The USPTO issued its stance on AI-assisted innovation on February 13, 2024 and July 17, 2020 stating that patent protection will require significant human contribution. https://www.cuecardslegal.com/post/artificial-intelligence-the-us-patent-office-takes-a-stand-on-ai-assisted-inventions
The Office also issued examples of how examiners should review patent applications involving AI going forward. The two examples included a transaxle for a remote controlled car created by AI in response to a prompt, and developing a therapeutic compound for treating cancer, where the AI is used to predict drug target interactions instead of performing wet-lab experiments. https://www.uspto.gov/subscription-center/2024/uspto-issues-inventorship-guidance-and-examples-ai-assisted-inventions
Transaxle Scenario:
In this hypothetical, two engineers Ruth and Morgan work at a toy company that is developing a new RC car. Recognizing that development of a transaxle for the car from scratch is time consuming, they prompt an AI for a preliminary design stating: create an original design for a transaxle for a model car, including a schematic and description of the transaxle.
Ruth and Morgan review the design provided by the AI and find it suitable for their RC car. The company then files a patent application claiming the transaxle produced by the AI. In this scenario, Ruth and Morgan are not proper inventors. Central to the analsysis are the Pannu factors from Pannu v. Jolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998). Under Pannu, a person makes a significant contribution to an invention if they: 1) contributed in some significant manner to the conception of the invention; 2) made a contribution to the claimed invention that is not insignificant in quality, when that contribution is measured against the dimension of the full invention; and 3) did more than merely explain the well-known concepts and/or the current state of the art.
While Ruth and Morgan identified a problem to be solved, i.e. they needed a transaxle, this did not rise to the level of conception. The prompt they provided to the AI was merely a restatement of the problem. In making the prompt, there is no evidence that Ruth and Morgan had made an inventive contribution. When reviewing the output, they agreed the AI design should work, they may have appreciated the invention of the new design, but did not make any significant contribution to its conception. In this scenario, Ruth and Morgan are not inventors and the application would have no significant human contribution. Under the Office's position, this application would not be patentable as solely an AI invention.
A second scenario, where Morgan reviews the AI design and decides to use steel as the material for the transaxle case, leads to the same result since merely substituting a known and commonly used material for transaxle casings does not provide a significant contribution. In a third hypothetical scenario, however, Ruth and Morgan decide to reconfigure the AI transaxle with a horizontal separation of the case. This design change causes them to modify the initial design and perform efficacy experiments that lead to further modification. The claim in the patent application reflects these modifications in the context of the AI design. In this scenario, the change in the AI design and resulting modifications to make it effective are considered significant contributions making Ruth and Morgan proper inventors.
Therapeutic Compound For Cancer Example:
In this example, Marisa a college professor researching a cancer drug uses a deep neural network (DNN) based prediction model to predict drug interactions. Lauren, a data scientist, trained the model using previous drug test experiments. The model Lauren created only accepts ASCII characters for drug compounds. Marisa consults Raghu, an AI expert, to try in silico drug-target interaction (DTI) predictions methods to limit the need for wet lab experiments and speed up drug discovery.
In a first scenario, Marisa and Naz, a postdoctoral fellow, synthesize several drug compounds that may be effective. Wet-lab experimentation shows a preferred method for synthesizing an effective compound. A patent application is filed claiming the method of identifying and synthesizing a lead drug compound. In this scenario, Marisa and Naz are proper inventors having contributed significantly to the method. Applying the Pannu factors, the Office points out that Marisa contributed significantly by developing the preferred treatment therapy, identifying relevant datasets and top compound candidates, and performing wet-lab experiments as well as determining the methodology for preparing and synthesizing the effective compounds. Likewise, Naz significantly contributed by performing wet-lab testing, determining the modifications that would make the compounds effective, and conducting synthetic methods to prepare the effective compound. The significance was noted as being more than merely identifying the problem and going beyond mere explanation of well-known concepts. Their contributions led to the development of the novel method in the claim.
In contrast, Lauren and Raghu would not be proper inventors since their contributions were not significant in the context of the claim. Lauren trained the model based on existing data and maintained the dataset. These were not significant contributions as they were not developed with a specific problem in mind and were not made to elicit a particular output solving a problem. Similarly, Raghu's contribution is limited to prompting the AI model to test the binding affinity of the compounds identified by Marisa and Naz. This contribution is not significant in the full scope of the claim.
A subsequent scenario considers optimization of the drug molecule using a specific generative AI system developed by Raghu. In this scenario, the office considers Raghu and Marisa to be proper inventors for a claim to the optimized drug molecule. Raghu's contributions of developing the AI model needed to identify the optimized drug molecule, identifying proper datasets for the model and work with Marisa to fine tune the model were considered significant.
Overall, the Office has taken the approach of removing the AI contribution from the analysis focusing on human contributions under the guidance of existing case law governing proper naming of inventors. This guidance is useful in developing AI policies that govern the use of AI in product development. Specifically, it is clear that merely having the AI do all of the design work will not yield a patentable invention under the current position that significant human contribution is required. Policies should reinforce the need for significant human involvement in the innovation process and documentation of those contributions to support any attempt to patent resulting innovations. The later scenarios also suggest that those responsible for developing AI models specific to an innovation should be considered as inventors when preparing any patent application. Failure to name all of the proper inventors, can jeopardize the validity of a patent. The nuance reflected in the scenarios also suggests that IP counsel should be involved or at least receive detailed information about the use of AI in the context of any product development before seeking patent protection.