DETAILED ACTION
Introduction
1. This office action is in response to Applicant’s submission filed on 11/25/2024. Claims 1-20 are pending in the application and have been examined.
Notice of Pre-AIA or AIA Status
2. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Drawings
3. The drawings filed on 11/25/2024 have been accepted and considered by the Examiner.
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.
4. Claims 1-4, 8-14, 18-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claims 1, 11, and 20, the claim(s) recite(s) “receiving, at a device, a natural language input for processing by a first agent executed by the device; generating, by the first agent and based on the natural language input, an output in a first embedding space using an artificial intelligence model; identifying, by the first agent, a second agent for further processing of the output; and providing, by the first agent, the output for further processing by a second agent.”
These may be practically performed in the human mind using pen and paper. For example, limitation (a) can be done by evaluation and judgement, where person A can receive natural language input in the form of a verbal question at a device from person B, person A can assign or identify person C who would start working on generating an answer to the question by verbally asking support and be provided to person B and eventually verbally offered to person A.
Under its broadest reasonable interpretation when read in light of the specification, the actions “receiving…generating…identifying…providing” encompasses mental processes practically performed in the human mind. Accordingly, the claim recites an abstract idea (Step 2A, Prong one).
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of (a) “…one or more network interfaces…a processor…a memory…a device…a first embedding space using an artificial intelligence model…” which are mere data gathering and output recited at a high level of generality, and thus are insignificant extra-solution activity. See e.g., MPEP 2106.05(g) (“whether the limitation is significant”). Further, limitation (a) is recited as being performed by a computing device potentially including at least one processor (e.g., mobile device). The computing device is described as any generic computer device (Specification, see e.g., pages 3-4, 13, 14 “…an example node/device 200 (e.g., an apparatus) that may be used with one or more implementations described herein, e.g., as any of the devices shown in FIG. 1 above. Device 200 may comprise one or more network interfaces, such as interfaces 210 (e.g., wired, wireless, network interfaces, etc.), at least one processor (e.g., processor 220), and a memory 240 …”; “…artificial intelligence model is a large language model (LLM)…”). As such, the processor (e.g., device) is recited at a high level of generality. In limitation (a), the computing device potentially including at least one processor is used to perform an abstract idea, as discussed above in Step 2A, Prong One, such that it amounts to no more than mere instructions to apply the exception using a generic computer. See e.g., MPEP 2106.05(f). Even when viewed in combination, these additional elements do not integrate the recited judicial exception into a practical application (Step 2A, Prong Two: NO), and the claim is directed to the judicial exception. (Step 2A: YES).
The claim does not include additional elements that are sufficient to amount to more than the judicial exception. As discussed above, the recitation of a computing device including at least one processor to perform limitation (a) amounts to no more than mere instructions to apply the exception using a generic computer component. Limitation (a) is/are considered mere data gathering and output, and are additionally well-understood, routine, conventional activity. See e.g., MPEP 2106.05(d) and 2106.07(a)III. Even when considered in combination, these additional elements represent mere instructions to implement an abstract idea or other exception on a computer, which do/does not provide an inventive concept (Step 2B).
Accordingly, claims 1, 11, and 20 are directed towards patent ineligible subject matter under 35 U.S.C. 101.
The remaining dependent claims fail to add patent eligible subject matter to their respective parent claims:
Claims 2, 12 regard a human performing interactive data gathering by observing statistical representations and preprocessing information with respect to asked questions and answers shared between person A, B, and C (e.g., converting, by the device, the natural language input into the first embedding space for input to the artificial intelligence model).
Claims 3, 4, 13, 14 regard consulting an available artificial intelligence model such as a large language model (LLM) and also using a device receiving natural language inputs via a user interface evaluating further background about intent/purpose of the question/answers.
Claims 8, 9, 10, 18, 19 regard wherein the second agent provides a result of its further processing of the output back to the first agent; wherein the first agent generates the output further using one or more tool interfaces; and wherein the second agent returns a result of its further processing of the output back to a user interface through multiple verbal user inputs (e.g., questions/answers provided), mentally processing text from verbal user inputs received for understanding the claimed interactive information via interface tools based upon knowledge of a natural language, and mentally evaluating the intent/purpose of the multiple verbal user inputs received and transmitted between persons A, B, and C.
Accordingly, dependent claims 2, 3, 4, 8, 9, 10, 12, 13, 14, 18, and 19 are directed towards patent ineligible subject matter under 35 U.S.C. 101.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
5. Claim(s) 1-20 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Pham et al., (U.S. Patent Application Publication: 2024/0104125), cited in IDS filed 11/25/2024, and hereinafter referred to as PHAM.
With respect to Claim 1, PHAM discloses:
1. A method, comprising:
receiving, at a device, a natural language input for processing by a first agent executed by the device (See e.g., “…a conversational platform 110 may receive a question from a user 130. The question may be provided in a natural language…The conversational platform 110 includes one or more agents 120-1, 120-2,…,120-N (collectively or individually referred to as agents 120, where N indicates an integer larger than or equal to one) to process the question from the user 130 and provides a final answer 140… The agents 120-1, 120-2, . . . , 120-N are configured to use machine learning models 125-1, 125-2, . . . , 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” See e.g., PHAM, paras. 23, 24, Fig. 1);
generating, by the first agent and based on the natural language input, an output in a first embedding space using an artificial intelligence model (See e.g., “…and provides a final answer 140… The agents 120-1, 120-2,…,120-N are configured to use machine learning models 125-1, 125-2, . . . , 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” paras. 23, 24, 32-42, Figs. 1, 2);
identifying, by the first agent, a second agent for further processing of the output (See e.g., “…conversational platform 110 includes one or more agents 120-1, 120-2,…,120-N (collectively or individually referred to as agents 120, where N indicates an integer larger than or equal to one) to process the question from the user 130 and provides a final answer 140… The agents 120-1, 120-2,…,120-N are configured to use machine learning models 125-1, 125-2,…, 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2); and providing, by the first agent, the output for further processing by a second agent (See e.g., “…an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…” and also see e.g., how “…for the debating rounds, the multiple agents 120 (i.e., the model debaters) intake the embedded question, instruction, and (potentially) responses from the previous rounds to generate their embedding responses…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 2, PHAM discloses:
2. The method as in claim 1, wherein generating the output comprises: converting, by the device, the natural language input into the first embedding space for input to the artificial intelligence model (See e.g., “…convert the prompt (in the natural language) into embedding(s). First, an initial input includes a prompt “prompt” based on an answer from the user. Then the input is updated after debating with other agents, e.g., by considering answers from other agents, vocab(1), . . . vocab(l-1). The embedding outputs from the tokenizer 252 include emb(prompt), emb(vocab(1)), . . . , emb(vocab(l)), which are inputted into the agent 254, as a model input. The agent 254 provides probabilities of various embeddings (V embeddings corresponding to multiple vocabularies defined in a vocabulary table adopted by the agents),…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 3, PHAM discloses:
3. The method as in claim 1, wherein the artificial intelligence model is a large language model (LLM) (See e.g., “…implementations of the present disclosure, the LLMs may take a prompt (such as a query) as input and autoregressively generate a sequence of tokens according to the tokenizer as the response…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B)
With respect to Claim 4, PHAM discloses:
4. The method as in claim 1, wherein the device receives the natural language input via a user interface (See e.g., user interface capabilities with first and second agents (e.g., agent 120-1, and agent 120-2 )“…FIG. 3A, regarding a question 301 (also referred to as a query) in the natural language, an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 5, PHAM discloses:
5. The method as in claim 1, wherein providing the output for further processing by a second agent comprises: causing, by the first agent, the output to be transformed from being in the first embedding space to being in a second embedding space used by the second agent (See e.g., “…proposed an embedding communication protocol in a multiagent debate procedure…In a context of the present disclosure, “representation” is defined in a vector format (for example, embedding) that is understood by the models but has an invisible format to the outside human users. Therefore, the debate procedure is implemented in the embedding space within the conversational platform 110…” See e.g., PHAM, paras. 23, 24, 32-49, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 6, PHAM discloses:
6. The method as in claim 5, wherein causing the output to be transformed from being in the first embedding space to being in the second embedding space comprises: inputting the output to a transformation function executed by the device (See e.g., “…FIG. 3B, when generating a response, an agent 120 or the model 125 (e.g., a causal language model) autoregressively generates a token, for example, one at a time based on the words that come before each token. In a tokenizer, a vocabulary set…the tokenizer encodes the prompt and response(l-1) into embedding inputs for the agent. The agent then outputs a distribution p(l)=(p1 (l), . . . , pV (l))… implementations of CIPHER, instead of producing a single token vocab(l), CIPHER generates an average embedding vector emb utilizing p(l) as weights. Such an embedding vector bypasses the token decoding step and is fed directly back into the agent…” See e.g., PHAM, paras. 23, 24, 32-49, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 7, PHAM discloses:
7. The method as in claim 5, wherein causing the output to be transformed from being in the first embedding space to being in the second embedding space comprises: sending the output to a second device (See e.g., “…it is proposed to allow the models 125 to communicate in the form of embeddings during debates. With the embedding representations of the responses (referred to as “response embeddings”) from multiple models 125 (also referred to as debaters) collected, the response embeddings are feed into the debating procedure…” See e.g., PHAM, paras. 23, 24, 32-51, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 8, PHAM discloses:
8. The method as in claim 1, wherein the second agent provides a result of its further processing of the output back to the first agent (See e.g., “…for the debating rounds, the multiple agents 120 (i.e., the model debaters) intake the embedded question, instruction, and (potentially) responses from the previous rounds to generate their embedding responses…example algorithm 400 for multiagent debate via embeddings in accordance with some example implementations of the present disclosure…embedding communication protocol for debates among LLMs, to capitalize the rich information encoded in the belief Ideally, it is desired to encode as much belief information as possible during inter-LLM communication…” See e.g., PHAM, paras. 23, 24, 32-52, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 9, PHAM discloses:
9. The method as in claim 1, wherein the first agent generates the output further using one or more tool interfaces (See e.g., user interface capabilities with first and second agents (e.g., agent 120-1, and agent 120-2 )“…FIG. 3A, regarding a question 301 (also referred to as a query) in the natural language, an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 10, PHAM discloses:
10. The method as in claim 1, wherein the second agent returns a result of its further processing of the output back to a user interface (See e.g., user interface capabilities with first and second agents (e.g., agent 120-1, and agent 120-2 )“…FIG. 3A, regarding a question 301 (also referred to as a query) in the natural language, an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…”; and also see e.g., how “…for the debating rounds, the multiple agents 120 (i.e., the model debaters) intake the embedded question, instruction, and (potentially) responses from the previous rounds to generate their embedding responses…example algorithm 400 for multiagent debate via embeddings in accordance with some example implementations of the present disclosure…embedding communication protocol for debates among LLMs, to capitalize the rich information encoded in the belief Ideally, it is desired to encode as much belief information as possible during inter-LLM communication…” See e.g., PHAM, paras. 23, 24, 32-52, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 11, PHAM discloses:
11. An apparatus, comprising: one or more network interfaces; a processor coupled to the one or more network interfaces and configured to execute one or more processes; and a memory configured to store a process that is executable by the processor, the process when executed configured to: receive, a natural language input for processing by a first agent executed by the apparatus(See e.g., “…a conversational platform 110 may receive a question from a user 130. The question may be provided in a natural language…The conversational platform 110 includes one or more agents 120-1, 120-2,…,120-N (collectively or individually referred to as agents 120, where N indicates an integer larger than or equal to one) to process the question from the user 130 and provides a final answer 140… The agents 120-1, 120-2, . . . , 120-N are configured to use machine learning models 125-1, 125-2, . . . , 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” See e.g., PHAM, paras. 23, 24, Fig. 1); generate, by the first agent and based on the natural language input, an output in a first embedding space using an artificial intelligence model (See e.g., “…and provides a final answer 140… The agents 120-1, 120-2,…,120-N are configured to use machine learning models 125-1, 125-2, . . . , 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” paras. 23, 24, 32-42, Figs. 1, 2); identify, by the first agent, a second agent for further processing of the output (See e.g., “…conversational platform 110 includes one or more agents 120-1, 120-2,…,120-N (collectively or individually referred to as agents 120, where N indicates an integer larger than or equal to one) to process the question from the user 130 and provides a final answer 140… The agents 120-1, 120-2,…,120-N are configured to use machine learning models 125-1, 125-2,…, 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2); and provide, by the first agent, the output for further processing by a second agent(See e.g., “…an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…” and also see e.g., how “…for the debating rounds, the multiple agents 120 (i.e., the model debaters) intake the embedded question, instruction, and (potentially) responses from the previous rounds to generate their embedding responses…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 12, PHAM discloses:
12. The apparatus as in claim 11, wherein the apparatus generates the output by: converting the natural language input into the first embedding space for input to the artificial intelligence model (See e.g., “…convert the prompt (in the natural language) into embedding(s). First, an initial input includes a prompt “prompt” based on an answer from the user. Then the input is updated after debating with other agents, e.g., by considering answers from other agents, vocab(1), . . . vocab(l-1). The embedding outputs from the tokenizer 252 include emb(prompt), emb(vocab(1)), . . . , emb(vocab(l)), which are inputted into the agent 254, as a model input. The agent 254 provides probabilities of various embeddings (V embeddings corresponding to multiple vocabularies defined in a vocabulary table adopted by the agents),…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 13, PHAM discloses:
13. The apparatus as in claim 11, wherein the artificial intelligence model is a large language model (LLM) (See e.g., “…implementations of the present disclosure, the LLMs may take a prompt (such as a query) as input and autoregressively generate a sequence of tokens according to the tokenizer as the response…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 14, PHAM discloses:
14. The apparatus as in claim 11, wherein the apparatus receives the natural language input via a user interface (See e.g., user interface capabilities with first and second agents (e.g., agent 120-1, and agent 120-2 )“…FIG. 3A, regarding a question 301 (also referred to as a query) in the natural language, an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 15, PHAM discloses:
15. The apparatus as in claim 11, wherein the apparatus provides the output for further processing by a second agent by: causing, by the first agent, the output to be transformed from being in the first embedding space to being in a second embedding space used by the second agent(See e.g., “…proposed an embedding communication protocol in a multiagent debate procedure…In a context of the present disclosure, “representation” is defined in a vector format (for example, embedding) that is understood by the models but has an invisible format to the outside human users. Therefore, the debate procedure is implemented in the embedding space within the conversational platform 110…” See e.g., PHAM, paras. 23, 24, 32-49, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 16, PHAM discloses:
16. The apparatus as in claim 15, wherein the apparatus causes the output to be transformed from being in the first embedding space to being in the second embedding space by: inputting the output to a transformation function executed by the apparatus(See e.g., “…FIG. 3B, when generating a response, an agent 120 or the model 125 (e.g., a causal language model) autoregressively generates a token, for example, one at a time based on the words that come before each token. In a tokenizer, a vocabulary set…the tokenizer encodes the prompt and response(l-1) into embedding inputs for the agent. The agent then outputs a distribution p(l)=(p1 (l), . . . , pV (l))… implementations of CIPHER, instead of producing a single token vocab(l), CIPHER generates an average embedding vector emb utilizing p(l) as weights. Such an embedding vector bypasses the token decoding step and is fed directly back into the agent…” See e.g., PHAM, paras. 23, 24, 32-49, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 17, PHAM discloses:
17. The apparatus as in claim 15, wherein the apparatus causes the output to be transformed from being in the first embedding space to being in the second embedding space by: sending the output to a second device (See e.g., “…it is proposed to allow the models 125 to communicate in the form of embeddings during debates. With the embedding representations of the responses (referred to as “response embeddings”) from multiple models 125 (also referred to as debaters) collected, the response embeddings are feed into the debating procedure…” See e.g., PHAM, paras. 23, 24, 32-51, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 18, PHAM discloses:
18. The apparatus as in claim 11, wherein the second agent provides a result of its further processing of the output back to the first agent (See e.g., “…for the debating rounds, the multiple agents 120 (i.e., the model debaters) intake the embedded question, instruction, and (potentially) responses from the previous rounds to generate their embedding responses…example algorithm 400 for multiagent debate via embeddings in accordance with some example implementations of the present disclosure…embedding communication protocol for debates among LLMs, to capitalize the rich information encoded in the belief Ideally, it is desired to encode as much belief information as possible during inter-LLM communication…” See e.g., PHAM, paras. 23, 24, 32-52, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 19, PHAM discloses:
19. The apparatus as in claim 11, wherein the first agent generates the output further using one or more tool interfaces (See e.g., user interface capabilities with first and second agents (e.g., agent 120-1, and agent 120-2 )“…FIG. 3A, regarding a question 301 (also referred to as a query) in the natural language, an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
With respect to Claim 20, PHAM discloses:
20. A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising: receiving, at the device, a natural language input for processing by a first agent executed by the device (See e.g., “…a conversational platform 110 may receive a question from a user 130. The question may be provided in a natural language…The conversational platform 110 includes one or more agents 120-1, 120-2,…,120-N (collectively or individually referred to as agents 120, where N indicates an integer larger than or equal to one) to process the question from the user 130 and provides a final answer 140… The agents 120-1, 120-2,…, 120-N are configured to use machine learning models 125-1, 125-2, . . . , 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” See e.g., PHAM, paras. 23, 24, Fig. 1); generating, by the first agent and based on the natural language input, an output in a first embedding space using an artificial intelligence model (See e.g., “…and provides a final answer 140… The agents 120-1, 120-2,…,120-N are configured to use machine learning models 125-1, 125-2, . . . , 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” paras. 23, 24, 32-42, Figs. 1, 2); identifying, by the first agent, a second agent for further processing of the output (See e.g., “…conversational platform 110 includes one or more agents 120-1, 120-2,…,120-N (collectively or individually referred to as agents 120, where N indicates an integer larger than or equal to one) to process the question from the user 130 and provides a final answer 140… The agents 120-1, 120-2,…,120-N are configured to use machine learning models 125-1, 125-2,…, 125-N(collectively or individually referred to as machine learning models 125 or models 125 for short), respectively, to process the natural language question. In some implementations, the models 125 may be language models, which are trained on large-scale corpus to learn the capability of understanding one or more types of natural languages. In some implementations, the models 125 may be generative models for content generation. The models 125 may be constructed based on various machine learning technologies, for example, the models 125 may be implemented based on Large Language Models (LLMs)…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2); and providing, by the first agent, the output for further processing by a second agent (See e.g., “…an agent 120-1 generates an initial response embedding 310, which is convertible into a natural language response (also referred to as an answer) 330, and an agent 120-2 also generates an initial response embedding 320, which is convertible into a natural language response 340…” and also see e.g., how “…for the debating rounds, the multiple agents 120 (i.e., the model debaters) intake the embedded question, instruction, and (potentially) responses from the previous rounds to generate their embedding responses…” See e.g., PHAM, paras. 23, 24, 32-42, Figs. 1, 2, 3A-B, 4A-B).
Conclusion
6. The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
PNG
media_image1.png
220
533
media_image1.png
Greyscale
Eloff et al., (K. M. Eloff and H. A. Engelbrecht, "Toward Collaborative Reinforcement Learning Agents that Communicate Through Text-Based Natural Language," 2021 (SAUPEC/RobMech/PRASA), Potchefstroom, South Africa, 2021, pp. 1-6), discloses an agentic architecture comprising, see e.g., “…communication between agents in collaborative multi-agent settings is in general implicit or a direct data stream. This paper considers text-based natural language as a novel form of communication between multiple agents trained with reinforcement learning. This could be considered first steps toward a truly autonomous communication without the need to define a limited set of instructions, and natural collaboration between humans and robots. Inspired by the game of Blind Leads, we propose an environment where one agent uses natural language instructions to guide another through a maze. We test the ability of reinforcement learning agents to effectively communicate through discrete word-level symbols and show that the agents are able to sufficiently communicate through natural language with a limited vocabulary. Although the communication is not always perfect English, the agents are still able to navigate the maze…” (See e.g., Eloff et al., Abstract, Fig. 1).
Please, see PTO-892 for more details.
7. Any inquiry concerning this communication or earlier communications from the examiner should be directed to Edgar Guerra-Erazo whose telephone number is (571) 270-3708. The examiner can normally be reached on M-F 7:30a.m.-5:00p.m. EST. If attempts to reach the examiner by telephone are unsuccessful, the examiner's supervisor, Bhavesh Mehta can be reached on (571) 272-7453. The fax phone number for the organization where this application or proceeding is assigned is (571) 273-8300.
8. 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.
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.
/EDGAR X GUERRA-ERAZO/ Primary Examiner, Art Unit 2656
Prosecution Insights