that introduce an AI character to be created based on an input of a user” -- This limitation is directed to receiving or setting user-provided introduction content. Setting introduction contents based on user input is mere data gathering because it merely obtains or stores the information used in the later analysis of tone and conversation contents of interest. Therefore, this limitation is insignificant extra-solution activity that cannot integrate the judicial exception into a practical application (see MPEP 2106.05(g)). Furthermore, under Step 2B, receiving, storing, or setting user input data using generic computer functionality is well-understood, routine, and conventional activity and does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“graphically displaying and rendering, by the electronic device, the user interface further comprising a matter-of-interest window for setting at least one tag topic to be assigned to the AI character” -- This limitation is directed to displaying a second generic window of the user interface for receiving a user-selected tag topic. Receiving a tag topic through an input window is mere data gathering that provides information used in the mental determination of conversation contents of interest, and displaying the window is insignificant extra-solution activity (see MPEP 2106.05(g)). The window is recited at a high level of generality and is used merely as a tool to receive user input, without any improvement to the user interface or computer functionality (see MPEP 2106.05(f)). Furthermore, under Step 2B, displaying data and receiving input through a generic computer interface is a well-understood, routine, and conventional activity (WURC) that does not provide significantly more than the judicial exception (see MPEP 2106.05(d)(II)).
“AI character” -- This limitation merely identifies the generic AI-based environment or tool in which the character/persona information is used. The claim does not recite a specific AI model, data structure, natural language processing architecture, training process, or improvement to computer functionality. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component or environment, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 28 is non-patent eligible.
Regarding claim 29,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
Step 2A Prong 1:
“The AI character creation method of claim 28, further comprising: assigning at least one hash tag to the AI character wherein the at least one hash tag is associated with the tag topic assigned to the AI character” -- This limitation is directed to labeling a character with a tag that corresponds to its topic. A person can decide which label describes a character’s topic and write that label next to the character, such as categorizing characters in a list. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
“searching an AI character by one or more hash tags” -- This limitation is directed to finding a character that has a particular tag. A person can look through a list of tagged characters and identify the characters having a given tag. This is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“providing the user interface for searching” -- This limitation is directed to providing a generic user interface through which the search is performed. The user interface is recited at a high level of generality and is used merely as a tool to perform the abstract idea of looking up characters by tag. The claim does not recite how the search is technically performed, any particular data structure or indexing technique, or any improvement to the user interface or computer functionality. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 29 is non-patent eligible.
Regarding claim 30:
Step 1:
The claim is directed to a method, which is one of the four statutory categories. Therefore, the claim satisfies Step 1.
There are no elements to be evaluated under Step 2A Prong 1.
Step 2A Prong 2 and Step 2B:
“creating the AI character by combining the speech tone and conversation contents of interest in a preset conversation engine or by adding the speech tone and conversation contents of interest to the preset conversation engine” -- This limitation is directed to adding, combining, or associating determined informational attributes, namely speech tone and conversation contents of interest, with a preset conversation engine. Combining or adding these informational attributes to a preset conversation engine is merely using or storing data in a generic computer component and does not recite a specific technical modification to the engine. Therefore, this limitation is insignificant extra-solution activity and/or mere instructions to apply the judicial exception using a generic computer component, which cannot integrate the judicial exception into a practical application (see MPEP 2106.05(g) and/or MPEP 2106.05(f)). Furthermore, under Step 2B, storing, adding, associating, or combining user/content attributes with a generic conversation engine is well-understood, routine, and conventional computer activity and does not provide significantly more than the judicial exception. (see MPEP 2106.05(d)(II)).
“preset conversation engine” -- This limitation merely identifies the generic computer-based environment or tool in which the speech tone and conversation contents of interest are combined or added. The claim does not recite a particular conversation engine architecture, improved data structure, improved training process, parameter update rule, or improvement to computer functionality. Therefore, this limitation amounts to no more than mere instructions to apply the judicial exception using a generic computer component, which cannot integrate the judicial exception into a practical application and cannot provide significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 30 is non-patent eligible.
Regarding claim 31,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
Step 2A Prong 1:
“The AI character creation method of claim 28, wherein the at least one tag topic comprises one of coin, stock, game, and alcohol.” -- This limitation further specifies the subject matter of the tag topic used in the mental determination of conversation contents of interest recited in claim 28. A person can determine a character’s conversation contents of interest based on a topic such as coins, stocks, games, or alcohol just as readily as any other topic. Limiting the topic to particular subjects narrows the abstract idea but does not make it any less abstract, and the determination remains a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Therefore, claim 31 is non-patent eligible.
Regarding claim 32,
Step 1: The claim is directed to a method, which falls under the category of process. The claim satisfies step 1.
Step 2A Prong 1:
“The AI conversation engine training method of claim 16, wherein the personality comprises one of a caring type, a bad boy type, and a courageous type” -- This limitation further specifies the personality on which the selection of the AI character and the determined response data of claim 16 are based. A person can choose a caring, bad boy, or courageous persona and decide on a reply that reflects that personality, such as a writer composing a caring reply in a character’s voice. Limiting the personality to particular types narrows the abstract idea but does not make it any less abstract, and the selection and determination remain a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Therefore, claim 32 is non-patent eligible.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this
Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not
identically disclosed as set forth in section 102, if the differences between the claimed invention and the
prior art are such that the claimed invention as a whole would have been obvious before the effective filing
date of the claimed invention to a person having ordinary skill in the art to which the claimed invention
pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are
summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 16-20 are rejected under 35 U.S.C. 103 as being unpatentable over the NPL reference “Learning from dialogue after deployment: Feed yourself, chatbot!” by Hancock et al. (referred herein as Hancock) in view of the NPL reference “Characterchat: Supporting the creation of fictional characters through conversation and progressive manifestation with a chatbot” by Schmitt et al. (referred herein as Schmitt), in view of the NPL reference “Smart reply: Automated response suggestion for email” by Kannan et al. (referred herein as Kannan) further in view of US 20190354594A1 by Foster et al. (referred herein as Foster).
Regarding claim 16, Hancock teaches:
An artificial intelligence (AI) conversation engine training method comprising: ([Hancock, page 1] “we propose the self-feeding chatbot, a dialogue agent with the ability to extract new training examples from the conversations it participates in”, wherein the examiner interprets a dialogue agent extracting new training examples from conversations it participates in to be the same as an artificial intelligence conversation engine training method because they are both directed to training or improving a conversational artificial intelligence system using conversation-based data.)
maintaining a conversation session graphically displayed and rendered in the virtual chatroom between a user and at least one conversation partner; ([Hancock, page 3, sec. 3] “In the deployment phase, the agent engages in multi-turn conversations with users, extracting new deployment examples of two types.” AND [Hancock, page 4, sec. 3.1] “The chatbot’s primary task (DIALOGUE) is to carry on a coherent and engaging conversation with a speaking partner.” AND [Hancock, page 4, sec. 3.1] “The Human-Bot (HB) portion of the DIALOGUE dataset is extracted during deployment as described earlier, where the user is again a crowdworker instructed to chat naturally.”, wherein the examiner interprets the agent engaging in multi-turn conversations with users and carrying on a coherent and engaging conversation with a speaking partner who is instructed to chat naturally to be the same as maintaining a conversation session between a user and at least one conversation partner because they are both directed to an ongoing exchange of conversational turns between a user and a conversation partner.)
Hancock does not teach graphically displaying and rendering, by an electronic device, a virtual chatroom on a graphical interface of the electronic device…graphically displayed and rendered in the virtual chatroom…generating recommendation data for a response using an AI character selected by the user in response to receiving conversation data from the at least one conversation partner…determining response data using the recommendation data…the determined response data is graphically displayed and rendered in the virtual chatroom on the graphical interface of the electronic device presented in a form of a chat message.
Schmitt teaches:
graphically displaying and rendering, by an electronic device, a virtual chatroom on a graphical interface of the electronic device; ([Schmitt, page 5, sec. 3.1] “Fig. 1 shows the user interface of our prototype, implemented as a web app. It has three main parts: a chat view for the conversation, a character view that shows the attributes defined so far, and a pinboard view that shows dialogue pieces that the user has “bookmarked” in the chat view.” AND [Schmitt, page 6, sec. 3.2.3] “Conceptually, with this model, we condition text generation both on the conversation in the chat view (Fig. 1a) and the attribute statements in the character view (Fig. 1b).”, wherein the examiner interprets the chat view for the conversation within the user interface of the prototype implemented as a web app to be the same as a virtual chatroom on a graphical interface of the electronic device because they are both directed to a displayed area of a device’s interface in which the conversation between participants is presented.)
graphically displayed and rendered in the virtual chatroom ([Schmitt, page 6, sec. 3.2.3] “Conceptually, with this model, we condition text generation both on the conversation in the chat view (Fig. 1a) and the attribute statements in the character view (Fig. 1b).”, wherein the examiner interprets the conversation being held in the chat view of the user interface to be the same as the conversation session being graphically displayed and rendered in the virtual chatroom because they are both directed to displaying the exchanged conversational turns within the chat area of the interface.)
generating recommendation data for a response ([Schmitt, page 2] “the bot responds to anything (here: user asks about a favourite meal). The figure shows the improved version (i.e. after study 1), in which the bots suggests three potential replies (here: user selects the 2nd one, “pizza”), instead of just giving one reply directly”, wherein the examiner interprets the bot suggesting three potential replies after the user asks about a favourite meal to be the same as generating recommendation data for a response in response to receiving conversation data from the at least one conversation partner because they are both directed to generating suggested response content based on received conversational input.)
using an AI character selected by the user ([Schmitt, page 1] “writers progressively turn the bot into their imagined character through conversation”, wherein the examiner interprets a writer turning the bot into an imagined character through conversation to be the same as using an AI character selected by the user because they are both directed to a user-defined character/persona being used as the conversational bot.)
the determined response data is graphically displayed and rendered in the virtual chatroom on the graphical interface of the electronic device presented in a form of a chat message ([Schmitt, page 9, sec. 4.4.1] “We changed the system such that the UI shows three suggested replies in Open Conversation Mode, instead of giving just one reply directly (Fig. 2b). Only the user’s chosen reply is added to the chat history.” AND [Schmitt, page 5, sec. 3.1] “It has three main parts: a chat view for the conversation, a character view that shows the attributes defined so far, and a pinboard view that shows dialogue pieces that the user has “bookmarked” in the chat view.”, wherein the examiner interprets the user’s chosen reply from the suggested replies being added to the chat history of the conversation shown in the chat view to be the same as the determined response data being graphically displayed and rendered in the virtual chatroom presented in a form of a chat message because they are both directed to presenting the chosen conversational response as a message within the displayed conversation.)
Hancock and Schmitt do not teach in response to receiving conversation data from the at least one conversation partner…determining response data using the recommendation data.
Kannan teaches:
in response to receiving conversation data from the at least one conversation partner, ([Kannan, page 2, sec 1] “our input sequence is an incoming message and our output distribution is over the space of possible replies.”, wherein the examiner interprets “incoming message” to be the same as “receiving conversation data from .. at least one conversation partner”, because they are both describing the system generating a response upon receiving an incoming message (conversation data) from the conversation partner.)
determining response data using the recommendation data ([Kannan, page 3, sec 3] “The fundamental task of the Smart Reply system is to find the most likely response given an original message.” AND [Kannan, page 3, sec 3] “To find this response, we will construct a model that can score responses and then find the highest scoring response”, wherein the examiner interprets finding the most likely response by scoring responses and finding the highest scoring response to be the same as determining response data using the recommendation data because they are both directed to selecting or determining response content from generated candidate response information.)
Hancock, Schmitt, and Kannan do not teach wherein the AI character is selected based on a personality of the AI character;…the response data being associated with the personality of the selected AI character.
Foster teaches:
wherein the AI character is selected based on a personality of the AI character ([Foster, [0019]] “The selector component 120 can expose an interface, for example, that renders available personas associated with language generation and receives a selection of a persona…For example, predefined personas can include, but are not limited to, helpful and friendly for support, fun and cool targeting teens or more casual audiences, formal for trusted guidance (for example, for recommendations and directions), smart and informed targeting education or information seeking audiences, and child oriented for young audiences.” AND [Foster, [0028]] “Furthermore, conversational agent 310 can correspond to a chatbot, agent, personal assistant, game character, or conversation-enabled physical robot, among other things.”, wherein the examiner interprets the interface that renders available personas, such as helpful and friendly or fun and cool, and receives a selection of a persona for a conversational agent such as a chatbot or game character to be the same as the AI character being selected based on a personality of the AI character because they are both directed to choosing a conversational character according to the set of personal qualities that character exhibits.)
the response data being associated with the personality of the selected AI character ([Foster, [0019]] “The selected persona can thus be infused into responses.” AND [Foster, [0022]] “For example, if uploaded conversational data is friendly and helpful, responses produced by a language generation model will also be consistent with the persona, namely friendly and helpful.” AND [Foster, [0028]] “In accordance with one embodiment, the model is able to generate fluent small talk, or chitchat, in accordance with a particular persona in response to a message, wherein small talk corresponds to social conversation regarding trivial or uncontroversial matters.” AND [Foster, [0045]] “Further, responses can be tailored to a particular persona by embedding speaker/user identifiers or persona identifiers (e.g., speaker/user identifiers matched to persona) within the model.”, wherein the examiner interprets the selected persona being infused into responses that are consistent with the persona, including responses generated in response to a message in accordance with a particular persona, to be the same as the response data being associated with the personality of the selected AI character because they are both directed to generating responses that reflect the personal qualities of the chosen conversational character.)
Hancock, Schmitt, Kannan, Foster, and the instant application are analogous art because they are all directed to artificial intelligence conversation systems that maintain conversational interactions and generate conversation data.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the self-feeding chatbot disclosed by Hancock to include the bot recommendation process disclosed by Schmitt. One would be motivated to do so to effectively generate character-based suggested response content during an open conversation, as suggested by Schmitt ([Schmitt, Abstract] “Open conversation with the chatbot helps writers explore their character and get inspiration”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the LSTM neural network processes disclosed by Kannan. One would be motivated to do so to efficiently assist a user in composing and selecting suitable response options to incoming conversational messages, as suggested by Kannan ([Kannan, page 1, sec 1] “Thus we raised the following question: can we assist users with composing these short messages? More specifically, would it be possible to suggest brief responses when appropriate, just one tap away?”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the “selection of a persona” disclosed by Foster. One would be motivated to do so to effectively generate conversation that is relatable to the particular type of personality chosen for the conversational character, as suggested by Foster ([Foster, [0015]] “Conversation can be generated that is human like and relatable with respect to a particular type of person, and therefore useful in responding to human conversation.”).
Regarding claim 17, Hancock, Schmitt, Kannan, and Foster teach The AI conversation engine training method of claim 16 (see rejection of claim 16).
Hancock further teaches:
wherein the AI character is capable of participating in the conversation session as a conversation participant independent of the user. ([Hancock, page 3] “the bot continues the conversation with its own response” AND [Hancock, page 4] “The chatbot’s primary task (DIALOGUE) is to carry on a coherent and engaging conversation with a speaking partner”, wherein the examiner interprets the bot continuing the conversation with its own response and “carry on a coherent and engaging conversation” with a speaking partner to be the same as the AI character being capable of participating in the conversation session as a conversation participant independent of the user because they are both directed to a chatbot/conversation engine generating its own conversational response and carrying on the conversation without requiring the user to provide the response.)
Regarding claim 18, Hancock, Schmitt, Kannan, and Foster teach The AI conversation engine training method of claim 16 (see rejection of claim 16).
Schmitt further teaches:
wherein the AI character does not participate in the conversation session as a conversation participant, ([Schmitt, page 9] “the UI shows three suggested replies in Open Conversation Mode, instead of giving just one reply directly”, wherein the examiner interprets the UI showing three suggested replies instead of giving one reply directly to be the same as the AI character does not participate in the conversation session as a conversation participant because they are both directed to the AI character/chatbot not directly providing its own response into the conversation, but instead providing selectable suggested replies for the user.)
and a conversation engine of the AI character is activated by the user, ([Schmitt, page 2] “Saying “Let’s chat” transitions to Open Conversation Mode, in which the bot responds to anything”, wherein the examiner interprets saying ‘Let’s chat’ “transitions to Open Conversation Mode” to be the same as a conversation engine of the AI character is activated by the user because they are both directed to a user input initiating operation of the character-based conversation mode.)
and generates the recommendation data for the response. ([Schmitt, page 9] “the UI shows three suggested replies in Open Conversation Mode, instead of giving just one reply directly” AND [Schmitt, page 6] “It generates text based on both the chat history and a few sentences about the chatbot’s ‘personality’”, wherein the examiner interprets the UI showing three “suggested replies in Open Conversation Mode” and generating text based on chat history and chatbot personality to be the same as generates the “recommendation data for the response” because they are both directed to producing candidate response content based on conversation context and character/personality information.)
Hancock, Schmitt, Kannan, Foster, and the instant application are analogous art because they are all directed to artificial intelligence conversation systems.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 16 disclosed by Hancock, Schmitt, Kannan, and Foster to include the Conversation Mode tool disclosed by Schmitt. One would be motivated to do so to effectively provide user-selectable response recommendations while allowing the user to control whether the AI character directly participates in the conversation or merely provides suggested response content, as suggested by Schmitt ([Schmitt, page 9] “the UI shows three suggested replies in Open Conversation Mode, instead of giving just one reply directly”).
Regarding claim 19, Hancock, Schmitt, Kannan, and Foster teach The AI conversation engine training method of claim 18 (see rejection of claim 18).
Hancock further teaches:
and the conversation engine is maintained by learning the conversation data through the conversation session. ([Hancock, page 1] “we propose the self-feeding chatbot, a dialogue agent with the ability to extract new training examples from the conversations it participates in. As our agent engages in conversation, it also estimates user satisfaction in its responses. When the conversation appears to be going well, the user’s responses become new training examples to imitate...learning from dialogue with a self-feeding chatbot significantly improves performance” AND [Hancock, page 3] “The chatbot is periodically retrained with the available examples from all four datasets, improving its DIALOGUE performance without collecting any new supervised examples”, wherein the examiner interprets extracting new training examples from conversations, using the user’s responses as new training examples to imitate, learning from dialogue, and periodically retraining the chatbot with available examples to improve dialogue performance to be the same as the conversation engine being maintained by learning the conversation data through the conversation session because they are both directed to improving or updating a conversational AI model using conversation data obtained during the ongoing conversation session.)
Kannan further teaches wherein: the generating comprises generating the recommendation data for the response using the conversation engine in response to receiving the conversation data from the at least one conversation partner without the AI character; ([Kannan, page 1] “Smart Reply: Automated Response Suggestion for Email…It generates semantically diverse suggestions that can be used as complete email responses with just one tap on mobile” AND [Kannan, page 2] “our input sequence is an incoming message and our output distribution is over the space of possible replies…Response selection: At the core of our system, an LSTM neural network processes an incoming message, then uses it to predict the most likely responses”, wherein the examiner interprets the Smart Reply system generating “semantically diverse suggestions” that can be used as complete responses, using an LSTM neural network that processes an incoming message and predicts the most likely responses, to be the same as generating the recommendation data for the response using the conversation engine in response to receiving the conversation data from the at least one conversation partner because they are both directed to a computer-implemented conversation model receiving incoming conversational message data and generating suggested response data based on that received message data. The examiner further interprets the “Smart Reply” system to be an automated email response suggestion system for assisting a user in responding to incoming messages to be the same as generating the recommendation data without the AI character because they are both directed to a response recommendation engine providing suggested response content for a user’s communication without requiring an AI character to participate as the conversation partner.)
Hancock, Schmitt, Kannan, Foster, and the instant application are analogous art because they are all directed to using conversational artificial intelligence systems to generate response data.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 18 disclosed by Hancock, Schmitt, Kannan, and Foster to include the Smart Reply system disclosed by Kannan. One would be motivated to do so to effectively provide suggested response content for a user without requiring the AI character to directly participate as the conversation partner, as suggested by Kannan ([Kannan, page 1] “It generates semantically diverse suggestions that can be used as complete email responses with just one tap on mobile.”).
Regarding claim 20, Hancock, Schmitt, Kannan, and Foster teach The AI conversation engine training method of claim 16 (see rejection of claim 16).
Kannan further teaches wherein the generating comprises generating the recommendation data based on an ON state or an OFF state of a response recommendation function selected by the user. ([Kannan, page 2] “Triggering model: A feedforward neural network decides whether or not to suggest responses” AND [Kannan, page 7] “If the decision is negative, we finish the execution and do not show any suggestions… If the score is above some threshold, we trigger and run the LSTM scoring”, wherein the examiner interprets the feedforward neural network deciding whether or not to suggest responses, not showing suggestions when the decision is negative, and triggering and running LSTM scoring when the score is above a threshold to be the same as generating the recommendation data based on an ON state or an OFF state of a response recommendation function because they are both directed to selectively activating or deactivating generation of response suggestions based on whether a response suggestion function is triggered or not triggered.)
Hancock, Schmitt, Kannan, Foster, and the instant application are analogous art because they are all directed to generating suggested conversational response content using an artificial intelligence.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 16 disclosed by Hancock, Schmitt, Kannan, and Foster to include the Triggering model disclosed by Kannan. One would be motivated to do so to effectively improve the usefulness of generated response recommendations by avoiding presentation of response suggestions when they are unlikely to be useful to the user, as suggested by Kannan ([Kannan, page 2] “This further improves utility by not showing suggestions when they are unlikely to be used.”).
Claims 21-24 and 26-27 are rejected under 35 U.S.C. 103 as being unpatentable over Hancock in view of US 20140032471A1 by Reddy et al. (referred herein as Reddy) further in view of US 20140122083A1 by Xiaojiang et al. (referred herein as Xiaojiang).
Regarding claim 21, Hancock teaches:
A method for training a conversation engine of an artificial intelligence (AI) character, comprising: ([Hancock, page 1] “we propose the self-feeding chatbot, a dialogue agent with the ability to extract new training examples from the conversations it participates in”, wherein the examiner interprets a dialogue agent extracting new training examples from conversations it participates in to be the same as a method for training a conversation engine because they are both directed to improving a conversational artificial intelligence system using examples obtained from conversations.)
collecting question and answer data of each of the AI character and followers that follow the AI character with respect to an initial question to generate collected question and answer data; ([Hancock, page 2] “the conversations the chatbot participates in are sliced into two complementary datasets--one largely protected from the chatbot’s mistakes (DIALOGUE examples), and one which directly addresses them (FEEDBACK examples)” AND [Hancock, Fig. 1] “Have you been to France?” AND [Hancock, page 4] “the agent engages in multi-turn conversations with users, extracting new deployment examples of two types”, wherein the examiner interprets conversations involving the chatbot being sliced into DIALOGUE examples and FEEDBACK examples, the agent engaging in multi-turn conversations with users while extracting new deployment examples, and the chatbot asking what it should have said in relation to a prior conversational prompt to be the same as collecting question and answer data of each of the AI character and followers that follow the AI character with respect to an initial question to generate collected question and answer data because they are both directed to obtaining conversational exchanges between a chatbot/conversational agent and users, including user responses and feedback tied to a preceding question or prompt, and converting those exchanges into collected datasets or examples for training.)
training the conversation engine by applying the collected question and answer data to the conversation engine of the AI character, wherein the conversation engine is trained in a direction desired by the creator by including the modified answer data as part of the collected answer data. ([Hancock, page 2] “the dialogue ability of the chatbot is always improved by adding the automatically extracted examples of either type, and improves the most by adding both”, wherein the examiner interprets improving the dialogue ability of the chatbot by adding automatically extracted examples to be the same as training the conversation engine by applying the collected question and answer data because they are both directed to improving chatbot/conversation-engine performance by using collected conversational examples.)
Hancock does not teach creating an AI character based on an input of a creator; graphically displaying and rendering, by an electronic device, the AI character on a graphical interface of the electronic device;…providing a user interface to the creator on the electronic device to allow the creator to modify at least one answer data; transmitting the modified answer data to at least one follower; or wherein the conversation engine is trained in a direction desired by the creator by including the modified answer data as part of the collected answer data.
Reddy teaches creating an AI character based on an input of a creator; ([Reddy, [0005]] “Systems and methods are described for creating interactive synthetic characters using an artificial intelligence Scripting tool…A set of conversation rules can be received from a user” AND [Reddy, [0011]] “The graphical user interface may include an interactive Script editor through which a set of customized rules and responses can be entered by a user.”, wherein the examiner interprets creating interactive synthetic characters using an artificial intelligence scripting tool, receiving conversation rules from a user, and allowing a user to enter customized rules and responses through an interactive script editor to be the same as creating an AI character based on an input of a creator because they are both directed to a user/creator providing rules, responses, or other configuration input used to create or define the behavior and conversational content of an artificial intelligence character.)
Hancock and Reddy do not teach graphically displaying and rendering, by an electronic device, the AI character on a graphical interface of the electronic device; providing a user interface to the creator on the electronic device to allow the creator to modify at least one answer data; transmitting the modified answer data to at least one follower; or wherein the conversation engine is trained in a direction desired by the creator by including the modified answer data as part of the collected answer data.
Xiaojiang teaches:
graphically displaying and rendering, by an electronic device, the AI character on a graphical interface of the electronic device; ([Xiaojiang, [0059]] “Chatbot system 108 includes a graphical image representing chatbot dialog interface 116, the graphical image including, without limitation, an avatar, a talking head, a text-to-speech engine, etc.” AND [Xiaojiang, [0074]] “In FIG. 2, chatbot dialog interface 216 includes a visual graphic that encompasses two main areas namely dialog box 248 and output display 211.”, wherein the examiner interprets the graphical image of the chatbot, such as an avatar or a talking head, presented with the chatbot dialog interface to be the same as the AI character being graphically displayed and rendered on a graphical interface of the electronic device because they are both directed to a visual representation of a conversational agent presented on a device display.)
providing a user interface to the creator on the electronic device to allow the creator to modify at least one answer data; ([Xiaojiang, [0081]] “An advantage of the present invention is that chat log 300 can be employed to edit input/ output messages 310, 312, 314 and consequently corresponding knowledge entries.” AND [Xiaojiang, [0088]] “Where users find an unsatisfying reply such as output message 310B given by the chatbot system 108 when reviewing the chat logs, users can select the unsatisfying reply, and select enter a better reply.”, wherein the examiner interprets the chat log through which the merchant who trains the chatbot selects an unsatisfying chatbot reply and enters a better reply to be the same as a user interface provided to the creator to allow the creator to modify at least one answer data because they are both directed to an interface through which the party responsible for the chatbot replaces a chatbot answer with a preferred answer.)
transmitting the modified answer data to at least one follower; ([Xiaojiang, [0072]] “Generally, merchant 106 can then enter ‘100 Main Street’ into chatbot dialog interface 116. Thereafter, when the customer enters an input message ‘Where is your business located?’ chatbot system 108 replies, ‘100 Main Street’”., wherein the examiner interprets the chatbot system replying to the customer with the answer entered by the merchant to be the same as transmitting the modified answer data to at least one follower because they are both directed to delivering the creator-modified answer to a user who interacts with the conversational agent.)
wherein the conversation engine is trained in a direction desired by the creator by including the modified answer data as part of the collected answer data. ([Xiaojiang, [0071]] “Merchant 106 trains chatbot system 108 to provide modified chatbot messages 109 during a chat session.” AND [Xiaojiang, [0079]] “In this manner, merchant 106 can train chatbot system 108 to provide enhanced responses to input messages from user 102 or other customers.” AND [Xiaojiang, [0089]] “Chatbot system 108 checks if the old reply is provided by a script; if yes, chatbot system 108 updates the corresponding output message node; if not, chatbot system 108 creates a new script with the user message and the better reply as its child.”, wherein the examiner interprets the merchant training the chatbot system with the merchant’s better replies, which are stored with the corresponding user messages, to be the same as the conversation engine being trained in a direction desired by the creator by including the modified answer data as part of the collected answer data because they are both directed to adding the creator’s preferred answers to the conversational data used to train the conversation engine.)
Hancock, Reddy, Xiaojiang, and the instant application are analogous art because they are all directed to artificial intelligence conversation systems for creating or improving character-based conversational agents.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of training a conversational agent disclosed by Hancock to include the scripting tool disclosed by Reddy. One would be motivated to do so to effectively allow the conversational agent of Hancock to be configured as a user-created interactive character with customized conversational rules and responses, as suggested by Reddy ([Reddy, [0007]] “the Script allows for dynamic interactions that can depend on multiple variables and/or change over time”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of training a conversational agent disclosed by Hancock to include the “chat log 300 can be employed to edit input/output messages” technique disclosed by Xiaojiang. One would be motivated to do so to effectively improve the accuracy and usefulness of chatbot responses by allowing creator-provided replies to replace unsatisfactory replies and train the chatbot for future interactions, as suggested by Xiaojiang ([Xiaojiang, [0079]] “In this manner, merchant 106 can train chatbot system 108 to provide enhanced responses to input messages from user 102 or other customers”).
Regarding claim 22, Hancock, Reddy, and Xiaojiang teach The method of claim 21 (see rejection of claim 21).
Xiaojiang further teaches:
wherein the collecting comprises: a first operation of providing a first answer of the AI character to the initial question to each of the followers that follow the AI character based on an answer input of the creator; ([Xiaojiang, [0076]-[0077]] “Responsive to prompt message “What can I do for you today?” 213, the user (e.g. merchant 106) enters an input message "How to train my robot?” 210 into dialog box 248 and then clicks send button 250. Chatbot system 108 responds by displaying output message “Chat with her 209... Merchant 106 then enters user-generated message "Chat with her and give better replies' 200 into dialog box 248.” AND [Xiaojiang, [0079]] “Future input inquiries relating to “How to train my robot?" entered into dialog box 248 would generate the response "Chat with her and give better replies”. In this manner, merchant 106 can train chatbot system 108 to provide enhanced responses to input messages from user 102 or other customers.”, wherein the examiner interprets the chatbot system generating a response to future input inquiries based on the merchant-entered user-generated message to be the same as providing a first answer of the AI character to the initial question based on an answer input of the creator because they are both directed to a conversational artificial intelligence entity providing an answer to an initial inquiry using answer content supplied by a creator/merchant. The examiner further interprets user 102 or other customers to be the same as followers that follow the AI character because they are both directed to multiple end users who interact with the conversational artificial intelligence entity.)
a second operation of receiving a response to the first answer from each of the followers and providing a second answer of the AI character to the response received from each of the followers to each of the followers based on the answer input of the creator; ([Xiaojiang, [0081]] “chat log 300 displays input/output messages 310, 312,314 communicated between a user and chatbot system 108 during a chat session.” AND [Xiaojiang, [0083]] “Here, the initial user input message is "Are all plans free’?” 310A. Responsive thereof, chatbot system 108 responds with output message “Sorry, but I need to check with my colleagues to see if it is free. I will email the answer to you.' 310B. This output message 310 is crossed out and replaced with user-generated output message “No, there are several available plans. One is free.”310C.” AND [Xiaojiang, [0084]] “A follow up input message "Do you have an affiliate program?” 312A is entered into chatbot system 108.” AND [Xiaojiang, [0084]] “The responsive output message ‘I need to check it first. I will get back to you soon by email.’312B is crossed out. And, it is replaced with user-generated output message ‘No, but we will inform you when one becomes available.’ 312C.”, wherein the examiner interprets a follow up input message entered into the chatbot system after an initial user input message and chatbot response, followed by replacement with a user-generated output message, to be the same as receiving a response to the first answer from each of the followers and providing a second answer of the AI character to the response received from each of the followers based on the answer input of the creator because they are both directed to receiving a later user/customer message in a conversation and providing a subsequent conversational artificial intelligence answer based on creator/merchant-provided answer content.)
and a third operation of collecting question and answer data in relation to the initial question by repeating, one or more times: the first answer of the AI character; the second answer of the AI character; and/or the response of each of the followers to the first answer of the AI character. ([Xiaojiang, [0081]] “chat log 300 displays input/output messages 310, 312,314 communicated between a user and chatbot system 108 during a chat session... chat log 300 can use Scripts to create or update input/output messages 310, 312. 314 that modify knowledge entries for chatbot system 108... A script typically consists of user messages responsive to the chatbot messages and chatbot messages responsive to the user messages.” AND [Xiaojiang, [0084 - 0085]] “A follow up input message "Do you have an affiliate program?” 312A is entered into chatbot system 108... As for input/output message 314, the follow up inquiry is ‘How can I install Aco on my website?’314A.”, wherein the examiner interprets displaying multiple input/output (I/O) messages communicated during a chat session, using scripts to create or update those I/O messages to modify knowledge entries, and including user messages responsive to chatbot messages and chatbot messages responsive to user messages to be the same as collecting question and answer data in relation to the initial question by repeating the first answer of the AI character, the second answer of the AI character, and/or the response of each of the followers to the first answer of the AI character because they are both directed to collecting repeated conversational exchanges that include artificial intelligence answers and user responses and using those exchanges to update chatbot knowledge or training data.)
Hancock, Reddy, Xiaojiang, and the instant application are analogous art because they are all directed to training or updating a conversational artificial intelligence system.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 21 disclosed by Hancock, Reddy, and Xiaojiang to include the chatbot system disclosed by Xiaojiang. One would be motivated to do so to effectively improve the accuracy and usefulness of chatbot responses by allowing user/creator-provided conversational replies to update the chatbot’s knowledge entries for future interactions, as suggested by Xiaojiang ([Xiaojiang, [0079]], “merchant 106 can train chatbot system 108 to provide enhanced responses to input messages from user 102 or other customers.”)
Regarding claim 23, Hancock, Reddy, and Xiaojiang teach The method of claim 21 (see rejection of claim 21).
Hancock further teaches:
executing a function for collecting the question and answer data and collecting the question and answer data. ([Hancock, page 4] “In the deployment phase, the agent engages in multi-turn conversations with users, extracting new deployment examples of two types” AND [Hancock, page 4] “The rate at which new DIALOGUE or FEEDBACK examples are collected can be adjusted by raising or lowering the satisfaction threshold”, wherein the examiner interprets extracting new deployment examples from multi-turn conversations and collecting new DIALOGUE or FEEDBACK examples to be the same as executing a function for collecting the question and answer data and collecting the question and answer data because they are both directed to performing a data-collection operation that obtains conversational examples from user/chatbot interactions for training or improving the conversation engine.)
Reddy further teaches and the answer conversation window is pushed for a certain period of time preset by the creator, ([Reddy, [0042]] “support for time-based rules that allow a toy's character to evolve over usage time” AND [Reddy, [0044]] “Chat window 230 allows the user to test out the latest content by typing words and seeing how the toy would respond with the given conversation rules”, wherein the examiner interprets support for time-based rules that allow a character to evolve over usage time, together with a chat window for testing how the toy responds according to conversation rules, to be the same as the answer conversation window being pushed for a certain period of time preset by the creator because they are both directed to applying creator/user-defined time-based conversation rules to a chat-window interaction involving an AI/synthetic character.)
Xiaojiang further teaches wherein the collecting comprises, when an answer conversation window is provided from the conversation engine of the AI character with respect to the initial question ([Xiaojiang, [0074]-[0075]] “chatbot dialog interface 216 includes a visual graphic that encompasses two main areas namely dialog box 248 and output display 211…Once chatbot system 108 is initialized, chatbot dialog interface 216 is displayed and is ready to initiate a chat session.” AND [Xiaojiang, [0076]] “Responsive to prompt message ‘What can I do for you today?’ 213, the user (e.g. merchant 106) enters an input message ‘How to train my robot?’ 210 into dialog box 248 and then clicks send button 250. Chatbot system 108 responds by displaying output message ‘Chat with her 209”, wherein the examiner interprets the chatbot dialog interface having a dialog box and output display, being displayed and ready to initiate a chat session, and displaying a chatbot output message responsive to an input message to be the same as an answer conversation window being provided from the conversation engine of the AI character with respect to the initial question because they are both directed to presenting a conversational interface in which a user provides an initial question or message and the chatbot/conversation engine provides a corresponding answer through the interface.)
Hancock, Reddy, Xiaojiang, and the instant application are analogous art because they are all directed to collecting conversational question-and-answer data through a chatbot or AI character interface.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 21 disclosed by Hancock, Reddy, and Xiaojiang to include the “time-based rules” disclosed by Reddy. One would be motivated to do so to effectively allow the creator to control how the character’s conversational behavior changes over a period of use, as suggested by Reddy ([Reddy, [0042]] “support for time-based rules that allow a toy's character to evolve over usage time”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the chatbot disclosed by Xiaojiang. One would be motivated to do so to effectively provide a user-facing conversation window that receives user questions and displays chatbot answers during the chat session, as suggested by Xiaojiang ([Xiaojiang, [0074]-[0075]] “chatbot dialog interface 216 is displayed and is ready to initiate a chat session.”).
Regarding claim 24, Hancock, Reddy, and Xiaojiang teach The method of claim 22 (see rejection of claim 22).
Xiaojiang further teaches:
wherein the second operation comprises providing the second answer to each of the followers by inputting the second answer through the answer input of the creator ([Xiaojiang, [0071]] “Merchant 106 trains chatbot system 108 to provide modified chatbot messages 109 during a chat session” AND [Xiaojiang, [0072]] “Merchant 106 being dissatisfied with this message may wish to replace this chatbot response with the merchant’s business address. Generally, merchant 106 can then enter “100 Main Street' into chatbot dialog interface 116. Thereafter, when the customer enters an input message "Where is your business located?” chatbot system 108 replies, “100 Main Street'.”, wherein the examiner interprets the merchant entering 100 Main Street into the chatbot dialog interface and the chatbot thereafter replying 100 Main Street to the customer to be the same as providing the second answer to each of the followers by inputting the second answer through the answer input of the creator because they are both directed to a creator/merchant inputting a replacement or modified chatbot answer through a chatbot input interface so that the chatbot provides that answer to later users/customers.)
in order to provide a notification to the creator when the response to the first answer is received ([Xiaojiang, [0037]] “FIG. 3 illustrates a chat log according to an exemplary embodiment of the present invention” AND [Xiaojiang, [0081]] “chat log 300 displays input/output messages 310, 312,314 communicated between a user and chatbot system 108 during a chat session. An advantage of the present invention is that chat log 300 can be employed to edit input/output messages 310, 312, 314 and consequently corresponding knowledge entries”, wherein the examiner interprets the chat log displaying input/output messages communicated between the user and chatbot system during the chat session, and being employed to edit the input/output messages, to be the same as providing a notification to the creator when the response to the first answer is received because they are both directed to presenting received user/customer conversational responses to the creator/merchant in an interface so that the creator/merchant can act on the received response.)
and to learn the second answer to the response received from each of the followers as an answer desired by the creator through the notification. ([Xiaojiang, [0071]-[0072]] “Merchant 106 trains chatbot system 108 to provide modified chatbot messages 109 during a chat session... Those skilled in the art will recognize that user-generated message (provided by merchant 106) is more robust and descriptive than the initial chatbot message” AND [Xiaojiang, [0069]] “User-generated messages provide a context for subsequent chatbot messages of chatbot system 108, and for Subsequent user input messages... a new chat bot message is repetitively modified both prior to and during a chat session by entering a plurality of user-generated messages and training chatbot system 108 by preceding and/or following user-generated messages with unique identifiers”, wherein the examiner interprets the merchant-provided user-generated messages training the chatbot system to provide modified chatbot messages for subsequent chatbot messages and subsequent user input messages to be the same as learning the second answer to the response received from each of the followers as an answer desired by the creator because they are both directed to storing or applying a creator/merchant-provided answer as the desired chatbot answer for later use in response to user/customer messages.)
Hancock, Reddy, Xiaojiang, and the instant application are analogous art because they are all directed to training or improving a chatbot/conversation engine using conversation-session data.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 22 disclosed by Hancock, Reddy, and Xiaojiang to include the user-generated message incorporation technique disclosed by Xiaojiang. One would be motivated to do so to effectively improve the quality and usefulness of chatbot responses by allowing a creator/merchant to provide corrected or preferred answer content that can be used by the chatbot in later conversations, as suggested by Xiaojiang ([Xiaojiang, [0071]-[0072]] “user-generated message (provided by merchant 106) is more robust and descriptive than the initial chatbot message”).
Regarding claim 26, Hancock, Reddy, and Xiaojiang teach The method of claim 21 (see rejection of claim 21).
Hancock further teaches:
further comprising: providing a conversation service between the AI character and the followers; ([Hancock, page 3, sec. 3] “In the deployment phase, the agent engages in multi-turn conversations with users, extracting new deployment examples of two types.” AND [Hancock, page 4, sec. 3.1] “The Human-Bot (HB) portion of the DIALOGUE dataset is extracted during deployment as described earlier, where the user is again a crowdworker instructed to chat naturally.”, wherein the examiner interprets the deployed agent engaging in multi-turn conversations with users who are instructed to chat naturally with it to be the same as providing a conversation service between the AI character and the followers because they are both directed to an ongoing conversational exchange between a conversational agent and the users who interact with it.)
automatically collecting the question and answer data ([Hancock, page 2] “we propose the self-feeding chatbot, a dialogue agent with the ability to extract new training examples from the conversations it participates in” AND [Hancock, page 3] “the agent engages in multi-turn conversations with users, extracting new deployment examples of two types … the agent extracts a new Human-Bot (HB) DIALOGUE example using the previous context x and the human’s response y” AND [Hancock, page 4] “the resulting feedback response f is used to create a new example for the FEEDBACK task … Periodically, the agent is retrained using all available data, thereby improving performance on the primary DIALOGUE task”, wherein the examiner interprets the self-feeding chatbot extracting new training examples from conversations it participates in, extracting new deployment examples from multi-turn conversations using previous context and human responses, and creating new feedback examples from resulting feedback responses to be the same as automatically collecting the question and answer data because they are both directed to automatically collecting conversation-based input/response examples from dialogue for use as training data.)
to generate collected question and answer data; and applying the automatically collected question and answer data to the conversation engine. ([Hancock, page 2] “the conversations the chatbot participates in are sliced into two complementary datasets - one largely protected from the chatbot’s mistakes (DIALOGUE examples), and one which directly addresses them (FEEDBACK examples) … adding the automatically extracted examples of either type” AND [Hancock, page 3] “The chatbot is periodically retrained with the available examples from all four datasets, improving its DIALOGUE performance without collecting any new supervised examples” AND [Hancock, page 4] “Periodically, the agent is retrained using all available data, thereby improving performance on the primary DIALOGUE task”, wherein the examiner interprets slicing conversations into DIALOGUE examples and FEEDBACK examples and adding automatically extracted examples of either type to be the same as generating collected question and answer data because they are both directed to forming datasets of collected conversational examples from user-chatbot interactions. The examiner further interprets periodically retraining the chatbot with available examples from the datasets and retraining the agent using all available data to be the same as applying the collected question and answer data to the conversation engine because they are both directed to using collected conversational training examples to update or improve the chatbot/dialogue engine.)
Xiaojiang further teaches:
that is input from the creator through the AI character with respect to a question and a response of the follower ([Xiaojiang, [0071]] “Merchant 106 trains chatbot system 108 to provide modified chatbot messages 109 during a chat session…User/merchant 106 begins by entering ‘Where are your offices?’ into chatbot dialog interface 116” AND [Xiaojiang, [0072]] “Merchant 106 being dissatisfied with this message may wish to replace this chatbot response with the merchant’s business address” AND [Xiaojiang, [0072]] “merchant 106 can then enter ‘100 Main Street’ into chatbot dialog interface 116…Thereafter, when the customer enters an input message ‘Where is your business located?’ chatbot system 108 replies, ‘100 Main Street’”, wherein the examiner interprets the merchant entering a replacement chatbot response into the chatbot dialog interface and the chatbot later using that response when a customer enters a corresponding input message to be the same as question and answer data that is input from the creator through the AI character with respect to a question and a response of the follower because they are both directed to a creator/merchant providing response data through a chatbot interface so the chatbot can answer user/follower questions.)
in a state in which an automatic response function of the conversation service is turned OFF ([Xiaojiang, [0072]] “In response, chatbot system 108 might generate a response such as ‘Please check our website, thank you.’ Merchant 106 being dissatisfied with this message may wish to replace this chatbot response with the merchant’s business address” AND [Xiaojiang, [0088]] “Where users find an unsatisfying reply such as output message 310B given by the chatbot system 108 when reviewing the chat logs, users can select the unsatisfying reply, and select enter a better reply…Chatbot system 108 marks the old reply as deleted and also displays a new reply” AND [Xiaojiang, [0077]] “This user-generated message 200 is preceded by a predetermined unique identifier such as the double angle brackets “>>” 202.” AND [Xiaojiang, [0078]] “Predetermined unique identifier 202 directs chatbot system 108 to replace output message “Chat with her” 209 with user-generated message “Chat with her and give better replies” 200.”, wherein the examiner interprets replacing an unsatisfying chatbot-generated response with a user-provided better reply to be the same as collecting creator-input data while an automatic response function is turned OFF because they are both directed to disabling, bypassing, or replacing the chatbot’s automatic response with a creator/user-provided response for training or updating the chatbot. The examiner further interprets the predetermined unique identifier directing the chatbot system to replace its own output message with the merchant’s user-generated message to be the same as the automatic response function of the conversation service being turned OFF because they are both directed to suppressing the AI character’s automatically generated reply so that the creator’s input is provided through the AI character instead.)
Hancock, Reddy, Xiaojiang, and the instant application are analogous art because they are all directed to training or updating a conversation engine of an artificial intelligence character.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the method of claim 21 disclosed by Hancock, Reddy, and Xiaojiang to include the creator-input replacement-response technique disclosed by Xiaojiang. One would be motivated to do so to effectively allow a creator or administrator to correct unsatisfactory chatbot responses and train the chatbot to provide improved future responses, as suggested by Xiaojiang ([Xiaojiang, [0088]] “users can select the unsatisfying reply, and select enter a better reply”).
Regarding claim 27, Hancock, Reddy, and Xiaojiang teach The method of claim 26 (see rejection of claim 26).
Hancock further teaches:
wherein the automatic response function represents a function that allows the AI character to automatically converse with the follower in the conversation service based on the conversation engine being pretrained. ([Hancock, page 3] “The chatbot is first trained with any available supervised data…In the deployment phase, the agent engages in multi-turn conversations with users…the bot continues the conversation with its own response” AND [Hancock, page 3, sec. 3] “In the initial training phase, the dialogue agent is trained on two tasks—DIALOGUE (next utterance prediction, or what should I say next?) and SATISFACTION (how satisfied is my speaking partner with my responses?)—using whatever supervised training data is available.” AND [Hancock, page 3, sec. 3] “If the satisfaction score is above a specified threshold t, the agent extracts a new Human-Bot (HB) DIALOGUE example using the previous context x and the human’s response y and continues the conversation.”, wherein the examiner interprets the chatbot being first trained with available supervised data to be the same as the conversation engine being pretrained because they are both directed to training a conversational model before use in a deployed conversation environment, and further interprets the agent engaging in multi-turn conversations with users and the bot continuing the conversation with its own response to be the same as the automatic response function allowing automatic conversation with the follower in the conversation service because they are both directed to a chatbot/agent automatically generating its own responses during an ongoing user conversation.)
Claims 28 and 30-31 are rejected under 35 U.S.C. 103 as being unpatentable over Schmitt in view of the NPL reference “A persona-based neural conversation model” by Li et al. (referred herein as Li) further in view of Reddy.
Regarding claim 28, Schmitt teaches:
An artificial intelligence (AI) character creation method comprising: ([Schmitt, page 1] “We present CharacterChat, a concept and chatbot to support writers in creating fictional characters. Concretely, writers progressively turn the bot into their imagined character through conversation.”, wherein the examiner interprets a chatbot supporting writers in creating fictional characters and turning a bot into an imagined character through conversation to be the same as an artificial intelligence character creation method because they are both directed to creating an AI/chatbot character through user interaction.)
graphically displaying and rendering, by an electronic device, a user interface comprising a character introduction window for setting introduction contents that introduce an AI character to be created based on an input of a user; ([Schmitt, page 1] “Our prototype combines two modes: (1) Guided prompts help writers define character attributes (e.g. User: “Your name is Jane.”), including suggestions for attributes (e.g. Bot: “What is my main motivation?”) and values” AND [Schmitt, page 3] “writers can use the bot both to define the character (e.g. User: ‘Your name is Jane.’) and to get inspiration” AND [Schmitt, page 5, sec. 3.1] “Fig. 1 shows the user interface of our prototype, implemented as a web app. It has three main parts: a chat view for the conversation, a character view that shows the attributes defined so far, and a pinboard view that shows dialogue pieces that the user has “bookmarked” in the chat view.” AND [Schmitt, page 6, sec. 3.2.3] “Conceptually, with this model, we condition text generation both on the conversation in the chat view (Fig. 1a) and the attribute statements in the character view (Fig. 1b).”, wherein the examiner interprets a user input defining character attributes such as the character name to be the same as setting introduction contents that introduce an AI character to be created based on an input of a user because they are both directed to receiving user-provided descriptive character information used to define the AI character. The examiner further interprets the character view of the web app user interface that shows the attributes defined so far to be the same as a user interface comprising a character introduction window for setting introduction contents because they are both directed to a displayed area of the interface that presents the descriptive information the user enters to introduce the character.)
analyzing the introduction contents ([Schmitt, page 3] “This model takes into account the character information elicited through the prompts so far”, wherein the examiner interprets taking into account character information elicited through prompts to be the same as analyzing the introduction contents because they are both directed to processing user-provided character/persona information.)
creating the AI character to which the determined speech tone and conversation contents of interest are assigned. ([Schmitt, page 1] “writers progressively turn the bot into their imagined character through conversation” AND [Schmitt, page 3] “Open Conversation Mode: For exploring the character, a Deep Learning dialogue model enables open conversations with the writer. This model takes into account the character information elicited through the prompts so far”, wherein the examiner interprets progressively turning the bot into an imagined character and using a deep learning dialogue model that takes into account elicited character information to be the same as creating the AI character to which the speech tone and conversation contents of interest are assigned because they are both directed to forming an AI character whose conversation behavior is based on assigned character/persona information.)
Schmitt does not teach graphically displaying and rendering, by the electronic device, the user interface further comprising a matter-of-interest window for setting at least one tag topic to be assigned to the AI character;…and determining a speech tone and conversation contents of interest based on at least one of the introduction contents and the at least one tag topic;.
Li teaches and determining a speech tone and conversation contents of interest based on at least one of the introduction contents and the at least one tag topic; ([Li, page 1] “A speaker model encodes personas in distributed embeddings that capture individual characteristics such as background information and speaking style … A persona can be viewed as a composite of elements of identity (background facts or user profile), language behavior, and interaction style”, wherein the examiner interprets taking into account character information provided by the user and encoding persona information including speaking style, language behavior, and interaction style to be the same as analyzing the introduction contents and determining a speech tone and conversation contents of interest to be assigned because they are both directed to processing user-provided information to determine how the AI character speaks and what conversational character information governs the AI character’s responses. The examiner further interprets the persona being a composite of background facts or user profile from which the speaking style is captured to be the same as determining the speech tone and conversation contents of interest based on at least one of the introduction contents and the at least one tag topic because they are both directed to deriving how the character speaks from descriptive information that introduces the character.)
Schmitt and Li do not teach graphically displaying and rendering, by the electronic device, the user interface further comprising a matter-of-interest window for setting at least one tag topic to be assigned to the AI character;.
Reddy teaches graphically displaying and rendering, by the electronic device, the user interface further comprising a matter-of-interest window for setting at least one tag topic to be assigned to the AI character; ([Reddy, [0050]] “During display operation 410, a graphical user interface screen with a script editor can be displayed on a display device.” AND [Reddy, [0043]] “As illustrated in the embodiments shown in FIG. 2, a main window in script editor 200 can include a conversation rule editor 210, a category list 220, and a chat window 230.” AND [Reddy, [0044]] “Category list 220 allows all of the content for a single character to be split across any number of categories, such as introductions, games, stories, math, etc.”, wherein the examiner interprets the category list displayed in the main window of the script editor, which is shown as a graphical user interface screen on a display device and through which the content of a character is split into categories such as games, stories, and math, to be the same as a matter-of-interest window for setting at least one tag topic to be assigned to the AI character because they are both directed to a displayed area of the interface through which subject-matter topics are assigned to the conversational content of a character.)
Schmitt, Li, Reddy, and the instant application are analogous art because they are all directed to creating or configuring an artificial intelligence character or persona-based conversational agent.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the artificial intelligence character creation method of setting contents and creating an AI character/persona disclosed by Schmitt to include the persona encoder disclosed by Li. One would be motivated to do so to effectively determine and apply character-specific speech style and persona-based conversational behavior to the AI character, as suggested by Li ([Li, Abstract] “Our models yield qualitative performance improvements in both perplexity and BLEU scores over baseline sequence-to-sequence models, with similar gains in speaker consistency as measured by human judges.”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the “category list 220” disclosed by Reddy. One would be motivated to do so to efficiently organize the AI character’s conversational content according to selected topics, as suggested by Reddy ([Reddy, [0044]] “Category list 220 allows all of the content for a single character to be split across any number of categories, such as introductions, games, stories, math, etc.”).
Regarding claim 30, Schmitt, Li, and Reddy teach The AI character creation method of claim 28 (see rejection of claim 28).
Li further teaches wherein the creating comprises creating the AI character by combining the speech tone and conversation contents of interest in a preset conversation engine. ([Li, page 1] “A speaker model encodes personas in distributed embeddings that capture individual characteristics such as background information and speaking style” AND [Li, page 3] “speaker information is encoded and injected into the hidden layer at each time step and thus helps predict personalized responses throughout the generation process”, wherein the examiner interprets encoding personas that capture background information and speaking style, and encoding and injecting speaker information into the hidden layer to predict personalized responses, to be the same as creating the AI character by combining the speech tone and conversation contents of interest in a preset conversation engine because they are both directed to incorporating character/persona information, including speaking style and background or interest-based conversational information, into an existing neural conversation model so that the model generates personalized character-specific responses.)
Schmitt and Li do not teach or by adding the speech tone and conversation contents of interest to the preset conversation engine.
Reddy teaches or by adding the speech tone and conversation contents of interest to the preset conversation engine. ([Reddy, [0005]] “Systems and methods are described for creating interactive synthetic characters using an artificial intelligence Scripting tool” AND [Reddy, [0007]] “A traversable script can be generated from the set of conversation rules. When executed by the synthetic character, the Script allows for dynamic interactions that can depend on multiple variables and/or change over time.”, wherein the examiner interprets creating interactive synthetic characters using an artificial intelligence scripting tool and generating a traversable script from conversation rules that is executed by the synthetic character to allow dynamic interactions, to be the same as adding the speech tone and conversation contents of interest to the preset conversation engine because they are both directed to adding user-defined conversational rules, responses, and character behavior information to an artificial intelligence/synthetic-character engine that controls the character’s conversational interactions.)
Schmitt, Li, Reddy, and the instant application are analogous art because they are all directed to creating or configuring an artificial intelligence character/conversational agent.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the artificial intelligence character creation method of claim 28 disclosed by Schmitt, Li, and Reddy to include the persona encoder disclosed by Li. One would be motivated to do so to effectively determine and apply character-specific speech style and persona-based conversational behavior to the AI character, as suggested by Li ([Li, Abstract] “Our models yield qualitative performance improvements in both perplexity and BLEU scores over baseline sequence-to-sequence models, with similar gains in speaker consistency as measured by human judges.”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the artificial intelligence character creation method of claim 28 disclosed by Schmitt, Li, and Reddy to include the scripting tool disclosed by Reddy. One would be motivated to do so to efficiently add user-defined conversation rules and character behavior information to a conversational character engine so that the resulting AI character can perform dynamic character-specific interactions, as suggested by Reddy ([Reddy, [0007]] “When executed by the synthetic character, the script allows for dynamic interactions that can depend on multiple variables and/or change over time.”)
Regarding claim 31, Schmitt, Li, and Reddy teach The AI character creation method of claim 28 (see rejection of claim 28).
Reddy teaches wherein the at least one tag topic comprises one of coin, stock, game, and alcohol. ([Reddy, [0044]] “Category list 220 allows all of the content for a single character to be split across any number of categories, such as introductions, games, stories, math, etc.”, wherein the examiner interprets the games category of the content of a single character to be the same as the at least one tag topic comprising game because they are both directed to a game subject assigned to the conversational content of a character.)
Schmitt, Li, Reddy, and the instant application are analogous art because they are all directed to creating or configuring an artificial intelligence character or persona-based conversational agent.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the artificial intelligence character creation method of claim 28 disclosed by Schmitt, Li, and Reddy to include the “games” category disclosed by Reddy. One would be motivated to do so to efficiently organize the AI character’s conversational content according to a game topic, as suggested by Reddy ([Reddy, [0044]] “Category list 220 allows all of the content for a single character to be split across any number of categories, such as introductions, games, stories, math, etc.”).
Claim 29 is rejected under 35 U.S.C. 103 as being unpatentable over Schmitt in view of Li in view of Reddy in view of US 20180300414A1 by Chen et al. (referred herein as Chen) and further in view of US 10417260B2 by Kikin-Gil et al. (referred herein as Kikin-Gil).
Regarding claim 29, Schmitt, Li, and Reddy teach The AI character creation method of claim 28 (see rejection of claim 28).
Schmitt, Li, and Reddy do not teach further comprising: assigning at least one hash tag to the AI character wherein the at least one hash tag is associated with the tag topic assigned to the AI character; and providing the user interface for searching an AI character by one or more hash tags.
Chen teaches:
further comprising: assigning at least one hash tag to the AI character wherein the at least one hash tag is associated with the tag topic assigned to the AI character; and ([Chen, [0023]] “Developers are empowered to label and categorize their bots’ experiences via a taxonomy to improve discovery and search.” AND [Chen, [0079]] “In some embodiments, a bot’s category may be assigned by a bot developer for the bot.”, wherein the examiner interprets the bot developer labeling and categorizing the bot via a taxonomy, including assigning a category to the bot, to be the same as assigning at least one tag to the AI character that is associated with the tag topic assigned to the AI character because they are both directed to the creator attaching a topic label to a conversational agent so that the agent is associated with that subject matter.)
providing the user interface for searching an AI character by one or more hash tags ([Chen, [0052]] “A search interface may comprise a search field 310, which may comprise a complete or partial entry of text.” AND [Chen, [0053]] “Alternatively, in some embodiments, a search interface may be provided that is exclusive to bots in which all search results 320 will be bots.” AND [Chen, [0058]] “Selecting a category entry on the bot category listing 340 opens a view of bots within that category.”, wherein the examiner interprets the search interface exclusive to bots, together with selecting a category entry to open a view of the bots within that category, to be the same as providing the user interface for searching an AI character by one or more tags because they are both directed to an interface through which a user locates conversational agents by their assigned topic labels.)
Schmitt, Li, Reddy, and Chen do not teach [that the tag is a] hash tag.
Kikin-Gil teaches hash tag ([Kikin-Gil, page 13, col. 3, lines 36-39] “If the user chooses to track a new topic (identified as the hash tag followed by the topic (e.g., one or more descriptive and identifying words)), the selected new topic can be added to the user’s topics page 204.” AND [Kikin-Gil, page 13, col. 3, lines 17-19] “The hash tag 104 can be searched from the personal feed page 114 and tracked for all instances by the tracking component 102.” AND [Kikin-Gil, page 13, col. 3, lines 22-25] “The system 200 includes the feed page 114 from which a search can be performed. The search results 202 list tags, one or more of which can be selected for following (tracking).”, wherein the examiner interprets the hash tag followed by one or more descriptive and identifying words of a topic, which can be searched from a page of the user interface, to be the same as a hash tag because they are both directed to a searchable topic label formed of a hash symbol followed by descriptive words.)
Schmitt, Li, Reddy, Chen, Kikin-Gil, and the instant application are analogous art because they are all directed to associating topic labels with conversational content or agents and enabling users to locate that content through a user interface.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the artificial intelligence character creation method of claim 28 disclosed by Schmitt, Li, and Reddy to include the “label and categorize their bots” technique disclosed by Chen. One would be motivated to do so to effectively improve how users discover a desired AI character, as suggested by Chen ([Chen, [0023]] “Developers are empowered to label and categorize their bots’ experiences via a taxonomy to improve discovery and search.”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the “hash tag followed by the topic” disclosed by Kikin-Gil. One would be motivated to do so to efficiently allow the topic labels of AI characters to be searched and tracked, as suggested by Kikin-Gil ([Kikin-Gil, page 13, col. 3, lines 17-19] “The hash tag 104 can be searched from the personal feed page 114 and tracked for all instances by the tracking component 102.”).
Claim 32 is rejected under 35 U.S.C. 103 as being unpatentable over Hancock in view of Schmitt in view of Kannan in view of Foster and further in view of the NPL reference “Learning to speak and act in a fantasy text adventure game” by Urbanek et al. (referred herein as Urbanek).
Regarding claim 32, Hancock, Schmitt, Kannan, and Foster teach The AI conversation engine training method of claim 16 (see rejection of claim 16).
Hancock, Schmitt, Kannan, and Foster do not teach wherein the personality comprises one of a caring type, a bad boy type, and a courageous type.
Urbanek teaches wherein the personality comprises one of a caring type, a bad boy type, and a courageous type. ([Urbanek, page 4, sec. 3] “Each character has a textual description, a persona (defined as a set of 3-5 profile sentences describing their traits, modeled after the Persona-Chat dataset (Zhang et al., 2018)), and a set of objects that are currently being carried, wielded, or worn.” AND [Urbanek, page 5, Fig. 1] “I am brave and fearless.” AND [Urbanek, page 8, sec. 5.4] “The persona features were found to be impactful, which makes sense as they shape the things the character says (and does).”, wherein the examiner interprets the character persona describing the character’s traits as brave and fearless, which shapes what the character says, to be the same as the personality comprising a courageous type because brave, fearless, and courageous are all directed to a personality trait of facing danger without fear.)
Hancock, Schmitt, Kannan, Foster, Urbanek, and the instant application are analogous art because they are all directed to artificial intelligence conversation systems that generate responses for a conversational character.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the AI conversation engine training method of claim 16 disclosed by Hancock, Schmitt, Kannan, and Foster to include the “brave and fearless” persona disclosed by Urbanek. One would be motivated to do so to effectively give the AI character a distinct personality that shapes its conversational responses, as suggested by Urbanek ([Urbanek, page 8, sec. 5.4] “The persona features were found to be impactful, which makes sense as they shape the things the character says (and does).”).
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DEVAN KAPOOR whose telephone number is (703)756-1434. The examiner can normally be reached Monday - Friday: 9:00AM - 5:00 PM EST (times may vary).
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, David Yi can be reached at (571) 270-7519. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/DEVAN KAPOOR/Examiner, Art Unit 2126
/DAVID YI/Supervisory Patent Examiner, Art Unit 2126