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 .
Claim Objections
Claims 9 and 18 are objected to because of the following informalities: claims 9 and 18 recite “an identify of an industrial asset”, it’s probably supposed to be an identity of an industrial asset. Appropriate correction is required.
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 of a mental concept and mathematical relationship without significantly more. The claims recite a mental concept of:
1… natural language input, a query describing a performance issue relating to an … for which technical support is requested; and
… in response to receipt of the query, formulate a prompt, directed to a generative AI model, designed to obtain a response from the generative AI model comprising information …generate a natural language technical support response describing a recommendation for addressing the performance issue, wherein … generates the prompt based on analysis of the query and a selected subset of industrial training data encoded in one or more custom models,
…
9. The system of claim 1, wherein … formulate the prompt to include at least one of information extracted or inferred from the query, an identify of an industrial asset affected by the performance issue, a description of the performance issue being experienced, a type of industrial application being performed by the industrial automation system, an industrial vertical in which the industrial automation system operates, or the selected subset of the industrial training data.
Applicant claims a mathematical relationship of:
8. …generative AI component is configured to formulate the prompt directed to the generative AI model in response to inferring that the response from the generative AI model will cause the natural language technical support response to have a probability of accurately addressing the performance issue described by the query that exceeds a probability threshold.
This judicial exception is not integrated into a practical application because the additional limitations of an industrial control system, domain specific data, and industrial domain merely link the abstract ideas to a technical field. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the memory and processor are generic computer parts, and the training and domain-specific training are generic model training using collected information with no particular technological improvement. Further, the steps of collecting data and displaying results are insignificant extra-solution activity.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, because the specification, while being enabling for and AI mode being an LLM or GPT, does not reasonably provide enablement for all generative AI models, including the listed “diffusion model… VAE… GAN… or other such models.” Spec. 44 The specification does not enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make the invention commensurate in scope with these claims. The not-enabled models don’t usually take a prompt and output natural language, not without some explanation. Therefore the scope of what is claimed “diffusion model… VAE… GAN… or other such” generative AI models is beyond what is enabled by the specification. Spec. 44.
Claims 8 and 17 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Applicant claims, “a probability of accurately addressing the performance issue described by the query…” Claims 8 and 17. There is no description of how to formulate this probability in the specification. Spec. 24 states, “generate a probability distribution over states…” However, this doesn’t describe what is claimed. The other mentions of probability in the specification does nothing to describe how to formulate a probability of accurately addressing the performance issue.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
Claims 8 and 15-17 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
The term “accurately” in claims 8 and 17 is a relative term which renders the claim indefinite. The term “accurately” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. Further, Applicant does not describe how to formulate a probability that a response accurately addresses a query.
Claim 15 is probably supposed to depend on claim 10, if claim 15 is properly dependent on claim 1 then it is a duplicate claim of claim 6 and has an indefinite preamble.
Claim 16 should probably depend on claim 15, as it is right now depending on itself, which makes it indefinite.
The following is a quotation of 35 U.S.C. 112(d):
(d) REFERENCE IN DEPENDENT FORMS.—Subject to subsection (e), a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
Claim 16 is rejected under 35 U.S.C. 112(d) or pre-AIA 35 U.S.C. 112, 4th paragraph, as being of improper dependent form for failing to further limit the subject matter of the claim upon which it depends, or for failing to include all the limitations of the claim upon which it depends. Claim 16 depends on itself. Applicant may cancel the claim(s), amend the claim(s) to place the claim(s) in proper dependent form, rewrite the claim(s) in independent form, or present a sufficient showing that the dependent claim(s) complies with the statutory requirements.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over WO2025003049A1 to Brehas et al and US20230419079A1 to Shazeer et al. Brehas is available under 102(a)(2) because MPEP 2154.01 states,
AIA 35 U.S.C. 102(a)(2) sets forth three types of patent documents that are available as prior art as of the date they were effectively filed with respect to the subject matter relied upon in the document if they name another inventor: (1) U.S. patents; (2) U.S. patent application publications; and (3) certain WIPO published applications. These documents are referred to collectively as ‘U.S. patent documents.’
(emphasis added).
MPEP 2154.01(b) goes on to say,
The AIA also eliminates the so-called Hilmer doctrine. Under the Hilmer doctrine, pre-AIA 35 U.S.C. 102(e) limited the effective filing date for U.S. patents (and published applications) as prior art to their earliest U.S. filing date. In re Hilmer, 359 F.2d 859, 149 USPQ 480 (CCPA 1966). In contrast, AIA 35 U.S.C. 102(d) provides that if the U.S. patent document claims priority to one or more prior-filed foreign or international applications under 35 U.S.C. 119 or 365, the patent or published application was effectively filed on the filing date of the earliest such application that describes the subject matter. Therefore, if the subject matter relied upon is described in the application to which there is a priority or benefit claim, the U.S. patent document is effective as prior art as of the filing date of the earliest such application, regardless of where filed.
(emphasis added)
Brehas teaches claims 1, 10 and 19. A system, comprising:
a memory that stores executable components; and
a processor, operatively coupled to the memory, that executes the executable components, the executable components comprising: (Brehas abs “data processing apparatus (4)”)
a user interface component configured to receive, from a client device as natural language input, a query describing a performance issue relating to an industrial automation system for which technical support is requested; and (Brehas p. 28 ln. 28 “the language model 42 receives as input a problem statement 402 in natural language from a user indicating a problem that has occurred during operation of the packaging line 1. The problem statement 402 may be received via the user interface 10, in particular via an input device thereof. “)
a generative artificial intelligence (AI) component configured to, in response to receipt of the query, formulate a prompt, directed to a generative AI model, designed to obtain a response from the generative AI model comprising information used by the generative AI component to generate a natural language technical support response describing a recommendation for addressing the performance issue, wherein the generative AI component generates the prompt based on analysis of the query and a selected subset of industrial training data (Brehas “the step of performing one or more conversation cycles is based on Chain of Thought (CoT) prompting. CoT prompting should be understood as a prompting method which encourages the language model to explain its reasoning. … One approach is to employ recurrent neural network (RNN) or transformer-based models that have the ability to retain and utilize information from previous steps or tokens. These models can be trained to learn the dependencies between prompts and generate responses that align with the ongoing conversation.” The RNN is the gen AI component feeding into a gen AI model “Examples of language models include, without limitation, GPT-3 and GPT-4 from OpenAI, LLaMA from Meta, and PaLM2…” Brehas.
wherein the user interface component is configured to render the natural language technical support response on the client device. (Brehas abs “generating (604), by a trained language model (42) and based at least in part on the problem statement (402), one or more instructions (404) indicating operations to be performed by the user on the packaging line (1) to alleviate the problem; and outputting (606), via an output device, the one or more instructions (404) in natural language to the user.”)
Brehas doesn’t teach multiple models receiving input from the RNN.
However, Shazeer teaches a generative artificial intelligence (AI) component configured to, in response to receipt of the query, formulate a prompt, directed to a generative AI model, designed to obtain a response from the generative AI model comprising information used by the generative AI component to generate a natural language technical support response describing a recommendation for addressing the performance issue, wherein the generative AI component generates the prompt based on analysis of the query and a selected subset of industrial training data encoded in one or more custom models. (Shazeer fig. 1 shows AI component 104 fed to models 114-122, below)
PNG
media_image1.png
541
706
media_image1.png
Greyscale
Shazeer, Brehas and the claims are all AI models. It would have been obvious to a person having ordinary skill in the art, at the time of filing, to implement Shazeer’s mixture of experts architecture in Brehas’ Chain of Thought because in Shazeer “only a small number of the expert neural networks are selected during the processing of any given network input by the neural network, and hence the processing time and computing resources necessary to process an inference using the neural network can be maintained at a reasonable level.” Shazeer para 10.
Brehas teaches claims 2, 11 and 20. The system of claim 1, further comprising a training component configured to train the one or more custom models with the industrial training data, wherein the industrial training data comprises at least one of libraries of product manuals for different types of industrial devices or software platforms, help files, vendor knowledgebase data, information defining industrial standards, technical specifics for different types of industrial control applications, information describing specifics of different industrial verticals, information regarding industrial best practices, or archived technical support chat sessions with the system. (Brehas “According to another aspect of the present disclosure, the language model has been trained using at least one operation manual associated with the packaging line. The operation manual may be a comprehensive document that provides detailed instructions, guidelines, and/or information on how to properly operate, maintain, troubleshoot, and/or perform routine tasks on the packaging line.” A product manual is an operation manual. Brehas “Fig. 2 illustrates a conceptual block diagram 200 of training data usable for training the language model 42 in accordance with embodiments of the present disclosure. As can be seen, the language model 42 can be used using one or more operation manuals 202…”)
Brehas doesn’t teach multiple models.
However, Shazeer fig. 1 teaches multiple models as expert NNs.
Brehas teaches claims 3 and 12. The system of claim 1, wherein the query comprises at least one of a description of observed behavior of a device or machine of the industrial automation system, a description of an error code or alarm observed on an industrial device, a request for example control code for performing a described control function, a request for recommended configuration settings for an industrial device that will cause the industrial device to operate in a described manner, a question regarding how to perform a specified maintenance task on a machine of the industrial automation system, a question regarding an estimated amount of time to perform a specified maintenance task, or a request for suggested maintenance actions to perform on a device or machine of the industrial automation system. (Brehas “Non-limiting examples of problem statements may include: Package number 20 has a dent on the top.The fill level of the liquid in the packages varies inconsistently, resulting in underfilled (or overfilled) packages. The packaging line frequently experiences leaks or spills during the filling (or sealing) process. Packages are not properly sealed. The labeling process results in misaligned (or unreadable) labels on the packages. The packaging line experiences jams or blockages.”)
Brehas teaches claims 4 and 13. The system of claim 1, wherein
the industrial training data comprises at least archived technical support chat sessions comprising previous queries submitted to the system and corresponding technical support responses generated by the generative AI component, and the selected subset of the industrial training data comprises a subset of the archived technical support chat sessions determined to address a technical support issue similar to the performance issue described by the query. (Brehas “According to another aspect of the present disclosure, the language model has been trained using stored information which indicates a previously encountered problem and a corresponding previously generated one or more instructions indicating operations to be performed to alleviate the problem.”)
Brehas teaches claims 5 and 14. The system of claim 1, further comprising a training component configured to train the one or more custom models using the query and the natural language technical support response. (Brehas “the language model has been trained using stored information which indicates a previously encountered problem and a corresponding previously generated one or more instructions indicating operations to be performed to alleviate the problem. Accordingly, a feedback loop may be provided in order to (re-)train the language model with problems that have already occurred, and/or corresponding recommended one or more instructions.” This retraining is what teaches using the query and response to train on.)
Brehas teaches claims 6 and 15. The system of claim 1, wherein
the system stores multiple sets of domain-specific custom models, including the one or more custom models, that are trained with respective sets of domain-specific training data corresponding to respective different industrial domains, and (Brehas “the language model has been trained using stored information which indicates a previously encountered problem and a corresponding previously generated one or more instructions indicating operations to be performed to alleviate the problem.” Two different domains are taught here, Brehas “Non-limiting examples of problem statements may include: Package number 20 has a dent on the top. The fill level of the liquid in the packages varies inconsistently, resulting in underfilled (or overfilled) packages.” Brehas just doesn’t have several models at once. Each type of problem is its own domain.)
the generative AI component is configured to at least one of formulate the prompt or the natural language technical support response based on analysis of the query and the selected subset of the industrial training data encoded in a set of domain-specific custom models, of the multiple sets of domain-specific custom models, corresponding to an industrial domain to which the query pertains. (Brehas “One approach is to employ recurrent neural network (RNN) or transformer-based models that have the ability to retain and utilize information from previous steps or tokens.”)
Brehas doesn’t teach multiple models.
However, Shazeer fig. 1 teaches multiple models as expert NNs.
Brehas teaches claims 7 and 16. The system of claim 6, wherein the industrial domain is at least one of food and beverage, pharmaceutical, automotive, textiles, mining, oil and gas, power generation, semiconductors, or life sciences. (Brehas “Non-limiting examples of problem statements may include: Package number 20 has a dent on the top. The fill level of the liquid in the packages varies inconsistently, resulting in underfilled (or overfilled) packages.” Brehas “non-limiting example in the figures a filling machine 2, configured for producing packages from a tube of packaging material filled with pourable (food) product.”)
Brehas teaches claims 8 and 17. The system of claim 1, wherein the generative AI component is configured to formulate the prompt directed to the generative AI model in response to inferring that the response from the generative AI model will cause the natural language technical support response to have a probability of accurately addressing the performance issue described by the query that exceeds a probability threshold. (Brehas “The one or more conversation cycles may be performed until the one or more candidate instructions reach a predetermined confidence score.” Confidence score is the probability of accurately addressing the performance issued described in the query.)
Brehas teaches claims 9 and 18. The system of claim 1, wherein the generative AI component is configured to formulate the prompt to include at least one of information extracted or inferred from the query, an identify of an industrial asset affected by the performance issue, a description of the performance issue being experienced, a type of industrial application being performed by the industrial automation system, an industrial vertical in which the industrial automation system operates, or the selected subset of the industrial training data. (Brehas “One approach is to employ recurrent neural network (RNN) or transformer-based models that have the ability to retain and utilize information from previous steps or tokens. These models can be trained to learn the dependencies between prompts and generate responses that align with the ongoing conversation.”)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Austin Hicks whose telephone number is (571)270-3377. The examiner can normally be reached Monday - Thursday 8-4 PST.
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, Mariela Reyes can be reached at (571) 270-1006. 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.
/AUSTIN HICKS/ Primary Examiner, Art Unit 2142
Prosecution Insights