DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-20 are pending for examination. Claims 1, 13, and 17 are independent.
Information Disclosure Statement
The information disclosure statement (IDS) was filed on 04/11/2024. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1: The claim is a process, machine, manufacture, or composition of matter.
In the instant application, Claims 1-12 are directed to a machine, Claims 13-16 are directed to a machine, and Claims 17-20 are directed to a process. Thus, each of the claims falls within one of the four statutory categories (i.e. process, machine, manufacture, or composition of matter).
With regard to Claim 1:
2A Prong 1: The claim recites an abstract idea, law of nature, or natural phenomenon.
analyzing the digital content (This step of analyzing data is practically performable in the human mind and is understood to be a recitation of a mental process with the aid of pen and paper (i.e. evaluation).)
based on results of analyzing the digital content, predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine. (This step of predicting a classification is practically performable in the human mind and is understood to be a recitation of a mental process with the aid of pen and paper (i.e. judgement).)
2A Prong 2: The judicial exception is not integrated into a practical application.
A system comprising: a processing device, and memory configured to store a program having logic instructions that, when executed, enable the processing device to perform the steps of (The processing device and memory are understood as mere instructions to apply the exception using a generic computer component — see MPEP 2106.05(f).)
receiving digital content to be tested, (Receiving digital content is understood as an insignificant extra-solution activity — see MPEP 2106.05(g).)
analyzing the digital content with respect to both a human classification model associated with a specific individual and a computer classification model associated with a specific Generative Artificial Intelligence (GenAI) engine, and (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. observation/evaluation/judgement/opinion) — see MPEP 2106.05(f).)
The additional elements as disclosed above alone or in combination do not integrate the judicial exception into practical application as they are applying the exception using a generic computer component or tool in combination with insignificant extra-solution activity.
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
A system comprising: a processing device, and memory configured to store a program having logic instructions that, when executed, enable the processing device to perform the steps of (The processing device and memory are understood as mere instructions to apply the exception using a generic computer component — see MPEP 2106.05(f).)
receiving digital content to be tested, (Receiving digital content is understood as a well-understood, routine, and conventional function, being exemplary of receiving or transmitting data over a network — see MPEP 2106.05(d)(II)(i).)
(This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. observation/evaluation/judgement/opinion) — see MPEP 2106.05(f).)
The additional elements as disclosed above alone or in combination do not recite significantly more than a judicial exception as they are applying the exception using a generic computer component or tool in combination with well-understood, routine, conventional activity.
With regard to Claim 13:
Claim 13 has similar limitations to Claim 1, therefore it is rejected under the same rationale. Additional limitations of Claim 13 are addressed below.
2A Prong 2 & 2B: A non-transitory computer-readable medium configured to store a contribution differentiating program having computer logic with instructions for enabling one or more processing devices to execute steps of: (The non-transitory computer-readable medium is understood as mere instructions to apply the exception using a generic computer component — see MPEP 2106.05(f).)
With regard to Claim 17: Claim 17 has similar limitations to Claim 1, therefore it is rejected under the same rationale.
With regard to Claim 2:
2A Prong 1: The claim recites an abstract idea, law of nature, or natural phenomenon.
The system of claim 1, wherein the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine further includes a sub-step of determining whether a source of consequential portions of the digital content is to be credited to the specific individual or the GenAI engine. (This step of determining the crediting is practically performable in the human mind and is understood to be a recitation of a mental process with the aid of pen and paper (i.e. judgement).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
With regard to Claim 3:
2A Prong 1: The claim recites an abstract idea, law of nature, or natural phenomenon.
The system of claim 1, wherein the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine further includes a sub-step of determining portions of the digital content that are credited to the specific individual and/or GenAI engine. (This step of determining the portions to be accredited is practically performable in the human mind and is understood to be a recitation of a mental process with the aid of pen and paper (i.e. judgement).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
With regard to Claim 4:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2: The judicial exception is not integrated into a practical application.
The system of claim 1, further comprising a step of providing an output including details of a prediction associated with the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine. (Outputting information is understood as an insignificant extra-solution activity — see MPEP 2106.05(g).)
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The system of claim 1, further comprising a step of providing an output including details of a prediction associated with the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine. (Outputting information is understood as a well-understood, routine, and conventional function, being exemplary of presenting offers and gathering statistics — see MPEP 2106.05(d)(II)(iv).)
With regard to Claim 5:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The system of claim 1, wherein the digital content includes software code. (The specification of the digital content is understood to be a field of use limitation — see MPEP 2106.05(h).)
With regard to Claim 6:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The system of claim 5, further comprising a step of training the human classification model by learning programming habits, styles, patterns, syntax, function generation techniques, and human-readable comments of the specific individual from samples of software code obtained from an Integrated Development Environment (IDE) associated with the specific individual. (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
With regard to Claim 7:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The system of claim 1, wherein the human classification model is trained with respect to a group of collaborating individuals, and wherein the computer classification model is trained with respect to a group of GenAI engines. (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
With regard to Claim 8:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The system of claim 1, further comprising steps of training a plurality of human classification models respectively associated with a plurality of individuals, and
training a plurality of computer classification models respectively associated with a plurality of GenAI engines. (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
With regard to Claim 9:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The system of claim 1, further comprising steps of training the human classification model based on a first set of one or more digital content samples verified as being created by the specific individual, and
training the computer classification model based on a second set of one or more digital content samples verified as being created by the specific GenAI engine. (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
With regard to Claim 10:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2: The judicial exception is not integrated into a practical application.
The system of claim 1, further comprising steps of receiving a first set of label information associated with the specific individual for supervised training of the human classification model, and
receiving a second set of label information associated with the specific GenAI engine for supervised training of the computer classification model. (Receiving information is understood as an insignificant extra-solution activity — see MPEP 2106.05(g).)
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The system of claim 1, further comprising steps of receiving a first set of label information associated with the specific individual for supervised training of the human classification model, and
receiving a second set of label information associated with the specific GenAI engine for supervised training of the computer classification model. (Receiving information is understood as a well-understood, routine, and conventional function, being exemplary of receiving or transmitting data — see MPEP 2106.05(d)(II)(i).)
With regard to Claims 11 and 19:
2A Prong 1: The claim recites an abstract idea, law of nature, or natural phenomenon.
predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine (This step of predicting a classification is practically performable in the human mind and is understood to be a recitation of a mental process with the aid of pen and paper (i.e. judgement).)
2A Prong 2 & 2B: The system of claim 1 and the method of claim 17 respectively, wherein the step of predicting (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
With regard to Claims 12 and 20:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The system of claim 1 and the method of claim 17 respectively, wherein the digital content includes one or more of videos, photographs, artwork, Non-Fungible Tokens (NFTs), digital assets, music, news, and literary works. (The specification of the digital content is understood to be a field of use limitation — see MPEP 2106.05(h).)
With regard to Claim 14:
2A Prong 1: The claim recites an abstract idea, law of nature, or natural phenomenon.
The non-transitory computer-readable medium of claim 13, wherein the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine (This step of determining the crediting is practically performable in the human mind and is understood to be a recitation of a mental process with the aid of pen and paper (i.e. judgement).)
determining portions of the digital content that are credited to the specific individual and/or GenAI engine. (This step of determining the portions to be accredited is practically performable in the human mind and is understood to be a recitation of a mental process with the aid of pen and paper (i.e. judgement).)
2A Prong 2 & 2B: wherein the step of predicting ((This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).))
With regard to Claim 15:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2: The judicial exception is not integrated into a practical application.
The non-transitory computer-readable medium of claim 13, wherein the instructions further enable the one or more processing devices to execute a step of providing an output including details of a prediction associated with the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine. (Outputting information is understood as an insignificant extra-solution activity — see MPEP 2106.05(g).)
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The non-transitory computer-readable medium of claim 13, wherein the instructions further enable the one or more processing devices to execute a step of providing an output including details of a prediction associated with the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine. (Outputting information is understood as a well-understood, routine, and conventional function, being exemplary of presenting offers and gathering statistics — see MPEP 2106.05(d)(II)(iv).)
With regard to Claim 16:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The non-transitory computer-readable medium of claim 13, wherein the digital content includes software code, and wherein the instructions further enable the one or more processing devices to execute a step of training the human classification model by learning programming habits, styles, patterns, syntax, function generation techniques, and human-readable comments of the specific individual from samples of software code obtained from an Integrated Development Environment (IDE) associated with the specific individual. (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
With regard to Claim 18:
2A Prong 1: The claim does not recite an Abstract idea.
2A Prong 2 & 2B: The method of claim 17, further comprising steps of: training a plurality of human classification models, each human classification model being associated with one individual or a group of collaborating individuals, each human classification model trained on one or more digital content samples verified as being created by the one individual or group of collaborating individuals, and each human classification model being further trained with supervised label information, and (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
training a plurality of computer classification models respectively associated with a plurality of GenAI engines, each computer classification model trained on one or more digital content samples verified as being created by the respective GenAI engine, and each computer classification model being further trained with supervised label information. (This step is reciting a judicial exception with the words “apply it” (or an equivalent), or merely invoking computers or machinery as a tool to perform the abstract idea (i.e. evaluation) — see MPEP 2106.05(f).)
Claim Rejections - 35 USC § 103
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 (i.e., changing from AIA to pre-AIA ) 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.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-3, 5, 7-14, 17, and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Wang et al. (“Evaluating AIGC Detectors on Code Content”), hereinafter "Wang", in view of Ferrari et al. (“On the Personalization of Classification Models for Human Activity Recognition”), hereinafter "Ferrari".
With regard to Claim 1:
Wang teaches: A system comprising:
a processing device; and memory configured to store a program having logic instructions that, when executed, enable the processing device to perform steps of ([Page 7 Tables 3-4, §3.3.5] Wang discusses evaluation results of numerous detector software such as GPT2-Detector and AITextClassifier that are scored based on accuracy. [Page 5 Col. 2 “AITextClassifier [11]”] Wang specifies that at least this model is a “fine-tuned GPT model”, indicating that it is a software that implies the use of processor and memory in order to function as a GPT type of large language model.)
receiving digital content to be tested, ([Pages 3-4 §3.1] Wang explains how the datasets to be analyzed are sampled from a variety of sources such as [§3.1.2] CodeSearchNet, a digital platform that has samples consisting of code and descriptions across multiple programming languages, as well as [§3.1.1] question-answer samples from Stack Exchange and [§3.1.3] ChatGPT-generated code.)
analyzing the digital content with respect to both a human classification model ([Page 5 §3.2.1] Wang discusses using commercial detectors such as GPTZero which is a classification model that has been trained on “a large corpus of human-written and AI-generated text.” [Page 5 §3.2.2] Wang discusses the use of GPT2-Detector as one of the detectors selected for analysis of digital content, trained specifically from the outputs of the specific GPT-2 engine.)
based on results of analyzing the digital content, predicting whether credit for creating the digital content is to be assigned to the ([Page 7 Tables 3-4, §4.1] Wang displays evaluation results of the various GPT detectors and their accuracies, false positive rates, and false negative rates on “detecting ChatGPT-generated content.” [Page 7 Last ¶] Wang further recites in regard to the evaluation results seen in Tables 3-4 that “some detectors have higher FPR but lower FNR, indicating a tendency to predict positive results (i.e., generated by ChatGPT), while others have higher FNR but lower FPR, indicating a tendency to predict negative results (i.e., generated by humans).”)
Wang does not explicitly teach “a human classification model associated with a specific individual,” but rather a human classification model associated with a group or corpus individuals.
However, Ferrari teaches in the same field of endeavor: a human classification model associated with a specific individual ([Page 32068 Col. 2 ¶2] Ferrari mentions the approach of using a machine learning “subject-dependent (also called personal) model” that is a classification model specific to a final end user and “is able to capture her/his peculiarities.”)
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to have modified Wang’s teachings by limiting the human classification model to an individual instead of a general body or corpus as taught by Ferrari. One would have been motivated to make this modification in order to improve classification testing by dialing in the features that are more specific to an individual as opposed to a larger group that AI/Machine Learning models tend to extract information from. By targeting more specific features instead of averaged features in a group, it may be easier to classify target data as human-generated versus AI-generated.
With regard to Claim 13:
Wang in view of Ferrari teaches: A non-transitory computer-readable medium configured to store a contribution differentiating program having computer logic with instructions for enabling one or more processing devices to execute steps of: ([Page 7 Tables 3-4, §3.3.5] Wang discusses evaluation results of numerous detector software such as GPT2-Detector and AITextClassifier that are scored based on accuracy. [Page 5 Col. 2 “AITextClassifier [11]”] Wang specifies that at least this model is a “fine-tuned GPT model”, indicating that it is a software that implies the use of memory or a non-transitory computer-readable medium in order to function as a GPT type of large language model.)
Claim 13 is a non-transitory computer-readable medium claim that corresponds to Claim 1 and the remaining limitations are rejected for at least the same reasons therein.
With regard to Claim 17: Claim 17 is a method claim that corresponds to Claim 1 and the remaining limitations are rejected for at least the same reasons therein.
With regard to Claim 2:
Wang in view of Ferrari teaches: The system of claim 1, wherein the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine further includes a sub-step of determining whether a source of consequential portions of the digital content is to be credited to the specific individual or the GenAI engine. ([Page 5 Col. 1 ¶2] Wang recites “For code snippets, we only kept code blocks, while removing any extraneous contents surrounding them.” [Page 6 §3.3.3] Wang discloses “the content generated by ChatGPT may not be used directly, and some of the generated content can be modified for customization or to avoid detection,” and further discusses modifying and mutating the code data to reflect these types of customizations to be later tested by the detector software for evaluation. The modifications made for customization here are understood as consequential portions of the digital content, as they are unique and reflect individualistic characterization typically absent from template or predefined code. [Page 9 Table 7, §4.3] Wang illustrates and describes the results of the detectors on the modified consequential portions of code, reusing the AUC, FPR, and FNR metrics first seen in Tables 3-4 on page 7. [Page 7 Last Para] Wang recites in regard to FPR and FNR scores seen in Table 7 among other tables that “some detectors have higher FPR but lower FNR, indicating a tendency to predict positive results (i.e., generated by ChatGPT), while others have higher FNR but lower FPR, indicating a tendency to predict negative results (i.e., generated by humans).”)
With regard to Claim 3:
Wang in view of Ferrari teaches: The system of claim 1, wherein the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine further includes a sub-step of determining portions of the digital content that are credited to the specific individual and/or GenAI engine. ([Page 5 Col. 1 ¶2] Wang recites in regard to datasets like “Wiki-GPT”, “For code snippets, we only kept code blocks, while removing any extraneous contents surrounding them,” wherein the code snippets used as data to be analyzed is understood as portions of the digital content. [Page 7 Tables 3-4, §4.1] Wang displays evaluation results of the detectors on datasets including Wiki-GPT. [Page 7 Last ¶] Wang recites in regard to the evaluation results seen in Tables 3-4 that “some detectors have higher FPR but lower FNR, indicating a tendency to predict positive results (i.e., generated by ChatGPT), while others have higher FNR but lower FPR, indicating a tendency to predict negative results (i.e., generated by humans).”)
With regard to Claim 5:
Wang in view of Ferrari teaches: The system of claim 1, wherein the digital content includes software code. ([Pages 3-4 §3.1] Wang explains how the datasets to be analyzed are sampled from a variety of sources such as [§3.1.2] CodeSearchNet, a digital platform that has samples consisting of code and descriptions across multiple programming languages, as well as [§3.1.3] ChatGPT-generated code.)
With regard to Claim 7:
Wang in view of Ferrari teaches: The system of claim 1, wherein the human classification model is trained with respect to a group of collaborating individuals, and wherein the computer classification model is trained with respect to a group of GenAI engines. ([Page 5 §3.2.1] Wang discusses using commercial detectors such as GPTZero which is a classification model that has been trained on “a large corpus of human-written and AI-generated text,” with human-written text sourcing from student-written articles, news articles, and Q&A datasets across multiple disciplines.)
With regard to Claim 8:
Wang teaches: The system of claim 1, further comprising steps of
training a plurality of computer classification models respectively associated with a plurality of GenAI engines. ([Page 8 Col. 1 ¶1] Wang recites “GPT2-Detector uses the output of GPT-2 for training, while RoBERTa-QA uses the output of ChatGPT.”)
Wang does not explicitly teach: training a plurality of human classification models respectively associated with a plurality of individuals, and
However, Ferrari teaches in the same field of endeavor: training a plurality of human classification models respectively associated with a plurality of individuals, and ([Page 32068 Col. 2 ¶3 Ferrari discusses the “hybrid model” that “uses the end user data and the data of the other users” for its model training. “the classification model is trained both on the data of the users and on a part of the data of the final user.”)
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to have modified Wang’s teachings by training a plurality of models respectively on a plurality of individuals as taught by Ferrari. One would have been motivated to make this modification in order to apply the invention to multiple different instances involving different individuals, while keeping the classification models specific and personalized to each user, allowing for more widespread use of an individualistic application.
With regard to Claim 9:
Wang in view of Ferrari teaches: The system of claim 1, further comprising steps of training the human classification model based on a first set of one or more digital content samples verified as being created by the specific individual, and ([Page 5 Col. 2 “AITextClassifier”] Wang discusses the AITextClassifier detector model that “consists of AI-generated and human-written text where human-written text came from three sources including the Wikipedia dataset, the WebText dataset and a set of human demonstrations from InstrcutGPT.”
training the computer classification model based on a second set of one or more digital content samples verified as being created by the specific GenAI engine. ([Page 5 Col. 2 “GPT2-Detector”] Wang discusses using the GPT2-Detector model, “where the training dataset is from the outputs of the 1.5B-parameter GPT-2 model.”)
With regard to Claim 10:
Wang teaches: The system of claim 1, further comprising steps of receiving a first set of label information ([Page 6 Col. 1 ¶2 “RoBERTa-QA”] Wang discloses that a logistic regression model “leverages the token features from HC3 [“Human ChatGPT Comparison Corpus”] dataset for classification,” wherein the token features are understood as label information and the HC3 dataset includes content specifically generated by human experts.)
receiving a second set of label information associated with the specific GenAI engine for supervised training of the computer classification model. ([Page 6 Col. 1 ¶2 “RoBERTa-QA”] Wang discloses that a logistic regression model “leverages the token features from HC3 [“Human ChatGPT Comparison Corpus”] dataset for classification,” wherein the token features are understood as label information and the HC3 dataset includes content specifically generated by ChatGPT.)
Wang does not explicitly teach: label information associated with the specific individual for supervised training of the human classification model
However, Ferrari teaches in the same field of endeavor: receiving a ([Page 32069 Col. 1 ¶5] Ferrari recites “the proposed method initially uses a subject-independent model, collects and labels the data from the user based on the subject-independent model, trains a subject-dependent model on the collected and labelled data, and classifies activity based on the subject-dependent model,” wherein the subject-independent model is a human classification model.)
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to have modified Wang’s teachings by training a classification model and receiving labelled data specifically on an individual human as taught by Ferrari. One would have been motivated to make this modification in order to improve classification testing by dialing in the features that are more specific to an individual as opposed to a larger group that AI/Machine Learning models tend to extract information from, making it easier for classification models to draw distinctions between (individual) human-generated versus AI-generated.
With regard to Claims 11 and 19:
Wang in view of Ferrari teaches: The system of claim 1 and the method of claim 17 respectively, wherein the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine includes utilizing a Machine Learning (ML) engine encoded with the human classification model and computer classification model. ([Page 5 §3.2.1] Wang discusses using commercial detectors such as GPTZero which is a classification model that has been trained on “a large corpus of human-written and AI-generated text,” and may be understood as a human and computer classification model as it is trained on both human and computer data.)
With regard to Claim 12 and 20:
Wang in view of Ferrari teaches: The system of claim 1 and the method of claim 17 respectively, wherein the digital content includes one or more of videos, photographs, artwork, Non-Fungible Tokens (NFTs), digital assets, music, news, and literary works. ([Page 4 Col. 2 Last Para.] Wang discusses the utilization of “raw text from the Wikipedia dataset” as part of the digital content used for analysis, wherein the text from the Wikipedia dataset is understood as a literary work and/or digital asset.)
With regard to Claim 14:
Wang in view of Ferrari teaches: The non-transitory computer-readable medium of claim 13, wherein the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine further includes enabling the one or more processing devices to execute one or more sub-steps of: determining whether a source of consequential portions of the digital content is to be credited to the specific individual or the GenAI engine, and ([Page 5 Col. 1 ¶2] Wang recites “For code snippets, we only kept code blocks, while removing any extraneous contents surrounding them.” [Page 6 §3.3.3] Wang discloses “the content generated by ChatGPT may not be used directly, and some of the generated content can be modified for customization or to avoid detection,” and further discusses modifying and mutating the code data to reflect these types of customizations to be later tested by the detector software for evaluation. The modifications made for customization here are understood as consequential portions of the digital content, as they are unique and reflect individualistic characterization typically absent from template or predefined code. [Page 9 Table 7, §4.3] Wang illustrates and describes the results of the detectors on the modified consequential portions of code, reusing the AUC, FPR, and FNR metrics first seen in Tables 3-4 on page 7. [Page 7 Last Para] Wang recites in regard to FPR and FNR scores seen in Table 7 among other tables that “some detectors have higher FPR but lower FNR, indicating a tendency to predict positive results (i.e., generated by ChatGPT), while others have higher FNR but lower FPR, indicating a tendency to predict negative results (i.e., generated by humans).”)
determining portions of the digital content that are credited to the specific individual and/or GenAI engine. ([Page 5 Col. 1 ¶2] Wang recites in regard to datasets like “Wiki-GPT”, “For code snippets, we only kept code blocks, while removing any extraneous contents surrounding them,” wherein the code snippets used as data to be analyzed is understood as portions of the digital content. [Page 7 Tables 3-4, §4.1] Wang displays evaluation results of the detectors on datasets including Wiki-GPT. [Page 7 Last ¶] Wang recites in regard to the evaluation results seen in Tables 3-4 that “some detectors have higher FPR but lower FNR, indicating a tendency to predict positive results (i.e., generated by ChatGPT), while others have higher FNR but lower FPR, indicating a tendency to predict negative results (i.e., generated by humans).”)
Claims 4 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Wang in view of Ferrari, further in view of GPTZERO TEAM ("Understanding AI Detection Results with Deep Analysis"), hereinafter “GPTZero”.
With regard to Claim 4:
Wang in view of Ferrari teaches: The system of claim 1,
Wang in view of Ferrari does not explicitly teach: further comprising a step of providing an output including details of a prediction associated with the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine.
However, GPTZero teaches in the same field of endeavor: further comprising a step of providing an output including details of a prediction associated with the step of predicting whether credit for creating the digital content is to be assigned to the specific individual or the GenAI engine. ([Fig. 1; ¶2] GPTZero presents an output display of its Deep Analysis feature, wherein a sample text is provided with indications of orange highlights pertaining to a prediction of AI-generated text and green highlights indicating a prediction of human-generated text.)
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to have modified Wang in view of Ferrari’s teachings by providing details as to whether the classification model predicts a digital content to be assigned to a human or assigned to a generative AI engine as taught by GPTZero. One would have been motivated to make this modification in order to better highlight to a user what sections may be attributed to a human vs GenAI for the sake of more accurately giving due credit to original human creation and ingenuity instead of dismissing it as artificially-generated.
With regard to Claim 15: Claim 15 is a non-transitory computer-readable medium claim that corresponds to Claim 4 and the remaining limitations are rejected for at least the same reasons therein.
Claims 6, 16, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Wang in view of Ferrari, further in view of Obando Chacon et al. (US 2023/0385042 A1), hereinafter “Obando”.
With regard to Claim 6:
Wang in view of Ferrari teaches: The system of claim 5, further comprising a step of
Wang in view of Ferrari does not explicitly teach: training the human classification model by learning programming habits, styles, patterns, syntax, function generation techniques, and human-readable comments of the specific individual from samples of software code obtained from an Integrated Development Environment (IDE) associated with the specific individual.
However, Obando teaches in the same field of endeavor: training the human classification model by learning programming habits, styles, patterns, syntax, function generation techniques, and human-readable comments of the specific individual from samples of software code obtained from an Integrated Development Environment (IDE) associated with the specific individual. ([0108] Obando recites “present a user 104 with opportunities for improvement in their code,” improvements pertaining to “programming style 1118”. [0106] Obando explains that the program may be developed in an integrated development environment (IDE). In regard to improving a user’s code, Obando further teaches a trained machine learning model that is configured or trained to [0060] “perform source code analysis”, [0061] “perform source code comment analysis”, [0063] “ascertain departure from a specified coding pattern”, [0066] “detect programming errors, bugs, stylistic errors, or suspicious constructs”.)
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to have modified Wang in view of Ferrari’s teachings by training the human classification model on specific individual aspects as “programming patterns” and comments as taught by Obando. One would have been motivated to make this modification in order to specifically target the software/programming aspect of an individual contribution, allowing the classification model to be more accurately tuned to a specific task of classifying data in view of a specific programming style.
With regard to Claim 16: Claim 16 is a non-transitory computer-readable medium claim that corresponds to Claim 6 and the remaining limitations are rejected for at least the same reasons therein.
With regard to Claim 18:
Wang teaches: The method of claim 17, further comprising steps of: training a plurality of human classification models, each human classification model being associated with ([Page 5 §3.2.1] Wang discusses using commercial detectors such as GPTZero which is a classification model that has been trained on “a large corpus of human-written and AI-generated text,” with human-written text sourcing from student-written articles, news articles, and Q&A datasets across multiple disciplines. [Page 6 Col. 1 “RoBERTa-QA”] Wang discloses that a logistic regression model “leverages the token features from HC3 [“Human ChatGPT Comparison Corpus”] dataset for classification,” wherein the token features are understood as label information and the HC3 dataset includes content specifically generated by ChatGPT.)
training a plurality of computer classification models respectively associated with a plurality of GenAI engines, each computer classification model trained on one or more digital content samples verified as being created by the respective GenAI engine, and each computer classification model being further trained with supervised label information. ([Page 5 Col. 2 “GPT2-Detector”] Wang discusses using the GPT2-Detector model, “where the training dataset is from the outputs of the 1.5B-parameter GPT-2 model.” [Page 6 Col. 1 “RoBERTa-QA”] Wang discloses that a logistic regression model “leverages the token features from HC3 [“Human ChatGPT Comparison Corpus”] dataset for classification,” wherein the token features are understood as label information and the HC3 dataset includes content specifically generated by ChatGPT.)
Wang does not explicitly teach: a human classification model associated with and trained on one or more digital content samples verified as being created by one individual, and being further trained with supervised label information for that specific individual.
However, Ferrari teaches in the same field of endeavor: a human classification model associated with and trained on one or more digital content samples verified as being created by one individual, and being further trained with supervised label information. ([Page 32073 Col. 1 §C] Ferrari discusses the subject-dependent model, subject-independent model, and hybrid models used for human classification wherein the subject-dependent model is associated with a specific individual user while the subject-independent model is associated with a plurality of users. [Page 32069 Col. 1 ¶5] Ferrari recites “the proposed method initially uses a subject-independent model, collects and labels the data from the user based on the subject-independent model, trains a subject-dependent model on the collected and labelled data, and classifies activity based on the subject-dependent model.”)
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to have modified Wang’s teachings by training the classification model and gathering labeled digital content related to one specific individual or user as taught by Ferrari. One would have been motivated to make this modification in order to improve classification testing by dialing in the features that are more specific to an individual as opposed to a larger group that AI/Machine Learning models tend to extract information from, making it easier for classification models to draw distinctions between (individual) human-generated versus AI-generated.
Wang in view of Ferrari does not explicitly teach: a human classification model associated with and trained on one or more digital content samples verified as being created by a group of collaborating individuals.
However, Obando teaches in the same field of endeavor: a human classification model associated with and trained on one or more digital content samples verified as being created by a group of collaborating individuals. ([0063] Obando recites an example message from the machine learning model, “this code looks different than how your team usually writes this kind of call,” wherein “your team” indicates a group of collaborating individuals. Obando further teaches that the machine learning model performs “program source code analysis” and compared it to “a specified coding pattern” exemplified as code from a user or their team of collaborating individuals, wherein the source code is understood as a digital content verified to be created by the user or team in question.)
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to have modified Wang in view of Ferrari’s teachings by training the classification model and gathering labeled digital content related to a group of collaborating individuals or users as taught by Obando. One would have been motivated to make this modification in order to better analyze and classify digital content that may have been written by large teams or corporations, wherein it may be difficult to distinguish which portions of a large codebase are attributed to which programmer, but the content or style of the code is specific enough to identify the group as a separate entity from the code of a more general population. This allows for more varied application of the invention in determining whether to credit digital content to a specific individual, a team of individuals, or a generative AI model.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Karpman et al. teaches a classifier model that may classify and predict text/image/video data or portions thereof as AI-generated versus human-created.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BRIAN D. BUI whose telephone number is (571)270-0463. The examiner can normally be reached Monday - Friday 8:00am - 5:00pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, ABDULLAH AL KAWSAR can be reached at (571) 270-3169. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/BRIAN D. BUI/Examiner, Art Unit 2127
/ABDULLAH AL KAWSAR/Supervisory Patent Examiner, Art Unit 2127
