DETAILED ACTION
1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
2. In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
3. The following office action is a Final Office Action in response to the communications received on 08/04/2026.
Claims 1, 7, 8, 11 and 14 have been amended; claim 13 has been canceled. Therefore, claims 1-12 and 14-17 are currently pending in this application.
Response to Amendments
4. The amendment to each of claims 1, 7 and 11 is sufficient to overcome the objection to the claims set forth in the previous office action. Accordingly, the Office with draws the above objection.
Similarly, the amendments to each of claims 8 and 14 is sufficient to overcome the rejection set forth in the previous office action under section §112(b). Thus, the Office withdraws the above rejection.
Claim Rejections - 35 USC § 101
5. Non-Statutory (Directed to a Judicial Exception without an Inventive Concept/Significantly More)
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.
● 1-12 and 14-17 are rejected under 35 U.S.C.101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
The current claims fall within one of the four statutory categories of invention (MPEP 2106.03).
Step 2A [Wingdings font/0xE0] Prong One:
The claim(s) recite a judicial exception, namely an abstract idea, as shown below:
— Considering each of claims 1, 7 and 14 as the representative claim, the following claimed limitations recite an abstract idea:
I. Claim 1:
[collect] first data of a course material to be studied by [a] student for a specific course;
[collect] second data of exam relevant questions related to the course material to be studied by the student, a predetermined percentage of which is required to be answered by the student to pass an exam related to the course;
carry out an interactive dialogue with the student about the course material to be studied by the student and regarding the exam relevant questions, wherein the interactive dialogue includes a first answer given in response to a first question;
obtain first data; obtain second data; and use the first answer given by the student in the dialogue to generate a new question or answer in the dialogue to be answered or commented by the student based on at least the first data and the second data;
conduct] the interactive dialogue based on third data, such that the interactive dialogue deals with topics which are relevant to [personal] attributes [of the student] and covers the course material to be studied by the student and takes the exam relevant questions into account.
II. Claim 7:
[collect] data of a course material to be studied by the student for a specific course;
use the data to set up multiple problems to be solved in an exam;
determine the actual knowledge state the student;
select a problem generated dependent on the actual knowledge state [of] the student;
give quantitative feedback to the student in relation to his input solution of the problem;
give qualitative feedback to the student in relation to his input solution of the problem;
obtain a solution input from the student and provide quantitative feedback and qualitative feedback in response;
predict, based on a history of solutions input from the student, for each problem of a plurality of problems to be selected, an expected score of an input solution at a present date, and optionally predict for each problem of the plurality of problems to be selected, an expected score of an input solution at a future date by taking into account a history of the predicted expected score of the input solution at the respective present date; and
the predicted score at the present date or the predicted score at the future date is used to select an appropriate problem.
III. Claim 14:
[collect] data of a course material to be studied by the student for the specific course;
obtain the data; use the data to generate questions which may be potentially based on content of the course material;
use the data to generate a respective expected responses to each question based upon the content of the course material; and
create a data set where each question is assigned to a respective expected response;
the respective expected responses to the questions in the data set are reviewed by a human person having knowledge in the field of the specific course and approved or rewritten, so that the dataset comprises pairs of questions and related approved or rewritten answers;
analyze a question input from the student whether it has a similarity equal to or above a predetermined amount of similarity to at least one of the artificial questions;
generate, if the similarity is equal to or above the predetermined amount of similarity to at least one of the artificial questions, a response to the question, wherein the response is based on an approved or rewritten answer of the pairs of questions and related approved or rewritten answers, which corresponding question has the highest or at least a predetermined similarity
provide the answer to the student.
Thus, the limitations identified above recite an abstract idea since the limitations correspond to certain methods of organizing human activity, and/or mental processes, which are part of the enumerated groupings of abstract ideas identified according to the current eligibility standard (see MPEP 2106.04(a)).
I. Considering claim 1:
Regarding the abstract idea group certain methods of organizing human activity, each of claims 1-6 correspond to managing personal behavior. For instance, while considering (i) information regarding course material that the student is studying for a specific course, and (ii) information regarding exam relevant questions related to the course material, an interactive dialog is conducted with the student; wherein the student provides an answer in response to a question during interaction; and wherein, based on the answer that the student is providing, the student is presented with a new question or answer during the dialog, so that the student provides further answer or comment based on information regarding the course material and the exam relevant questions; and furthermore, such interaction proceeds based on the student’s personal attributes; so that the dialog is further conducted based on topics that are relevant to the student’s personal attributes, etc.
Similarly, given the limitations that recite the process of: carrying out an interactive dialog with a student about a course material and exam relevant questions; presenting—based on the student’s answer to a question—a new question or answer, which the student is required to provide answer or comment based on information regarding course material and exam relevant questions; presenting the interactive dialog based on the student’s personal attribute, etc., claims 1-6 also correspond to the group mental processes, which encompasses concepts that can be performed in the human mind and/or using a pen and paper (e.g., an evaluation, an observation, a judgment, etc.).
II. Considering claim 7:
Regarding the abstract idea group certain methods of organizing human activity, each of claims 7-13 correspond to managing personal behavior. For instance, based on the actual knowledge state of a student, a problem is selected and presented to the student; and furthermore, based on a solution that the student is providing to the problem, (a) a qualitative and quantitative feedback is presented to the user; and furthermore, (b) an appropriate problem is selected for the student based on predicted score of an input solution at the present date, after determining an expected/predicted score for each problem of a plurality of problems using history of solutions input from the student, etc.
Similarly, given the limitations that recite the concept of: determining the actual knowledge of a student; selecting a problem deepening on the actual knowledge state of the student; giving quantitative feedback to the student based on the student’s solution to the problem; giving qualitative feedback to the student based on the student’s solution to the problem; predicting, based on a history of solutions input
from the student, for each problem of a plurality of problems to be selected, an expected score of an input solution at a present date; selecting an appropriate problem based on the predicted score at the present date, etc., claims 7-13 also correspond the abstract idea group mental processes, which encompasses concepts that can be performed in the human mind and/or using a pen and paper (e.g., an evaluation, an observation, a judgment, etc.).
III. Considering claim 14:
Regarding the abstract idea group certain methods of organizing human activity, each of claims 14-17 correspond to managing personal behavior. For instance, questions that are potentially based on content for a course material, including respective expected responses to each question, are drafted as a set; and a subject matter expert (e.g., a teacher/instructor) reviews and approves or rewrites the answers, so that a certified set that comprises questions and related approved/rewritten answers is created; and accordingly, when a student is providing a question, an approved/rewritten answer, which corresponds to the question with the highest or predetermined degree of similarity, is selected and presented to the student based on the degree of similarity of the student’s question to at least one of the questions in the certified set above,, etc.
Similarly, given the limitations that recite the process of: generating questions that may be based on content of a course material; generating respective expected responses/answers to each question based on content of the course material; reviewing the answers to the questions and approving or rewriting the answers; analyzing whether a student’s question is similar to at least one of the generated questions; generating a response to the student’s question if the similarity is equal to or above a predetermined amount of similarity to at least one of the generated questions, etc., claims 7-13 also correspond the abstract idea group mental processes, which encompasses concepts that can be performed in the human mind and/or using a pen and paper (e.g., an evaluation, an observation, a judgment, etc.).
Step 2A [Wingdings font/0xE0] Prong Two:
The current claims recite additional element(s), wherein a computer-based system that executes artificial intelligence model is utilized to facilitate the recited functions/steps regarding: storing data (e.g. “a course database comprising . . . data of a course material . . . a question database comprising second data of exam relevant questions . . . percentage of which is required to be answered by the student to pass an exam related to the course”, “a personalization database comprising personal attributes of the student”); conducting input/output interaction (e.g., “a dialogue module which is configured to carry out an interactive dialogue . . . the interactive dialogue includes a first answer given in response to a first question”); analyzing the collected/stored data using an algorithm and further generate one or more relevant results (e.g., “an AI Bot module which is configured to: obtain first data from the course database; obtain second data from the question database . . . generate a new question . . . based on at least the first data and the second data”, “the AI Bot module generates the interactive dialogue . . . covers the course material . . . and takes the exam relevant questions into account”; “an AI Bot module which is configured to use the data to set up multiple problems . . . a knowledge state module which is configured to determine the actual knowledge state the student; a problem selection module which is configured to select a problem . . . a quantitative feedback module configured to give a quantitative feedback . . . a qualitative feedback module adapted to give a qualitative feedback . . . obtain a solution input from the student and provide quantitative feedback and qualitative feedback in response . . . the knowledge state module is adapted to predict, based on a history of solutions input from the student, for each problem of a plurality of problems to be selected, an expected score of an input solution at a present date . . . the predicted score at the present date or the predicted score at the future date is used by the problem selection module to select an appropriate problem”; “an AI Bot module which is configured to: obtain the data; use the data to generate artificial questions . . . use the data to generate a respective expected responses . . . create a data set where each artificial question is assigned to a respective expected response . . . the answers to the questions in the data set are reviewed by a human . . . so that the dataset comprises pairs of artificial questions and related approved . . . a paraphrase detector module . . . analyze a question input from the student . . . a response generation module which generates, if the similarity is equal to or above the predetermined amount of similarity to at least one of the artificial questions, a response to the question . . . the system in configured to provide the answer to the student”), etc.
However, the claimed additional element(s) fail to integrate the abstract idea into a patent-eligible practical application since the additional element(s) are utilized merely as a tool to facilitate the abstract idea. Accordingly, when each of the claims is considered as a whole, the additional element(s) fail to impose meaningful limits on practicing the abstract idea. For instance, when each of the claims is considered as a whole, none of the claims provides an improvement over the relevant existing technology.
The observations above confirm that the claims are indeed directed to an abstract idea.
Step 2B
Accordingly, when the claim(s) is considered as a whole (i.e., considering all claim elements both individually and in combination), the claimed additional elements do not provide meaningful limitations to transform the abstract idea into a patent eligible application of the abstract idea such that the claim(s) amounts to “significantly more” than the abstract idea itself (also see MPEP 2106). The claimed additional elements are directed to conventional computer elements, which are serving merely to perform conventional computer functions.
Thus, when each of the current claims is considered as a whole (e.g., see the discussion under Prong Two above about such consideration of the claim as a whole), none of the claims recites an element—or a combination of elements—directed to an inventive concept.
Note also that the utilization of the conventional computer/network technology to facilitate the presentation pertinent content/information to one or more users, including the process of implementing one or more artificial intelligence-based chatbots, which facilitate—via one or more dialogues—the presentation of pertinent educational materials to the user (e.g. a student), and/or the evaluation of the user’s skills with respect to one or more course materials, etc., is already directed to a well-understood, routine, conventional activity in the art (e.g., see US 2018/0130156; US 2017/0206797, etc.).
It is further worth noting—per the original disclosure—that the claimed invention is directed to a conventional and generic arrangement of the additional elements. For instance, per the original specification, the claimed interactive dialog is facilitated by a system that implements one or more commercially available conventional computing devices (e.g., a personal computer, a cellphone, etc.); and such computing device(s) communicates—over the conventional communication network (e.g., the Internet)—with one or more online servers (e.g., see [0128]; [0133], etc.).
In addition, the original specification already admits that the process of setting one or more computers with an artificial AI Bot is generally known; and thus, the skilled artisan can implement the software configuration using his/her technical skills ([0135]).
The observations above confirm that the current claimed invention fails to amount to “significantly more” than an abstract idea.
It is worth noting that the above analysis already encompasses each of the current dependent claims (i.e., claims 2-6, 8-12 and 15-17). Particularly, each of the dependent claims also fails to amount to “significantly more” than the abstract idea since each dependent claim is directed to a further abstract idea, and/or a further conventional computer element(s) utilized to facilitate the abstract idea.
Accordingly, the findings above demonstrate that none of the claims implements an element—or a combination of elements—directed to an inventive concept (e.g., none of the current claims is reciting an element—or a combination of elements—that provides a technological improvement over the existing/conventional technology).
Response to the Declaration under CFR 1.132
6. The declaration under CFR 1.132, which was filed on 08/04/2026, has been fully considered. However, the declaration fails to negate the Office’s findings at least for the following reasons:
(a) Firstly, regarding the Office’s eligibility analysis under section §101, the declarant asserts (see the 12th paragraph, emphasis added),
“. . . in my expert opinion, the claimed invention is not merely directed to the broad concept of organizing human activity or performing mental processes, but to particular technological solutions that implement specific, non-routine, and unconventional computer architectures to address concrete technical problems in computer-based educational systems. Further, even if any of the claimed inventions were characterized as directed to an abstract idea, the specific and unconventional combination of computer-implemented elements in each independent claim integrates any such abstract idea into a practical application and amounts to significantly more than the abstract idea itself, thereby providing a patent eligible invention” (emphasis added).
However, the above appears to be a generic assertion that fails to point out a claimed element (if any)—or a combination of claimed elements (if any)—that is assumed to be providing the “particular technological solutions” that the declarant is alleging. In fact, there appears to be no description regarding the alleged “concrete technical problems”, which the currently claimed (or the originally disclosed) system is assumed to be addressing.
Similarly, despite alleging “specific and unconventional combination of computer-implemented elements”, which purportedly “integrates any such abstract idea into a practical application and amounts to significantly more than the abstract idea itself”, the declarant fails to show any claimed/disclosed computer-implemented element—or any combination of claimed/disclosed elements—that allegedly integrates the abstract idea into a patent-eligible practical application (Prong Two of Step 2A). Of course, the declarant also fails to show a claimed/disclosed technological feature (if any)—or a combination of claimed/disclosed features (if any)—that allegedly implements an inventive concept that amounts to “significantly more” than the abstract idea (Step 2B).
It is worth noting that a given claim (e.g., see claim 1) may overcome the prior art; however, this does not necessarily mean that the claim is beyond well-understood, routine, conventional activity (hereinafter WRCA) in the art. This is because a claim that recites a new abstract idea (again see claim 1) easily overcomes the prior art despite relying merely on the conventional technology—i.e., the conventional and generic arrangement of the additional elements. In contrast, the WRCA test, under Step 2B, is evaluating the underlying technology, but not the abstract idea. Thus, none of the declarant’s theories is persuasive since the declarant is mistaking the abstract idea for the alleged “specific, non-routine, and unconventional computer architectures”, and/or the alleged “specific and unconventional combination of computer-implemented elements”.
(b) Regarding the alleged “technological problem”, which prior systems (systems prior to 2024) are assumed to be facing, the declarant asserts (emphasis added),
“At the time of the invention, June 26, 2024, conventional computer-based educational systems, including so-called massive open online courses, were configured to present information to students without adapting to the learning style and proclivities of individual students taking the course. This technological deficiency is specifically identified in the specification: systems implementing massive open online courses are usually configured to present information without adapting to the learning style and proclivities of individual students taking the course . . . a concrete technical limitation of the underlying computer system architecture . . . conventional educational software systems did not implement any mechanism by which a computer system could simultaneously access structured student attribute data, course content data, and exam question data across multiple concurrent user sessions, and dynamically integrate all three data sources in real time to generate adaptive instructional content tailored to the individual student . . . No prior art system implemented a machine learning architecture capable of automatically predicting a student's expected performance on specific exam problems at both present and future dates based on accumulated solution history, and using those automated predictions to dynamically select optimally calibrated problems for the student at scale across a continuously operating platform. No prior art system implemented a tiered natural language processing pipeline combining a trained paraphrase detection module operating against a human-curated question-answer knowledge base with a fallback response architecture grounded in course material to reduce hallucinations, together with automated indication of the origin of each response, to address the limitations of unverified AI-generated responses in a
computer-based educational context”
However, the declarant appears to be referring to a teaching/learning strategy, as opposed to a technological feature, in an attempt to substantiate the alleged technological improvement that the clamed/disclosed system is supposedly providing. In particular, the so-called “deficiency”, which the declarant and/or the specification is implying, appears to be related to a resource deficiency, as opposed to the alleged “technological deficiency”. For instance, an educational entity that faces shortage of resources may implement a fixed teaching strategy; and therefore, regardless of the particular learning style and/or preferences of each individual student, the above entity may provide the same educational material to each and every student in the same manner. In contrast, an educational entity with sufficient educational resources may implement a dynamic teaching strategy. Thus, while evaluating the learning profile of each individual student, this entity adapts the educational material to each individual student according to the particular learning style and/or preferences of the student. Note also that the fact above remains true regardless of whether one is considering a computer-based teaching or a traditional instructor-led teaching.
The observation above demonstrates that the main issue, which the declarant attempts to portray as a technological feature, is merely a teaching strategy that a given educational entity adopts based on resource availability. Of course, the inaccuracy regarding the alleged “technological deficiency” is further evident given the timeframe—namely, the year 2024—that the declarant is relying on as a reference. This is because even the technology available in the year 2020 was mature enough to incorporate dynamic and/or adaptive computer-based educational systems. Although a reference is not necessarily required to substantiate the above fact, the Office directs the declarant’s attention to Grau (US 2018/0130156). Grau discloses an exemplary existing computer-based educational system that is capable of determining and generating—based on information received from each user of a plurality of users—an individualized/customized or personalized course profile for each user (see [0042]). In particular, besides allowing the user to specify his/her specific preferences regarding the lesson material to be learnt and/or the time duration to achieve a desired goal (see [0060] to [0062]), the system implements a machine-learning or an artificial intelligence model that learns from the user’s responses and further personalizes the user’s course ([0065] to [0067]).
Of course, the evidence above effectively invalidates the declarant’s alleged “technological deficiency”, which the “conventional computer-based educational systems” are assumed to be facing. Accordingly, even assuming arguendo that the issue above—namely, the process of adapting the learning material to each individual student, is considered to be a technological issue, the declarant still fails to appreciate the existing computer technology; namely, the existing computer-based educational system. In particular, unlike the declarant’s assumptions, the existing computer-based educational system dynamically adapts—based on the profile of each student—the learning material to the student (i.e., the system provides the learning material to each individual student according to the specific learning style and/or preferences of that student, etc.).
Accordingly, the observation above confirms that the declarant is mistaking the abstract idea (a teaching strategy) for a technological feature. In particular, while relying on a teaching strategy, which is itself an abstract idea, the declarant is attempting to substantiate an alleged technological improvement.
In addition, the declarant appears to be relying on the specificity of the information, which the claimed (or the disclosed) system is manipulating and/or generating, to substantiate the alleged technological improvement. In contrast, the specific content of the information that the system is analyzing (e.g., the student attribute data, the course content data, the exam question data), and/or the specific content/result that the system is generating (e.g., a customized instructional content, predicted/expected performance result of a student, etc.), does not necessarily signify a technological improvement. This is again because the underlying technological process (e.g., the process of collecting information, analyzing the information using one or more algorithms, and generating one or more textual and/or audiovisual results, etc.), which the claimed—and the disclosed—system is implementing, is still part of the existing computer/network technology. Accordingly, it is immaterial whether the declarant assumes that “No prior art system implemented a machine learning architecture capable of automatically predicting a student's expected performance on specific exam problems at both present and future dates based on accumulated solution history, and . . . dynamically select optimally calibrated problems for the student at scale across a continuously operating platform” (emphasis assed). This is because the above does not signify a technological improvement (if any) that claimed/disclosed system is implementing; rather, it is signifying merely the data that the claimed/disclosed system is manipulating and/or generating using the existing computer/network technology. However, unlike the declarant’s theory, the mere difference in the information being manipulated/generated does not necessarily signify a technological difference, much less a technological improvement.
Of course, even the MPEP confirms the invalidity of the declarant’s theory above (see MPEP 2106.06 (I) cited below). Note that a given claim (e.g., current claim 1) may overcome the prior art; however, this does not necessarily mean that the claim is patent-eligible. This is because neither the novelty analysis (section §102) nor the obviousness analysis (section §103) necessarily influences the eligibility analysis (section §101). In particular, despite relying merely on the existing technology, the given claim (again see current claim 1) may overcome the prior art due to the new abstract idea it is reciting. Of course, due to the new abstract idea, the claim may be considered to be novel or non-obvious over the prior art. However, such alleged novelty or non-obviousness does not signify eligibility. This because a claim for a new abstract idea is still an abstract idea. Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151 (Fed. Cir. 2016). Also see MPEP 2106.06 (I) (emphasis added),
Although the courts often evaluate considerations such as the conventionality of an additional element in the eligibility analysis, the search for an inventive concept should not be confused with a novelty or non-obviousness determination. See Mayo, 566 U.S. at 91, 101 USPQ2d at 1973 . . . As made clear by the courts, the “‘novelty’ of any element or steps in a process, or even of the process itself, is of no relevance in determining whether the subject matter of a claim falls within the §101 categories of possibly patentable subject matter.” Intellectual Ventures I v. Symantec Corp., 838 F.3d 1307, 1315, 120 USPQ2d 1353, 1358 (Fed. Cir. 2016) . . . See also Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151, 120 USPQ2d 1473, 1483 (Fed. Cir. 2016) (“a claim for a new abstract idea is still an abstract idea. The search for a §101 inventive concept is thus distinct from demonstrating §102 novelty.”) . . . the search for an inventive concept is different from an obviousness analysis under 35 U.S.C. 103 . . . patentability of the claimed invention under 35 U.S.C.102 and 103 with respect to the prior art is neither required for, nor a guarantee of, patent eligibility under 35 U.S.C.101.
The observation above demonstrates that the declarant’s attempt to substantiate the alleged technological improvement, while emphasizing the lack of prior art that teaches/suggests the claimed system, is certainly not valid.
It is also noted that the declarant is relying on the inherent features of already developed AI algorithms to substantiate the alleged technological improvement. For instance, the developers (e.g. OpenAI) of Large Language Models (e.g., GPT-4, etc.) already established various techniques for mitigating the issue of hallucination that AI modles typically encounter during use. Some of these techniques involve, for example: (a) using Retrieval-Augmented Generation (RAG)—such as, a pipeline that essentially forces the AI model—when formulating an answer—to ground its answers in approved source documents (i.e. requiring it to use only human provided/approved documents to formulate a relevant answer), wherein the model is also instructed to say “I don’t know” if it is uncertain, so that it is refrained from guessing, (b) applying a Chain-of-Thought (CoT), wherein the model is required to outline its reasoning in a step-by-step fashion before providing a response, (c) lowering its temperature, which requires the model to choose the most statistically likely word (via greedy decoding), so that it is discouraged from using broader sampling and creative combinations, etc. It is worth noting that the developers also provide additional options for customizing such models according to the objectives and/or policies of service providers (e.g., call centers, etc.). For instance, when considering a call center that implements an AI agent/bot to handle customers, the call center typically customizes the AI agent/bot to route—to a human agent—a question (or an issue) that it cannot resolve. In fact, this has been a common practice even before the development of GPT-4.
Although the discussion above is not necessarily an exhaustive list, it shows the various options that the developers already provided in order to mitigate the issue of hallucinations in AI models. Accordingly, utilizing one or more of such options to reduce the chance of an AI model from hallucinating, etc., does not constitute a technological improvement. In particular, Applicant’s disclosed system relies on the existing Large Language Model (LLM)—such as, GPT-4—in order to implement an AI Bot that provides an answer/response to a student. In this regard, except for the generic description regarding the use of already developed options to mitigate the problem of hallucination, the specification does not provide any new/advanced technique that is assumed to be an improvement over the existing technology. For instance, the specification describes (see [0118], emphasis added),
“. . . the AI Bot module may be configured to provide the answer in an output depending on the conversation with the student using an LLM (e.g., GPT-4, to give an example). In embodiments, the input to the LLM may include large excerpts of the coursebook. An advantage to such an approach is that it grounds the model and reduces hallucinations. In this case, the (e.g. visual or audible) information that the answer has not been approved by a human person and is AI generated. In embodiments, such a response may be routed to a human person for verification . . .”
It is evident from the excerpt above that the specification does not even contemplate—much less implement—any advanced technique to mitigate the issue of hallucination; rather, it is relying merely on already developed techniques. Thus, the declarant’s conclusory assertions regarding the alleged “concrete technical problems rooted in the architecture of prior computer-based educational systems”, are not persuasive. In particular, the declarant is asserting that “person of ordinary skill in the art would understand that simultaneously cross-referencing multiple structured databases in real time across concurrent student sessions to generate dynamically adaptive natural language dialogue, automatically maintaining continuously updated probabilistic performance predictions across an entire problem bank for a given student, and implementing a tiered query routing architecture with automated provenance marking are complex technical capabilities that require specific and non-routine computer system architectures. None of these capabilities existed in prior art educational software systems” (emphasis added). Accordingly, it is evident from the conclusory assertion above that the declarant does not demonstrate whether the alleged “capabilities” above constitute a technological improvement. In fact, the declarant does not even refer to the claims (or the specification) in order to identify a claimed/disclosed feature (if any)—or a combination of features (if any)—that is assumed to be providing the alleged technological improvement. Instead, due to the lack of an advanced technological feature per the claimed and disclosed system, the declarant is relying on various prolix and grandiloquent assertions, which even the specification does not contemplate, in an attempt to portray the claimed/disclosed system as a technological marvel. Thus, the declarant’s assertions are not persuasive.
(c) In addition, while listing the limitations that claim 1 is reciting (see the 17th paragraph of the declaration), the declarant is attempting to substantiate an alleged technological improvement (emphasis added),
“[t]he technological improvement implemented by the system of claim 1 is a specific computer architecture in which an AI Bot module is configured to simultaneously access and integrate three distinct structured data sources - a course database, a question database, and a personalization database - in real time, and to use the combined output of that multi-source integration, together with the student's prior answer in the ongoing dialogue, to dynamically generate adaptive instructional content. This is not a mental process. No human teacher, however skilled, could simultaneously and in real time cross-reference a structured database of course material, a structured database of exam-relevant questions, and a structured database of personal student attributes including motivation for passing the course, professional background, personal interests, and admission documents, integrate all three data sources with the content of an ongoing dialogue, and dynamically generate contextually appropriate and personally adaptive instructional responses at the speed and scale required by a computer-based educational system serving multiple students. The specification confirms that the AI Bot module is configured to adapt the complexity and content of the dialogue based on one or more or all of those attributes, and that one or more of those attributes are used as part of prompts to a machine learning model of the AI Bot module along with an instruction to take the attributes into account. Specification, paragraph 0009”
However, unlike the declarant’s assertion, existing AI models already access multiple databases simultaneously (e.g., via already existing cloud data frameworks, etc.). Accordingly, neither the specific content that each database is storing (e.g., a course database, a question database, etc.), nor the specific topic of interaction that the model is considering (e.g., the student's prior answer in the ongoing dialogue, etc.) to generate the relevant information (e.g., the adaptive content), alone or in combination necessarily suggests a technological improvement. It is again worth noting that the content (e.g., the topic) of the information is merely an abstract idea. In contrast, while simply emphasizing the specific content of the information (e.g., the specific content that the different databases are storing, and/or the specific topic of dialogue that model is considering in order to generate the adaptive content), which the claimed/disclosed system is manipulating, the declarant is attempting to show an alleged technological improvement by labeling the features of the existing computer technology as “specific computer structure”; and therefore, the declarant’s assertions are not persuasive.
The declarant is also attempting to challenge the Office’s findings presented under Prong One—namely, the finding regarding mental processes—while relying on computer elements. In particular, the declarant is asserting that “No human teacher, however skilled, could simultaneously and in real time cross-reference a structured database of course material . . . generate contextually appropriate and personally adaptive instructional responses at the speed and scale required by a computer-based educational system serving multiple students” (emphasis added).
However, unlike the declarant’s theory above, the eligibility test regarding a mental process does not necessarily require one to evaluate whether a human can operate/function like a machine, regardless of whether the machine being considered is a computer-based educational system that executes an AI model. Instead, the test requires one to evaluate the claim and identify only the limitations that can be performed in the human mind and/or using a pen and paper. Thus, even if the claim includes limitations (i.e., additional elements) that are beyond a mental process, the test is concerned merely with the limitations that recite a mental process. For instance, when considering the limitation, “a course database comprising first data of a course material to be studied by the student for a specific course”, at least part of the limitation above—namely, “first data of a course material to be studied by the student for a specific course”, is indeed a mental process. This is because a teacher can readily create, at least using a pen and paper, data regarding a course material that a student is required to study for a specific course (e.g., a particular course that the teacher is teaching, etc.). In this regard, given the context of the claim, the claimed “course database” is assumed to be a computer structure (i.e., an additional element); and therefore, it is excluded from the abstract idea.
The observation above demonstrates the proper manner of executing the eligibility inquiry under Prong One of Step 2A—i.e., identifying the limitations that recite an abstract idea. In contrast, while conflating the computer elements with the abstract idea, the declarant is attempting to challenge the Office’s finding regarding mental processes. Consequently, the declarant’s arguments are not persuasive.
The declarant also appears to rely on the alleged skill improvement, which the student is supposedly acquiring as the result of interaction, in order to substantiate the alleged technological improvement. The declarant asserts (emphasis added),
“The specification further states that an advantage of the present invention is that personal attributes of the student are regarded in the dialogue, thereby improving the learning effect of the student, and that as the material to be studied is matched to the personal preferences of the student, due to his practical experiences, the learning effect is superior to a learning when the personal attributes and/or experiences are not regarded. Specification, paragraph 0073. Specifically, an AI Bot, using its natural language processing capability that is pre-trained on thousands of previous student interactions, can determine both the intent and sentiment of the student's response if given sufficient information. Therefore, by combining that AI Bot' s capability with the use of the non-traditional personal attributes data (e.g. motivation for passing, personal interests) it allows the dialog module to generate a reply that is more than just relevant to the course material. It allows the dialog module to use the student's intent and sentiment to generate an adaptive reply that will motivate and provide interest to the student based on those personal attributes”
However, neither the declarant’s assertion nor the section(s) that the declarant identified from the specification, alone or in combination, demonstrates any technological improvement. This is because it is irrelevant whether the claimed and/or the disclosed system is improving the student’s skill since improving the student’s skill has nothing to do with a technological improvement. In contrast, the presence of a technological improvement (if any) is established by identifying a structural and/or a functional feature—or a combination of structural and/or functional features—that is an advance over the existing computer/network technology.
Of course, besides the data stored in the course databases, the AI model may also consider the student’s profile (e.g., the student’s personal data, etc.) in order to adapt one or more of the responses that it is providing to the student. However, this is already part of the existing computer/network technology. Thus, the declarant is demonstrating, perhaps inadvertently, the benefit that the existing computer/network technology is providing to the student. In particular, the declarant is demonstrating the use of the existing computer/network technology—merely as tool—to facilitate an abstract idea—such as, facilitating the presentation of relevant information to a user based on the analysis of collected data (e.g., data related to the student, data related to course materials, etc.). Consequently, the alleged skill improvement, which the student is acquiring, does not necessarily signify a technological improvement. Thus, the declarant’s assertions are once again not persuasive.
The declarant also continues to emphasize the computer elements in further attempt to challenge the Offices findings under Prong One of Step 2A; namely, the Office’s finding regarding certain methods of organizing human activity and/or mental processes. The declarant asserts (emphasis added),
“[the Examiner’s] characterization fails to account for the specific technical architecture recited in the claim. The process of dynamically generating adaptive natural language dialogue by simultaneously integrating three distinct structured data sources in real time through a machine learning model is not a concept performable in the human mind or with a pen and paper. It is a specific technical capability that exists only within a computer-implemented architecture. Furthermore, the claimed system does not merely use a computer as a tool to perform a pre-existing mental process. The multi-source real-time data integration architecture is itself the technical improvement, enabling a category of adaptive instructional capability that did not exist in prior art educational software systems that presented information uniformly without adaptation”
However, as already pointed out above, the computer elements are excluded from consideration when properly applying the eligibility inquiry under Prong One of Step 2A. This is once again because none of the computer elements (e.g., the alleged process of “dynamically generating adaptive natural language dialogue by simultaneously integrating three distinct structured data sources in real time through a machine learning model”, the alleged “specific technical capability that exists only within a computer-implemented architecture”, etc.) is part of the abstract idea. So far, the declarant fails to challenge parts of the claim that Office positively identified as the limitations that recite the abstract idea. Instead, while simply listing the computer elements, which the Office never considered to be an abstract idea, the declarant is attempting to challenge the Office’s findings regarding mental processes and/or certain methods of organizing human activity. Consequently, the declarant’s arguments are not persuasive.
The declarant also appears to draft a new theory regarding the abstract idea group mental processes. For instance, when attempting to show the alleged eligibility of the claimed system, the declarant is asserting that “the claimed system does not merely use a computer as a tool to perform a pre-existing mental process” (emphasis added); however, it is immaterial whether the abstract idea (the mental process) is considered to be a preexisting one or a new one. Even assuming arguendo that the claimed abstract idea per each of the current claims is a new abstract idea, the alleged novelty of the abstract idea does not signify a technological improvement. This is once again because a claim for a new abstract idea is still an abstract idea. Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151 (Fed. Cir. 2016). Consequently, the declarant’s assertion is once again not persuasive.
Of course, the declarant’s conclusory assertion regarding the alleged “technical improvement” is also not persuasive. In particular, while mischaracterizing the existing technology as systems that “present[] information uniformly without adaptation”, the declarant is attempting to portray the claimed (or the disclosed) system as a technological improvement. However, quite contrary to the declarant’s conclusory assertion, the existing computer/network technology already adapts content items (e.g., educational materials) to each user based on the analysis of data (e.g., profile) of each user. Although a reference is not necessarily required to support the fact above, one of the references—i.e., Grau (US 2018/0130156)—discussed above already confirms this. For instance, Grau’s educational system not only gathers the user’s attributes—such as, the user’s specific preferences regarding the lesson material to be learnt and/or the time duration to achieve a desired goal (see [0060] to [0062]), but also learns—using an artificial intelligence model—from the user’s responses and further personalizes the user’s course ([0065] to [0067]).
In contrast, while simply dismissing existing computer-based educational systems as nonadaptive systems, the declarant is attempting to portray the claimed system as a technological improvement by labeling it as “multi-source real-time data integration architecture”. However, such labeling does not necessarily demonstrate a technological feature (if any)—or a combination of features (if any)—that provides a technological improvement. Consequently, the declarant’s assertion regarding the combination of the claimed features, “[t]he combination of the course database, question database, personalization database, dialogue module, and AI Bot module, operating together as an integrated system in the specific manner recited in claim 1” (emphasis added) fails to substantiate the declarant’s alleged “concrete technical improvement in the functioning of computer-based educational systems” (emphasis added). Thus, the declarant’s assertions are once again not persuasive.
(d) The declarant has also attempted to support the alleged eligibility of claim 7. In particular, while listing the limitations that the claim is reciting (see paragraph 22nd of the declaration), the declarant asserts, “[i]t should be understood that an AI Bot, with the requisite [natural] language processing, has already been trained on understanding and responding to thousands of [interactions] with the general population of students (i.e. using heuristic neural network training). Therefore, the AI Bot necessarily utilizes [its] previous trained knowledge of general students' population as a basis for evaluating a particular student's current and previous responses as well as predicted future responses.” (emphasis added).
Thus, the above appears to show the declarant’s admission regarding the existing technology that the claimed/disclosed system is implementing. This is because the original disclosure does not have any new/advanced implementation of a Natural Language Processing (NLP), and/or the use of heuristic neural network to train the claimed (or disclosed AI Bot). This is further evident given the fact that many AI-based chatbots, which are developed prior to the claimed/disclosed system, are trained using deep learning artificial neural networks (e.g., transformer-based LLM), which are more robust than the heuristic neural network that the declarant is specifying. Thus, the declarant is acknowledging, at least implicitly, the lack of technological improvement regarding the claimed/disclosed system.
In addition, while referring to part of the specification, the declarant appears to be attempting to support an alleged technological improvement regarding claim 7. The declarant asserts (emphasis added),
“[the] system is configured to obtain a solution input from the student and provide quantitative feedback and qualitative feedback in response, and wherein the knowledge state module is adapted to predict, based on a history of solutions input from the student, and previously trained responses from the general student population, for each problem of a plurality of problems to be selected, an expected score of an input solution at a present date, and optionally predict for each problem of the plurality of problems to be selected. These factors are used to produce an expected score of an input solution (quantitive feedback) at a future date by taking into account a history of the predicted expected score of the input solution at the respective present date, and wherein the predicted score at the present date or the predicted score at the future date is used by the problem selection module to select an appropriate problem”
However, the declarant is merely summarizing the existing computer functions, which the claimed/disclosed system is performing—such as, the process of gathering data related to the student (the student’s profile, the student’s response, etc.), and subsequently generating—based on the analysis of the gathered data—one or more pertinent results (e.g., one or more predicted scores, a relevant question, etc.). However, the above is indeed part of the existing computer/network technology; and therefore, the declarant still fails to demonstrate whether the claimed/disclosed system is implementing any technological improvement over the existing computer/network technology. Instead, the declarant is demonstrating the use of the existing computer technology—merely as a tool—to facilitate the abstract idea.
The declarant also continues to rely on the existing computer functions, which the claimed (and the disclosed) system is implementing, to support the alleged technological improvement. The declarant asserts (emphasis added),
“[t]he technological improvement implemented by the system of claim 7 is a specific machine learning architecture (i.e. neural network training) in which the knowledge state module does not merely track a student's historical performance, but generates forward-looking probabilistic estimates of the student's expected score on each problem across the entire available problem set, based on the accumulated history of prior solution inputs, and it is those predicted expected scores—rather than raw historical performance data—that are then consumed by the problem selection module as the direct basis for selecting the next appropriate problem for the student at each point in the training process . . . the problem selection module does not operate just on a particular students historical scores directly, but on predicted future performance estimates generated by the knowledge state module across the full problem set, enabling the system to select problems optimally calibrated not merely to where the student has been but to where the student's knowledge state is predicted to be, both at the present date and at a future date. Critically, the predictive scoring functionality is now expressly recited in the body of claim 7 itself: the knowledge state module is specifically configured to predict an expected score of the input solution at the present date for each problem in the problem bank, and optionally to predict an expected score at a future date by taking into account the history of the predicted expected scores at respective present dates, with those predicted scores used by the problem selection module to select an appropriate problem . . . the knowledge state module may be adapted to predict based on the history of the solutions input from the student for each problem out of the problems to be selected an expected score of the input solution at a preset date, and may also predict a score in the future by taking the history of the predicted expected score into account, with the predicted score at the present date or the predicted score at the future date used by the problem selection module to select an appropriate problem, such that better training goals and faster fulfillment of the requirements to pass an exam are obtained”
However, except for labeling the claimed/disclosed system as the so-called “specific machine learning architecture”, the declarant still fails to address the crux of the eligibility inquiry. In particular, the declarant fails to show a claimed/disclosed technological feature (if any)—or a combination of features (if any)—that provides a technological improvement over the existing computer/network technology. For instance, the use of a neural network to train an AI model, and/or the use of predicting algorithms to generate one or more predictive results, etc., are already part of the existing computer/network technology. Therefore, simply emphasizing such existing features, and/or labeling the claimed/disclosed system as the so-called “specific machine learning architecture”, does not constitute a technological improvement.
Similarly, simply emphasizing (i) the specific data that the claimed/disclosed system is analyzing (e.g., student’s historical performance, accumulated history of prior inputs, the history of predicted expected scores etc.), and/or (ii) the specific information that the claimed/disclosed system is generating to the user (e.g., estimates of the student's expected score on each problem, an expected score of the input solution at the present date for each problem in the problem bank, an appropriate problem to the student, etc.), also do not constitute a technological improvement. This is again because the information that the claimed/disclosed system is gathering as input, and/or the information that the system is generating as an output, is merely an abstract idea.
Note also that simply assigning one or more labels—such as, the so-called “problem selection module”, and/or “knowledge state module”, to one or more of the exiting computer functions above does not constitute a technological improvement. This is certainly because such labeling has nothing to do with technology, much less an advanced technology.
The observations above confirm that the declarant still fails to demonstrate whether the currently claimed (or originally disclosed) system is implementing any technological feature—or any combination of technological features—that provides a technological improvement. Instead, the declarant is attempting to show the alleged technological improvement by emphasizing the specificity of the student/course data that the system is analyzing, and/or the specificity of the information (e.g., predictive scores, relevant problem, etc.) that the system is proving to the student. Consequently, the declarant still fails to demonstrate whether the claimed (or the disclosed) system is “significantly more” than an abstract idea.
The declarant has also once again failed to properly apply the eligibility inquiry regarding a mental process. The declarant asserts (emphasis added),
“. . . No human educator could simultaneously maintain continuously updated probabilistic expected score predictions across an entire bank of exam problems for a given student, that is evaluated against the AI Bot's previous heuristic training, and where each prediction is based on the full accumulated history of that student's prior solution inputs. This information is then updated in real time after each interaction, and further used to generate forward-looking predictions of expected scores at future dates by taking into account the history of those present date predictions, all while dynamically selecting the next problem with optimal calibration to the student's current and predicted future knowledge state. The layered predictive architecture recited in the amended claim—present-date expected score prediction, future-date expected score prediction based on the history of present-date predictions, and dynamic problem selection based on either predicted score—is a specific and non-routine machine learning feedback architecture of a kind that has no human cognitive equivalent and that exists only within a specifically architected computer-implemented system”
However, as already pointed out above, the test for a menta process does not necessarily require one to evaluate whether a human can operate as a computer-based system. Instead, the test requires one only to determine whether the claim has limitations that recite an abstract idea—such as a mental process (i.e., limitations that can be performed in the human mind and/or using a pen and paper). For instance, current claim 7 recites the following limitations, “a course database comprising data of a course material to be studied by the student for a specific course; an AI Bot module which is configured to use the data to set up multiple problems to be solved in an exam” (emphasis added).
Thus, when considering the exemplary limitations above, the eligibility inquiry requires one to identify only the limitations that recite the abstract idea. In particular, while excluding the computer elements (e.g., the course database and the AI Bot module), the test requires one to identify the limitations that can be performed in the human mind and/or using a pen and paper. In this regard, the limitations, “data of a course material to be studied by the student for a specific course” and “[using] the data to set up multiple problems to be solved in an exam”, do recite a mental process. This is because a human—such as a teacher—can easily outline/create, using a pen and paper, data regarding a course material that a student is required to study for a specific course; and furthermore, once outlining the course material above, the teacher uses the data to create various problems that are going to be presented as part of an exam, which the student may take at some point.
The observation above is itself sufficient to confirm the fact that claim 7 does recite an abstract idea—such, as a mental process. Of course, similar type of observation is made with respect to the rest of the claimed limitations that recite an abstract idea. For instance, consider the following limitation, “the knowledge state module is adapted to predict, based on a history of solutions input from the student, for each problem of a plurality of problems to be selected, an expected score of an input solution at a present date, and optionally predict for each problem of the plurality of problems to be selected, an expected score of an input solution at a future date by taking into account a history of the predicted expected score of the input solution at the respective present date, and wherein the predicted score at the present date or the predicted score at the future date is used by the problem selection module to select an appropriate problem” (emphasis added).
Accordingly, while excluding the computer elements—such as, the so-called “knowledge state module” and the so-called “problem selection module”, the teacher can still perform—mentally and/or using a pen and paper—the limitations that recite the abstract idea. For instance, based on evaluating the student’s record, which shows the solutions that the student provided in the past when answering each of one or more problems, the teacher can estimate—e.g., using a prediction formula—an expected score of an input solution at a present date for each problem of a plurality of problems to be selected. Although the claimed optional feature (“optionally predict for each problem . . .”) is not necessarily required, the teacher can still predict, based on considering the history of the predicted expected score of an input solution at the respective present date, an expected score of the input solution at a future date. Of course, once obtaining one or both of the above scores/values, the teacher uses any one of the above values to select an appropriate problem for the student (e.g., using a table that correlates each of one or more scores with one or more corresponding problems, etc.).
The above is yet another evidence, which confirms that the claim does recite an abstract idea—namely, a mental process. In contrast, while repeatedly mistaking the above abstract idea for a technological feature, the declarant is attempting to portray an alleged technological improvement while simply labeling the claimed system as the so-called “specific and non-routine machine learning feedback architecture”. Accordingly, the declarant’s theory is not relevant since it is inconsistent with the basics of the eligibility inquiry under section §101.
The declarant is also attempting to characterize the existing computer/network technology, which the claimed—and the disclosed—system is relying on, as an alleged technological improvement. The declarant asserts (emphasis added),
“[t]he technical significance of this architecture is further supported by the quantitative and qualitative feedback modules, which together implement a specific transformer-based scoring and explanation architecture. Regarding the AI Bot's previous heuristic (i.e neural network) [training], the specification states that the AI Bot module may use a large open-source transformer-based model to generate a score based on the question, solution input by the student, the reference answer, and the maximum score, and that the AI Bot module is trained using sets of questions, answers, and corresponding grades. Specification, paragraph 0098. The qualitative feedback module generates a description explaining to the student what would have been needed to obtain a higher score or the maximum score. This specific multi-input transformer-based simultaneous generation of categorically distinct quantitative and qualitative feedback outputs is a concrete technical implementation that differs from the human process of grading not merely in speed or scale, but in architectural kind . . . in the obviousness analysis under 35 U.S.C. § 103, the Examiner found no prior art teaching or suggesting the predictive knowledge state architecture now expressly recited in the body of amended claim 7. This is consistent with my opinion that this technical feature represents an unconventional and non-routine computer system capability that did not exist in prior art educational software systems, and that the amended claim recites a specific and technically bounded computer-implemented architecture that constitutes significantly more than any alleged abstract idea”
However, as already demonstrated above, it is already part of the existing computer/network technology to utilize one or more existing models—such as, transformer-based Large Langue Models, to analyze collected data and generate pertinent results. So far, the declarant fails to demonstrate any claimed/disclosed technological feature—or any combination of features—that is considered to be an advance over the existing computer/network technology. Instead, while simply designating various labels—such as, the so-called “specific transformer-based scoring and explanation architecture”, the declarant is attempting to portray an alleged technological improvement while repeatedly mistaking the abstract idea for a technological feature. For instance, except for the usual labeling of the features of the existing technology, the declarant fails to point out a disclosed/claimed advanced technological feature (if any) that may render the claimed/disclosed system beyond the existing computer/network technology. Instead, the declarant is emphasizing the abstract idea, which the existing technology is utilized to facilitate. For instance, the declarant is emphasizing that “the AI Bot module may use a large open-source transformer-based model to generate a score based on the question, solution input by the student, the reference answer, and the maximum score” (emphasis added). In this regard, it has already been demonstrated above that the use of an AI chatbot, which implements a transformer-based Large Language Model to analyze data, is already part of the existing computer/network technology. In fact, one of the existing models that the specification admits—such as GPT-4 (see [0118])—already implements a transformer-based Large Language Model. Accordingly, the only features left regarding the claimed/disclosed system are the pieces of information (e.g., the score, the solution, the reference answer and the maximum score, etc.), which Applicant’s claimed/disclosed system is manipulating. However, neither the pieces of information that the claimed/disclosed system is analyzing, nor the pieces of information that the claimed/disclosed is generating, signify a technological improvement regardless of whether they are considered individually or in any ordered combination with any of the claimed/disclosed features. In particular, each piece of information is merely an abstract idea regardless of whether it is considered individually or in any ordered combination.
Of course, the same is true regarding the training process that the declarant is asserting, “the AI Bot module is trained using sets of questions, answers, and corresponding grades” (emphasis added). For instance, except for simply listing the pieces of information used to train the model, the declarant fails to demonstrate whether the claimed/disclosed system is implementing any technologically advanced training scheme. Note that, depending on the particular field, it is already part of the existing computer/network technology to train an AI model using one or more desired training datasets. For instance, an AI model intended for the medical field may be trained using one or more datasets related to symptoms and/or treatments of various diseases, etc. Similarly, an AI model intended for the educational field may be trained using one or more datasets related to students and/or course materials, etc. However, such use of the desired dataset to train an AI model for a particular objective and/or field does not necessarily signify a technological improvement. This is because the above is merely signifying the different piece of information (the different datasets) used to train the model. Of course, the same is true regarding the piece of information that the model is providing as an output. For instance, the results (the pieces of information) that the claimed/disclosed system is providing—such as, the alleged “description explaining to the student what would have been needed to obtain a higher score or the maximum score”, the alleged “categorically distinct quantitative and qualitative feedback”, etc., do not signify any technological improvement whether such information is considered individually or in any ordered combination with any of the claimed/disclosed features. This is again because such information is merely an abstract idea. In contrast, a technological improvement (if any) is shown based on a technologically advanced structural and/or functional feature—such as, an advanced training scheme for training the AI model (e.g., see Ex Parte Desjardins., Appeal No. 2024-000567).
Accordingly, the declarant’s repetitive attempt to substantiate an alleged technological improvement, while simply emphasizing the information that the claimed/disclosed system is analyzing and/or generating, is once again not persuasive. In particular, except for the various labels that the declarant is designating, including the so-called “concrete technical implementation”, and/or the alleged “unconventional and non-routine computer system capability”, etc., the declarant fails to demonstrate whether the claimed/disclosed system is providing any technological improvement over the existing computer/network technology. Consequently, the declarant’s assertions are not persuasive.
The declarant also appears to be attempting to replace the patent-eligibility analysis (section §101) with an obviousness analysis of the prior art (section §103). However, quite contrary to the declarant’s theory above, neither the novelty analysis (section §102) nor the obviousness analysis (section §103) influences the eligibility analysis (section §101). Also see MPEP 2106.06(I) cited above. In particular, the claimed/disclosed system may overcome the prior art (nonobvious over the prior art) merely due to the new abstract idea it is reciting, but not necessarily due to an advanced technological feature (if any) it is implementing. This is again because the claimed (and originally disclosed) system is utilizing the existing computer/network technology—merely as a tool—to facilitate the new abstract idea. In contrast, while mistaking the new abstract idea for an alleged technological feature, the declarant is making various conclusory assertions; such as, the alleged “technical feature [that] represents an unconventional and non-routine computer system capability that did not exist in prior art educational software systems” and/or the alleged “specific and technically bounded computer-implemented architecture that constitutes significantly more than any alleged abstract idea”, etc. Thus, none of the declarant’s conclusory assertions is persuasive. Of course, it is also immaterial whether the declarant is asserting that none of the current claims is obvious over the prior art. This is again because such non-obviousness (if any) is only due to the new abstract idea that the current claims are reciting, but not due to any of the technological features that the claimed/disclosed system is implementing.
(e) Regarding claim 14, the declarant has attempted to demonstrate an alleged technological improvement despite the exiting technology that the claimed/disclosed system is implementing. The declarant asserts (emphasis added),
“[t]he technological improvement implemented by the system of claim 14 is a specific two-tier natural language processing and query routing architecture that did not exist in prior art question-answering systems. The architecture operates as a unified pipeline in which the provenance indication is an integral output of the routing decision itself, produced automatically by the system as a direct consequence of which tier of the architecture processed the query. In the first tier, a paraphrase detector module implementing a trained neural search engine determines whether a student’s input question meets a learned similarity threshold against a curated dataset of AI-generated questions that have been human-reviewed and approved or rewritten, and if the threshold is met, routes the query to the response generation module which returns the corresponding human-approved answer with provenance marking confirming human verification. In the second tier, if the similarity threshold is not met, a distinct fallback mechanism routes the query to the AI Bot module, which formulates a response directly from course material using a large language model grounded in excerpts of the coursebook, with the advantage that it grounds the model and reduces hallucinations, and the system automatically marks that response as AI-generated and not human-approved. Specification, paragraph 0118. The combined architecture—threshold-based paraphrase detection against a human-reviewed dataset, a separate fallback mechanism for queries not meeting the threshold, and automatic provenance marking of responses according to their routing pathway—constitutes a specific and integrated technical improvement in the functioning of AI-driven question answering systems that did not exist in prior art educational software systems and that is categorically incapable of being performed manually”
However, quite similar to the observations made so far, here also the declarant is attempting to paint an alleged technological improvement while simply designating various labels to the features of the exiting technology. As an initial matter, it is worth noting that the declarant’s assertion relates to claim 17, as opposed to claim 14, since claim 14 does not even contemplate whether the AI Bot is going to formulate an answer based on the course material when the similarity “is not equal to or above the predetermined amount of similarity” (emphasis added). Consequently, the declarant’s assertion directed to claim 14 is not even consistent with the scope of claim 14.
Nevertheless, despite asserting that the existing computer/network technology is lacking the so-called “specific two-tier natural language processing and query routing architecture”, the declarant still fails to establish a specific technological feature (if any)—or a combination of technological features (if any)—that the existing computer/network technology is assumed to be lacking. Instead, while attempting to mask the lack of technological improvement using prolix and grandiloquent assertions, the declarant is simply referring to the basic functions of existing AI models in order to portray the alleged technological improvement. For instance, existing AI-based chatbots normally interpret a user’s input (e.g., one or more words/phrases) in order to recognize the user’s request; and this process may involve comparing the user’s input to one or more already approved/known data elements in its database, including determining their degree of similarity (e.g., using cosine similarity), etc. In fact, existing digital assistants (or chatbots) typically display to the user—after analyzing the user’s original input—one or more paraphrased versions of the user’s input; and further request the user to confirm the one that matches his/her original request (e.g., prompting the user to select at least one of the paraphrased versions that match the user’s request). Of course, once receiving the user’s selection or confirmation, the model provides the most relevant answer to the user. Therefore, implementing such existing functions to accomplish a desired objective in a particular field—the education field—does not necessarily signify a technological improvement.
In addition, as already discussed above, the developers (e.g., OpenAI, etc.) of existing AI models already provided various techniques for reducing the possibility of hallucinations in AI models. In particular, the developers provided various options that allow existing AI models to be customized according to the specific needs of a service provider (e.g., a call center, an educational institution, etc.). For instance, one of such existing options is of course RAG (i.e., Retrieval-Augmented Generation), which was formally introduced in 2020; and this option essentially forces the AI model, when formulating an answer, to ground its answers in human-approved source documents. Accordingly, when the AI model is used in a call center, it uses approved documents (e.g., manuals, policies, etc.) of that call center to formulate a proper response to a question that a customer is providing. Of course, besides such options, the model also routes a question to a human agent when it is uncertain about it. Thus, when such model is used in an educational institution, the model uses approved documents (e.g., course materials, instructor notes, etc.) to formulate a proper answer to a student’s question, etc.
Accordingly, simply incorporating one or more of such existing functions into the claimed/disclosed system does not constitute a technological improvement. In particular, except for simply assigning various labels to the existing functions (e.g., the so-called “a unified pipeline”, “a paraphrase detector module”, “a trained neural search engine”, “distinct fallback mechanism”, etc.), and/or simply categorizing the existing functions in to two or more categories (e.g., the alleged “first tier” and “second tier”), neither the declarant’s theories nor the section(s) cited from the specification shows any technological feature—or any combination of features—that provides a technological improvement. Consequently, none of the declarant’s conclusory assertions, including the alleged “specific two-tier natural language processing and query routing architecture”, and/or the alleged “specific and integrated technical improvement”, is persuasive.
The declarant has also attempted to refer to the specification in further attempt to portray an alleged technological improvement. The declarant asserts (emphasis added),
“[t]he specification describes the paraphrase detector module as implementing a neural search engine trained by calibrating a sensitivity threshold on a dataset of real student questions which are manually labelled or tagged as to whether they are similar to each other or not. Specification, paragraph 0115. This is a specifically trained machine learning component operating within a technically defined query routing architecture. The training of the neural search engine on manually labelled real student question data to calibrate a sensitivity threshold, the automated routing of queries between a human-curated knowledge base and an LLM-based fall back architecture, and the automated provenance marking of responses as human-approved or AI-generated are specific and non-routine technical implementations that constitute improvements in the functioning of AI-driven question answering systems. They are not mental processes. No human could simultaneously and in real time assess the semantic similarity of an incoming question against an entire database of artificial questions, apply a calibrated similarity threshold to route the query to the appropriate response source, retrieve the most similar approved answer, and mark the response with appropriate provenance information, all within the constraints of a live interactive computer system”
However, once again quite similar to the declarant’s patterns noted above, here also the declarant is attempting to paint the features of existing computer/network technology as an alleged technological improvement. For instance, the function that the declarant is describing regarding the so-called “paraphrase detector module”, which is supposedly trained using manually labelled or tagged documents to calibrate the alleged “sensitivity threshold”, etc., is essentially referring to one of the existing machine training techniques—namely, a supervised or a semi-supervised machine learning. In particular, regarding existing supervised or semi-supervised neural network, the model trains on input data paired with correct/human labeled data; and further, the model adjusts—by comparing its outputs to the true/labeled data—the weights of one or more of its parameters in order to reduce error.
Thus, except for relying on the usual prolix and grandiloquent assertions to mask the use of such existing features, the declarant fails to establish whether the claimed/disclosed system is implementing any new/advanced technological feature. Consequently, here also the declarant’s conclusory assertion regarding the alleged “improvements in the functioning of AI-driven question answering systems” is not persuasive.
The declarant also continues to misapply the eligibility inquiry that must be applied to determine an abstract idea—such as, a mental process. In particular, the declarant is repeatedly relying on the computer elements (the so-called “paraphrase detection module”, “neural search engine”, “trained machine learning component”, and/or “live interactive computer system”, etc.) in an attempt to dismiss the Office’s findings regarding a mental process. However, it is once again worth noting that none of the computer elements is considered when evaluating whether the claim is reciting an abstract idea. Moreover, when evaluating a mental process, the eligibility inquiry under Prong One does not require one to evaluate whether a human’s mind can operate/function with the same speed/efficiency of a computer system. Instead, the test requires one to identify only the limitations that can be performed in the human mind (and/or using a pen and paper), while excluding the limitations that are beyond a mental process—namely, the computer elements. Consequently, it is immaterial whether the declarant concludes that “No human could simultaneously and in real time assess the semantic similarity of an incoming question against an entire database of artificial questions . . . all within the constraints of a live interactive computer system” (emphasis added).
Moreover, exiting AI-based chatbots not only formulate, using documents stored in databases, a pertinent response to a user (e.g., a summary regarding a particular topic, etc.), but also generate a direct response(s) to the user from the documents (e.g., providing an excerpt from a document, including annotating the source, etc.). Accordingly, utilizing such existing technology to achieve a desired objective—such as, utilizing an AI-based Bot to provide a student with (i) an answer that the AI Bot drafted based on course materials stored in a database(s), and/or (ii) an answer that a teacher already provided, including annotating the answer based on the source, etc., does not demonstrate a technological improvement.
Note also that the declarant’s conclusions (see, the 33rd to 38th paragraphs of the declaration) do not provide any new rationale, much less a new/advanced technological feature (if any), that signifies a technological improvement. Instead, the declarant appears to simply recap the same assertions addressed above—such as, the alleged “multi-source real-time data integration architecture of claim 1”, the alleged “predictive knowledge state modeling and adaptive problem selection architecture of claim 7”, the alleged “tiered provenance-marked natural language query routing architecture of claim 14”, which are assumed to be “technical capabilities that are categorically incapable of being performed mentally or manually” and/or “specific and concrete improvements over conventional educational software systems”.
However, the Office’s analysis presented above already demonstrates the reasons why none of the assertions above is persuasive. Of course, the same is true regarding the declarant’s repetitive assumptions regarding the alleged “additional elements that unquestionably amount to significantly more than any alleged judicial exception”, and/or the alleged “specific, technically bounded, computer-implemented architectures that did not exist in prior art systems and that are incapable of being replicated through any human cognitive or manual workflow”, etc. This is again because the Office’s discussion presented above already demonstrates the reasons why none of the declarant’s subjective assumptions above signifies any technological feature—or any combination of features—that provides a technological improvement. Consequently, the declarant’s conclusory assertions are not persuasive.
The declarant has also attempted to summarize the dependent claims in further attempt to substantiate the alleged technological improvement. However, except for mischaracterizing the existing technology as systems that “presented information uniformly without adaptation and did not perform real-time multi-source data integration to generate personalized adaptive dialogue, did not maintain continuously updated probabilistic performance predictions across an entire problem bank to drive adaptive problem selection, and did not implement a tiered natural language query routing architecture combining a trained paraphrase detection module, a human-curated knowledge base, and a provenance-marked LLM-based fallback response pipeline”, the declarant fails to demonstrate whether any of the dependent claims is implementing a technological feature—or a combinations of features—that is an advance over the existing computer/network technology. Thus, the declarant’s prolix and grandiloquent assertions above, which do not rely on factual findings, once again fail to even challenge—much less negate—the Office’s ineligibility findings.
Not surprisingly, the declarant has also attempted to once again improperly rely on the computer elements to challenge the Office’s finding regarding a mental process (e.g., see the 37th paragraph). However, as repeatedly pointed out above, the eligibility test for a mental process (under Prong One) does not require one to consider the computer elements, much less compare the human’s mind with the speed or functions of a computer system. Consequently, the declarant’s repetitive arguments are once again not persuasive.
Note also that the last paragraph (the 39th paragraph) is yet another repetitive argument directed to the independent claims. In particular, the declarant appears to be attempting to summarize the alleged technological improvement while once again relying merely on prolix and grandiloquent assertions (emphasis added),
“the combination of the particular computer-implemented elements in each independent claim creates a technological solution to the technological problem of how to deliver personalized, adaptive, computer-based education at scale with sufficient technical sophistication to adapt to individual student attributes, predict and respond to student knowledge states, and provide quality-assured responses to student questions, all without requiring the continuous presence of a human instructor. By this novel and inventive combination of computer processes, including multi-source real-time data integration for adaptive dialogue generation, predictive knowledge state modeling integrated with adaptive problem selection, and tiered provenance-marked natural language query routing, the claimed invention improves upon prior art educational software systems that lacked the technical architecture necessary to achieve these capabilities, to produce technological solutions that were not previously available in either manual or computer-implemented educational systems. For the reasons discussed herein, it is my opinion that the claimed invention is directed to a specific technological solution to a concrete technological problem and is therefore directed to patent-eligible subject matter”
However, the Office’s discussion presented above already provides sufficient analysis regarding why none of the independent claims is implementing any technological improvement. In fact, the Office’s discussion encompasses the original disclosure as a whole. In particular, given the lack of technological improvement regarding the disclosed system as a whole, none of the current claims implements a technological improvement regardless of whether one is considering the independent claims or the dependent claims. Thus, due to the lack of any advanced technological feature per the claimed/disclosed system, the declarant appears to be attempting to substantiate the alleged technological improvement while relying on various tech jargon terms—such as, “sufficient technical sophistication”, “novel and inventive combination of computer processes”, “specific technological solution to a concrete technological problem”, etc. However, none of such assertions necessarily signifies any advanced technological feature(s); and therefore, the declarant’s arguments are one again not persuasive.
In view of the foregoing, when all of the evidence is considered, the totality of the rebuttal evidence of eligibility fails to outweigh the evidence of ineligibility.
► Applicant’s arguments directed to section §101 have been fully considered (the arguments filed on 08/04/2026). However, the arguments are not persuasive at least for the following reasons:
1. Prong One of Step 2A
Firstly, Applicant’s preliminary statement regarding the subject matter eligibility declaration (hereinafter SMED) is already addressed above. In particular, the Office has already fully considered and addressed each issue raised in the SMED. Thus, Applicant is advised to review the analysis presented above.
Regarding Prong One of Step 2A, Applicant has attempted to summarize the court’s decision regarding McRO. Applicant further concludes that “[t]he Office Action characterizes claim 1 as directed to managing personal behavior or
mental processes. Applicant respectfully submits that this characterization is incorrect” (emphasis added).
However, Applicant’s conclusory assertion above is erroneous since Applicant fails to point out the rationale (if any) to challenge the Office’s findings presented under Prong One of Step 2A. For instance, Applicant fails to demonstrate whether the specific limitations, which the Office identified under Prong One of Step 2A, are beyond an abstract idea.
Applicant has also quoted a section from the SMED in an attempt to support the conclusory assertion. However, the Office’s analysis presented above (see the analysis presented under the heading “Response to the Declaration under CFR 1.132”) already demonstrates the reasons why the SMED fails to negate the Office’s ineligibility findings. Thus, Applicant is expected to review the analysis above.
In addition, Applicant is misapplying McRO in an attempt to challenge the Office’s finding under Prong One. It is worth noting that McRO is considered to be patent-eligible due to the technological improvement it is implementing (e.g., an improvement in computer animation). However, a technological improvement is an issue to be evaluated under Prong Two of Step 2A, but not under Prong One. This is because the eligibility inquiry under Prong One requires one to identify merely the limitations that recite the judicial exception (e.g., an abstract idea). Accordingly, regardless of whether a given claim is implementing a technological improvement or not, this has nothing to do with the inquiry under Prong One. In contrast, Applicant is improperly relying on McRO in an attempt to challenge the Office’s finding under Prong One. For instance, Applicant is asserts (emphasis added),
“Similar to the claim element at issue in McRO Claim 1 of the present application is bounded by specific inputs and specific outputs. The specific inputs to the AI Bot module are first data from the course database, second data from the question database, third data from the personalization database comprising personal attributes of the, and the student's prior answer in the ongoing dialogue. The specific outputs are dynamically generated adaptive instructional content tailored to the individual student based on the integrated processing of all of those specific inputs simultaneously. Like the claim element at issue in McRO, the specificity of these inputs and outputs limits the scope of claim 1 to a specific and bounded technical process rather than an abstract idea achievable by any means”
However, besides misapplying McRO to challenge the Office’s findings under Prong One, Applicant also appears to be drafting a new theory in an attempt to correlate McRO with current claim 1. This is because, unlike Applicant’s new theory, the eligibility of McRO has nothing to do with the alleged “specific inputs and specific outputs”, which supposedly bound the system. Instead, the eligibility of McRO was due to the technological improvement resulted from the implementation of “particular rules to set morph weights and transitions through phonemes) to solve the problem of producing accurate and realistic lip synchronization and facial expressions in animated characters”, see MPEP 2106.05(a)(II) (emphasis added). This is indeed a technological improvement—namely, an improvement in computer animation. Thus, it is evident that the above has nothing to do with Applicant’s new theory, the alleged “specific inputs and specific outputs” that purportedly bounds the system. In fact, even common sense dictates that a system, which is assumed to take specific inputs and generates specific outputs, does not necessarily signify a technological improvement. For instance, a basic non-scientific calculator can be characterized as a system or an application “bounded by specific inputs and specific outputs” since the inputs and the outputs of the calculator are limited merely to numerical data. Nevertheless, Applicant still fails to show whether any of the current claims is not reciting an abstract idea. Instead, Applicant appears to simply summarize the inputs, which the claimed system is assumed to be receiving, and the outputs that the system is assumed to be providing. Thus, Applicant’s assertion fails to even challenge—much less negate—the Office’s findings under Prong One.
Similarly, regarding claim 7, Applicant is once again quoting a section from the SMED in an attempt to challenge the analysis presented under Prong One. However, here also the Office directs Applicant’s attention to the Office’s response presented above regarding the SMED. In particular, the analysis presented above already details the reasons why the SMED fails to negate the Office’s ineligibility findings.
In addition, quite similar to the pattern noted with respect to claim 1, here also, regarding claim 7, Applicant is attempting to challenge the findings under Prong One while misapplying McRO. Applicant asserts (emphasis added),
“Similar to the claim element at issue in McRO, claim 7 of the present application is bounded by specific inputs and specific outputs. The specific inputs to the knowledge state module are the history of solutions input from the student for each problem of the plurality of problems to be selected. The specific outputs are a predicted expected score of the input solution at the present date for each problem in the problem bank with those predicted scores then consumed by the problem selection module as the direct basis for selecting the next appropriate problem. Like the claim element at issue in McRO, the specificity of these inputs and outputs limits the scope of claim 7 to a specific and bounded technical process — a layered predictive machine learning architecture operating on a defined input to produce a defined output that directly drives system behavior — a rather than an abstract idea achievable by any means”
However, once again Applicant’s new theory directed to the eligibility of McRO—namely, the alleged “specific inputs and specific outputs” that supposedly bound the system, is irrelevant to challenging the Office’s finding under Prong One. This is because Applicant not only fails to properly construe the court’s eligibility analysis directed to McRO, but also fails to properly construe the eligibility inquiry set forth under Prong One of Step 2A. Nevertheless, even assuming arguendo that the alleged “specific inputs and specific outputs” are assumed to be limiting claim 7 to the so-called “specific and bounded technical process”, this does not necessarily mean that the claim is no longer reciting an abstract idea. This is because neither the alleged “specific inputs and specific outputs” nor the so-called “specific and bounded technical process” precludes the claim from implementing limitations that recite an abstract idea. This is further evident given the findings under Prong One, which identifies the limitations that recite the abstract idea. So far, except for describing the types of inputs that the claimed system is assumed to be gathering, and/or the type of outputs that the system is assumed to be generating, Applicant still fails to challenge the Office’s findings presented under Prong One. Note also that the discussion presented above regarding claim 1 also applies to claim 7 since Applicant is relying on the same theory.
Consequently, at least for the reasons above, Applicant’s arguments are not persuasive.
Regarding claim 14, Applicant is again quoting a section from the SMED in an attempt to challenge the Office’s findings under Prong One. However, Applicant is once again advised to review the Office’s analysis regarding the SMED. In particular, the Office has already provided a comprehensive analysis, which details the reasons why the SMED fails to challenge—much less negate—the Office’s ineligibility findings.
Regarding claim 14, Applicant is also once again relying on McRO in an attempt to challenge the findings under Prong One. For instance, Applicant asserts (emphasis added),
“Similar to the claim element at issue in McRO, claim 14 of the present application is bounded by specific inputs and specific outputs. The specific inputs to the paraphrase detector module are a question input from the student and the curated dataset of artificial questions that have been human-reviewed and approved or rewritten, against which the student's input question is analyzed for similarity equal to or above a predetermined threshold. The specific outputs are determined by the routing decision itself: where the similarity threshold is met, the response generation module returns the corresponding human-approved answer with provenance marking confirming human verification; where the similarity threshold is not met, a distinct fallback mechanism routes the query to the AI Bot module, which formulates a response grounded directly in course material, and the system automatically marks that response as AI-generated and not human-approved. Like the claim element at issue in McRO, the specificity of these inputs and outputs limits the scope of claim 14 to a specific and bounded technical process – a two-tier natural language processing and query routing pipeline operating on defined inputs to produce defined outputs with integrated provenance marking as a direct consequence of the routing decision – rather than an abstract idea achievable by any means”
Accordingly, besides repeatedly misapplying McRO to address issues related to Prong One, Applicant also continues to rely on the new subjective theory regarding McRO while disregarding the court’s rationale. However, similar to the points made above, none of the current claims—including claim 14—is even remotely analogous to McRO. In particular, McRO corresponds to a technological improvement—namely, an improvement in computer animation (see discussion above). In contrast, each the current claims (including the original disclosure as a whole) is relying on the existing computer/network technology—merely as a tool—to facilitate the claimed/disclosed an abstract idea; such as, facilitating the presentation of educational materials to a student based on the analysis of data related to the student and/or course materials, etc. (see claim 14, or any of the current claims). Accordingly, Applicant’s new theory regarding the alleged “specific inputs and specific outputs”, which supposedly make the claim “a specific and bounded technical process”, does not change the fact above.
Moreover, except for the attempt made to describe the features that supposedly represent the alleged “specific inputs”, including the process that the claimed system is assumed to be performing to generate the alleged “specific outputs”, Applicant still fails to address the crux of the eligibility inquiry under Prong One. For instance, no specific rationale is presented to demonstrate why Applicant assumes that limitations that the Office identified under Prong One are not abstract idea. Thus, Applicant’s assertions are not persuasive.
2. Prong Two of Step 2A.
Secondly, regarding Prong Two, Applicant is once again referring to part of the SMED in an attempt to justify the alleged integration of the abstract idea into a patent-eligible practical application.
However, Applicant is once again remined to review the Office’s response presented above, which provides detail analysis regarding why the declarant’s arguments fail to demonstrate a technological improvement. In particular, besides mischaracterizing the existing technology as a nonadaptive system, the declarant is attempting to justify the alleged technological improvement while repeatedly mistaking the abstract idea for a technological feature. Of course, the analysis also demonstrates the declarant’s repetitive attempts to portray the features of the existing computer technology as advanced technological features while relying on prolix and grandiloquent assertions. Accordingly, simply quoting one or more sections from the SMED does not necessarily show whether the claimed (and/or the disclosed) system is implementing a technological improvement.
Applicant has also attempted to rely on sections from the specification in further attempt to substantiate the alleged technological improvement. Applicant asserts (emphasis added),
“. . . the specification, which states that the AI Bot module is configured to adapt the complexity and content of the dialogue based on one or more or all of those attributes, and that one or more of those attributes are used as part of prompts to a machine learning model of the AI Bot module along with an instruction to take the attributes into account. Specification, [0079]. The specification further confirms this improvement in paragraph [0073], where it states that the learning effect of the student is improved by regarding personal attributes in the dialogue and that the learning effect is superior when personal attributes and experiences are taken into account . . . the specific combination of elements recited in claim 1 - the course database, question database, personalization database, dialogue module, and AI Bot module operating together to simultaneously integrate three distinct structured data sources in real time to dynamically generate adaptive instructional content — integrates any alleged abstract idea into a specific and concrete improvement in the technical field of computer-based educational systems that was not achievable by prior art systems. Claim 1 therefore recites a practical application of any alleged abstract idea and is patent eligible”
However, except for the attempt made to summarize the tasks that the disclosed (or the claimed) system is assumed to be performing, the declarant fails to show a technological feature (if any)—or a combination of features (if any)—that provides a technological improvement over the existing computer/network technology. Instead, quite similar to the patterns noted with respect to the declarant, Applicant is also attempting to portray the features of the existing computer/network technology as the alleged technological improvement. For instance, during interaction with a human subject, existing AI-based chatbots normally adapt the complexity and the content of the dialogue based on various factors (the issue being discussed, the responses that the user is providing, etc.); and further, one or more of the factors are used as feedback to update/train the machine-learning model of the chatbot. Accordingly, implementing such features of the existing computer/network technology to achieve a desired objective (e.g., educating a student) does not signify a technological improvement.
Similarly, the alleged benefits that the claimed/disclosed system is assumed to be providing to the user also do not signify a technological improvement. For instance, it is immaterial whether one assumes that “the learning effect of the student is improved by regarding personal attributes in the dialogue” and/or “the learning effect is superior when personal attributes and experiences are taken into account” (emphasis added). This is because the above is merely signifying a teaching plan or strategy—i.e., an abstract idea—but not necessarily a technological feature. Of course, the teaching plan above may—or may not—be effective since such teaching plan does not always have the same effect on all students. Nevertheless, regardless of the effectiveness of the plan, the claimed/disclosed system is merely performing existing computer functions—such as, collecting student and/or course data, and subsequently generating—based on the analysis of the data—pertinent information to the student. Of course, it is also part of existing AI-based chatbots to access multiple data sources simultaneously (e.g., local and/or remote databases over the Internet) in in order to dynamically generate—in real-time—the most relevant response to the user regarding the issue being discussed.
Accordingly, the claimed/disclosed system, which relies merely on the features of the exiting computer/network technology to facilitate dialogues between an AI Bot and a student, does not constitute a technological improvement. Thus, Applicant’s conclusory assertion—namely, the alleged integration of “any” abstract idea into the so-called “specific and concrete improvement in the technical field of computer-based educational systems that was not achievable by prior art systems”, is not persuasive.
Regarding claim 7, Applicant is as usual quoting sections from the SMED in an attempt to justify the alleged technological improvement. However, here also the Office directs Applicant to review the analysis presented above (under the heading, “Response to the Declaration under CFR 1.132”), which already details the reasons why the declarant’s SMED fails to demonstrate a technological improvement. It appears that Applicant is repeatedly resorting to the declarant’s remarks due to the lack of technological improvement in the original disclosure.
Applicant has also cited some sections from the specification in an attempt to support the alleged technological improvement. Applicant asserts (emphasis added),
“. . . the specification, which states that the knowledge state module may be adapted to predict based on the history of the solutions input from the student for each problem out of the problems to be selected an expected score of the input solution at a preset date, and may also predict a score in the future by taking the history of the predicted expected score into account, with the predicted score at the present date or the predicted score at the future date used by the problem selection module to select an appropriate problem. Specification, [0103]. The specification further confirms this improvement at paragraph [0103], where it states that better training goals and faster fulfillment of the requirements to pass an exam are obtained through this predictive knowledge state architecture . . . the specific combination of elements recited in claim 7 — the knowledge state module generating present-date and future-date expected score predictions across the entire available problem set, the problem selection module consuming those predictions as the direct basis for selecting appropriate problems, and the transformer-based quantitative and qualitative feedback modules operating together as an integrated system — integrates any alleged abstract idea into a specific and concrete improvement in the technical field of computer-based educational systems that was not achievable by prior art systems. Claim 7 therefore recites a practical application of any alleged abstract idea”
However, here also except for summarizing the data that the disclosed system is analyzing, and/or the results that the system is generating, Applicant still fails to establish a technological feature (if any)—or a combination of features (if any)—that is assumed to be providing a technological improvement. For instance, the process of generating—based on the analysis of stored data or history—one or more predicted scores, including selecting pertinent problem to the student based on the predicted score(s), has nothing to do with technology, much less a technological improvement. This is because the above is merely signifying one’s strategy or approach for selecting a relevant problem to the student. Of course, it is questionable how/why selecting a given problem (the so-called “appropriate problem”) based on a predicted score of an input solution is going to be useful in enhancing the student’s skills (since a problem that the student already mastered may be selected if the predicted score of the input solution favors it). Nevertheless, such process of selecting an “appropriate problem” based on a predicted score of an input solution has nothing to do with any technology, regardless of whether one is considering the expected score of the input solution at (a) the present date, (b) a future date, or (c) both the present date and the future date. This is because the above is merely signifying an abstract idea, but not a technological feature, much less an advanced technological feature. Consequently, Applicant’s assertions are not persuasive. In particular, quite similar to the patterns noted with respect to the declarant, Applicant is also mistaking the abstract idea for an alleged technological improvement.
Note that the same is true regarding the alleged benefits that the claimed system is assumed to be providing to the student. For instance, it is immaterial whether one assumes that the claimed (or the disclosed) predictive scheme helps the student to achieve “better training goals and faster fulfillment of the requirements to pass an exam”. This is because the above is merely signifying a subjective assumption about a potential benefit, but not a technology, much less a technological improvement. Although Applicant has also attempted to summarize the alleged functions of the claimed/disclosed “knowledge state module” and “problem selection module”, once again Applicant fails to demonstrate whether any of the features above, individually or in any ordered combination with any of the claimed/disclosed features, is providing a technological improvement. Thus, Applicant’s conclusory assertion regarding the alleged integration of “any” abstract idea into the so-called “specific and concrete improvement in the technical field of computer-based educational systems” is not persuasive.
Regarding claim 14, Applicant is similarly quoting sections from the SMED in an attempt to justify the alleged technological improvement. However, as repeatedly pointed out above, the Office has already addressed the declarant’s SMED separately under the heading “Response to the Declaration under CFR 1.132”; and therefore, Applicant is once again advised to review the Office’s response above.
Applicant has also attempted to discuss some sections from the specification in further attempt to support the alleged technological improvement. Applicant asserts (emphasis added),
“the specification, which describes the paraphrase detector module as implementing a neural search engine trained by calibrating a sensitivity threshold on a dataset of real student questions which are manually labelled or tagged as to whether they are similar to each other or not. Specification, [0115] . . . confirms this improvement at paragraph [0118], where it states that the AI Bot module formulates an answer based on the course material without using the pairs of artificial questions and related approved or rewritten answers where the similarity threshold is not met, and that the system marks the provided answer as not approved from the human person having knowledge in the field of the specific course . . . with the advantage that grounding the model in course material reduces hallucinations . . . the specific combination of elements recited in claim 14 – the AI Bot module generating a human-reviewed artificial question-answer dataset, the paraphrase detector module performing threshold-based neural similarity analysis calibrated on manually labelled real student question data, the response generation module returning human-approved answers with provenance marking, and the distinct LLM-based fallback mechanism with automatic provenance marking, all operating together as a unified tiered pipeline — integrates any alleged abstract idea into a specific and concrete improvement in the technical field of AI-driven question answering systems that was not achievable by prior art systems”
However, neither the cited sections nor Applicant’s assertions demonstrates any technological improvement. Instead, quite similar to the errors noted in the declarant’s SMED, Applicant is also attempting to portray the features of the existing computer technology as the alleged technological improvement. For instance, the process of training a machine-learning model using human labeled/tagged dataset is one of the features of the existing computer/network technology—namely, a supervised or a semi-supervised machine learning. For instance, when considering existing supervised or semi-supervised neural network, the model trains on input data that is paired with correct/human labeled data; and accordingly, while comparing its outputs to the true/labeled data, the model adjusts the weights of one or more of its parameters in order to reduce error. Accordingly, except for designating various labels and/or expressions; such as the so-called “paraphrase detector module”, “calibrating a sensitivity threshold”, etc., Applicant fails demonstrate whether the claimed/disclosed system is implementing any technological feature, or a combination of technological features, that is assumed to be beyond the features of the existing computer/network technology. Note that the dataset itself—such as, the “real student questions”, which are manually labeled as being “similar to each other or not”, also does not signify any technological improvement.
Of course, the same is true regarding the annotation(s), and/or description of a source, that an AI-based chatbot provides to a user during interaction. For instance, besides formulating—using documents stored in databases—a pertinent response to a user (e.g., a summary regarding a particular topic, etc.), existing AI-based chatbots also generate direct responses from the documents (e.g., providing an excerpt from a document, including annotating the source, etc.). Thus, utilizing such features of the existing technology to achieve a desired objective—such as, utilizing an AI-based Bot to provide a student with (i) an answer that the AI Bot drafted based on course materials stored in a database(s), and/or (ii) an answer that a teacher already provided, including annotating the answer, etc., does not signify a technological improvement. In particular, neither the particular field/environment where the existing technology is being used, nor the particular objective being achieved using the technology, signifies any technological improvement. Therefore, Applicant’s assertions directed to the alleged technological improvement are still not persuasive.
In addition, except for the generic assertion regarding the use of course materials to reduce hallucination of the AI Bot, Applicant still fails to demonstrate whether the claimed/disclosed feature is implementing any technologically advanced feature. Nevertheless, existing AI-based chatbots already have various options for reducing hallucinations. Accordingly, some of these options include: (a) using Retrieval-Augmented Generation (RAG)—such as, a pipeline that essentially forces the AI model—when formulating an answer—to ground its answers in approved source documents (i.e. requiring the model to use only human provided/approved documents to formulate a relevant answer), wherein the model is also instructed to say “I don’t know” if it is uncertain, so that it is refrained from guessing, (b) applying a Chain-of-Thought (CoT), wherein the model is required to outline its reasoning in a step-by-step fashion before providing a response, (c) lowering its temperature, so that the model chooses the most statistically likely word (via greedy decoding); thereby discouraging the model from using broader sampling and creative combinations, etc.
It is further worth noting that additional options are also provided for customizing such models according to the objectives and/or policies of businesses or service providers (e.g., call centers, educational institutions, etc.). For instance, when considering a call center that implements an AI agent/bot, the call center typically customizes the AI agent/bot to route—to a human agent—a question (or an issue) that the bot cannot resolve. Accordingly, simply implementing such exiting feature in a particular field—e.g., the education field/environment, does not necessarily signify a technological improvement. Instead, the above is demonstrating the use of the exiting technology—merely as a tool—to facilitate similar objective in a different field.
So far, except for Applicant’s prolix and grandiloquent assertions (similar to the observation made with respect to the declarant), Applicant fails to show whether the claimed/disclosed system has any feature—or any combination of features—that is directed to a technological improvement.
The observations above demonstrates that Applicant’s conclusory assertion regarding the alleged integration of “any” abstract idea into the so-called “specific and concrete improvement in the technical field of AI-driven question answering systems” is not persuasive.
3. Step 2B.
Thirdly, quite similar to the patterns noted above, here also Applicant is referring to the declarant’s remarks made in the SMED. However, the Office has already provided a comprehensive response regarding the SMED; and the response already details the reasons why the declarant’s assertions fail to demonstrate a technological improvement (if any) that the claimed/disclosed system is supposedly implementing. Of course, the Office’s analysis also details the reason why the declarant’s assertions fail to demonstrate whether the claimed/disclosed system is beyond the conventional and generic arrangement of the additional elements. In particular, besides mischaracterizing the existing technology as a nonadaptive system, the declarant is improperly relying on the abstract idea in order to substantiate the alleged “specific, non-routine, and unconventional computer architectures” (see discussion presented above). Consequently, Applicant’s repetitive patterns of relying on the declarant’s assertions also fail to demonstrate whether the claimed/disclosed system is beyond WRCA.
Applicant further asserts that “[e]ven if each individual component — a course database, a question database, a dialogue module, or an AI Bot — could be characterized as conventional in isolation, it is their specific ordered combination operating together in the manner recited in claim 1 that produces the concrete technical improvement and constitutes the inventive concept under BASCOM. The absence of any such prior system in the published literature directly establishes that this ordered combination was not well understood, routine, or conventional in the field, satisfying the Step 2B inquiry” (emphasis added).
However, except for Applicant’s generic/conclusory assumption regarding the alleged “specific ordered combination” of the claimed elements (e.g., the course database, the question database, the dialogue module, the AI Bot), Applicant still fails to demonstrate what claimed/disclosed “manner” of operation is assumed to render the claim beyond WRCA. In contrast, given the claimed/disclosed arrangement of the additional elements, each of the claims—when considered as a whole—is directed to WRCA. In particular, the claimed/disclosed system is a computer-based system that implements the existing artificial inelegance technology—namely, an AI-based chatbot that implements a Large Language Model (LLM). Accordingly, based on the analysis of various data gathered about a student and educational course materials, the AI-based chatbot provides the student—during its interaction with the student—one or more relevant responses, etc. (e.g., see claim 1, or see the original specification). Of course, the claims (including the original disclosure) may describe the specific content or information being analyzed (e.g. a course database storing a course material to be studied by the student; a question database storing exam relevant questions related to the course material, etc.), and/or the specific content/information the system is outputting (e.g., an interactive dialog that regarding topics that are relevant to the course material and/or exam relevant questions, etc.). However, none of these elements changes the conventional nature of the claimed/disclosed system pointed out above, regardless of whether these elements are considered individually or in any ordered combination with any claimed/disclosed features. This is because the specific content of the information is merely an abstract idea.
Of course, the same is true regarding the various prediction scores that the system is generating regarding an input solution, including the alleged “appropriate” problem that the system is selecting based on the predicted values. This is because the above is merely an abstract idea, but not a technological feature, much less an advanced technology.
Applicant is also misapplying BASCOM in further attempt to substantiate the alleged eligibility of the claimed system under Step 2B. However, unlike Applicant’s claimed system, BASCOM does implement a non-generic and non-conventional arrangement of the additional elements. For instance, BASCOM implements Internet content filtering tool, which is installed at a specific location/network remote from the end user; and this solved the limitation of the conventional technology that required users to maintain large databases on local devices. In contrast, given the fact that the claimed/disclosed system is relying merely on the conventional computer/network technology, each of the current claims, when considered as a whole, is directed to the conventional and generic arrangement of the additional elements. Consequently, Applicant’s arguments are not persuasive.
Not surprisingly, Applicant is also blending the issue of novelty (section §102) and/or non-obviousness (section §103) with the issue of eligibility (section §101). However, unlike Applicant’s assertion, neither the alleged novelty nor the alleged non-obviousness of the claimed/disclosed system necessarily signifies eligibility under section §101. In particular, the claimed system (e.g., see claim 1) overcomes the prior art due to the new abstract idea it is reciting even though it is relying merely on the conventional computer/network technology to facilitate the new abstract idea. However, just because the claim overcomes the prior art, this does not necessarily mean that the claim is automatically patent-eligible under section §101. This is once again because a claim for a new abstract idea is still an abstract idea. Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151 (Fed. Cir. 2016). Applicant is also advised to review the MPEP section—namely, MPEP 2106.06 (I)—cited above.
Note that the lack of technological improvement (the inquiry under Prong Two of Step 2A) also confirms the lack of an inventive concept. This is because Step 2B also evaluates, as part of the inquiry regarding an inventive concept, whether the claim implements a technological improvement (see MPEP 2106.05(a), emphasis added),
While improvements were evaluated in Alice Corp. as relevant to the search for an inventive concept (Step 2B), several decisions of the Federal Circuit have also evaluated this consideration when determining whether a claim was directed to an abstract idea (Step 2A). See, e.g., Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1335-36, 118 USPQ2d 1684, 1689 (Fed. Cir. 2016) . . . an examiner should evaluate whether a claim contains an improvement to the functioning of a computer or to any other technology or technical field at Step 2A Prong Two and Step 2B, as well as when considering whether the claim has such self-evident eligibility that it qualifies for the streamlined analysis.
Consequently, Applicant’s new theory, “[t]he absence of any such prior system in the published literature directly establishes that this ordered combination was not well understood, routine, or conventional in the field”, is not valid.
Regarding claim 7, Applicant is once again resorting to the declarant’s assertions presented in the SMED. However, quite similar to the points made so far, the Office has already provided a comprehensive analysis regarding the SMED; and the analysis already details the reasons why the SMED fails to negate the Office’s ineligibility findings. Thus, Applicant is once again advised to review the analysis presented above under the heading “Response to the Declaration under CFR 1.132”.
Applicant further asserts, “[e]ven if each individual component — a knowledge state module, a problem selection module, or a quantitative and qualitative feedback architecture — could be characterized as conventional in isolation, it is their specific ordered combination operating together in the manner recited in claim 7 that produces the concrete technical improvement and constitutes the inventive concept under BASCOM. The absence of any such prior system in the published literature, confirmed both by Mr. Zatkovich's prior art review directly establishes that this ordered combination was not well understood, routine, or conventional in the field, satisfying the Step 2B inquiry” (emphasis added).
However, here also Applicant is relying on a conclusory assertion—namely, the alleged “specific ordered combination” of the claimed modules (i.e., the knowledge state module, the problem selection module, the quantitative feedback module, the qualitative feedback module) to justify the alleged inventive concept. However, except for the generic remark, Applicant still fails to demonstrate whether the alleged “manner” of operation recited in claim 7 is signifying a non-conventional and non-generic arrangement of the additional elements. It is worth noting that the disclosed system encompasses all of the claims. Accordingly, given the fact that the disclosed system as a whole is relying merely on the conventional computer/network technology, then one needs only a common sense to realize that each of the claims is also directed to the conventional and generic arrangement the additional elements.
Here also Applicant is relying on BASCOM and the alleged “absence” of the prior art in further attempt to justify the alleged inventive concept. Accordingly, the discussion presented above with respect to claim 1 also applies to claim 7.
The analysis above confirms that Applicant fails to demonstrate whether any of the current claims—when considered as a whole—is beyond WRCA.
Of course, regarding claim 14, Applicant is as usual resorting to the declarant’s assertions presented in the SMED. In this regard, the Office has already provided full analysis (again see analysis under the heading, “Response to the Declaration under CFR 1.132”), which already details the reasons why the SMED fails to negate the Office’s ineligibility findings. Thus, similar to the points made above, Applicant is once again advised to review the analysis above.
Applicant further asserts, “[e]ven if each individual component — a paraphrase detector module, a question- answer database, a response generation module, or an LLM-based fallback mechanism — could be characterized as conventional in isolation, it is their specific ordered combination operating together as a unified tiered pipeline in the manner recited in claim 14 that produces the concrete technical improvement and constitutes the inventive concept under BASCOM. The absence of any such prior system, as confirmed by Mr. Zatkovich's independent review, directly establishes that this ordered combination was not well understood, routine, or conventional in the field” (emphasis added).
As an initial matter, Applicant’s attempt to justify an alleged inventive concept, while improperly relying on BASCOM and the alleged “absence” of the prior, is once again invalid for the same reasons discussed with respect to claim 1. Thus, Applicant may review the Office’s reasoning above since it is equally applicable to all of the current claims, including claim 14.
In addition, Applicant’s conclusory assertion regarding claim 14 also fails to show whether claim 14 is a non-conventional and non-generic arrangement of the additional elements. Instead, despite the complete reliance on already developed functions of the existing computer/network technology, Applicant is attempting to portray an alleged inventive concept while simply assigning various labels to these existing functions—such as, “a paraphrase detector module”, “a response generation module” and the so-called “LLM-based fallback mechanism”, etc. However, such labeling does not necessarily signify a specific technological feature, much less an arrangement of technological features beyond WRCA. For instance, besides executing one or more algorithms for interpreting the user’s query, exiting AI-based chatbots also access various data sources/databases to retrieve and/or formulate one or more relevant responses to the user. Accordingly, simply labeling one or more of such existing functions as “a paraphrase detector module”, “a response generation module”, etc., does not signify a technological improvement or an inventive concept.
Of course, besides the features noted above, exiting AI-based chatbots also use existing transformer-based Large Language Models as their core engine in order to generate human-like language during interaction. Accordingly, simply labeling this already existing feature as the so-called “LLM-based fallback mechanism” does not demonstrate a technological improvement or an inventive concept. In fact, Applicant’s disclosure already confesses that the system is relying on such existing features, “the AI Bot module may use a large open-source transformer based model to generate a score based on the question, solution input by the student, the reference answer, and the maximum score” ([0098], emphasis added). This confirms that the disclosed system does not even contemplate a specific transformer-based model, much less an advanced one. Nevertheless, pioneers in the AI field, such as Google and OpenAI, introduced transformer-based pretrained language models (e.g., BERT and GPT) half a decade prior to Applicant’s claimed/disclosed system. This fact further confirms that the implementation of transformer-based Large Language Models is already part of the conventional computer/network technology.
Thus, at least for the reasons discussed above, the Office concludes that the claimed/disclosed system fails to implement an inventive concept that amounts to “significantly more” than the abstract idea.
● Considering each of the current claims as a whole, the prior art does not teach or suggest the current claim (regarding the state of the prior art, see the obviousness analysis presented in the office action dated 04/22/2026).
Conclusion
Applicant’s amendment necessitated the new grounds of rejection presented in this final office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 filled 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 extension fee 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 date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BRUK A GEBREMICHAEL whose telephone number is (571) 270-3079. The examiner can normally be reached from 7:00 AM - 3:00 PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, PETER VASAT can be reached on (571) 270-7625. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/BRUK A GEBREMICHAEL/Primary Examiner, Art Unit 3715
