By Michael Stein, Feb. 12, 2024

The United States Patent and Trademark Office (USPTO) has published new examination guidance for determining inventorship of inventions created using artificial intelligence (AI). This guidance comes in response to the recent Executive Order on AI issued by President Biden.

Introduction

The rise of artificial intelligence (AI) is bringing new capabilities to the invention process and raising complex questions about inventorship. If an AI system contributes substantially to developing a new invention, does it qualify as an inventor? The United States Patent and Trademark Office (USPTO) has provided guidance stating that under current law, only natural persons can be named as inventors on patents, even for AI-assisted inventions.

This new guidance aims to balance promoting human innovation and acknowledging AI’s increasing role in inventive work. The USPTO declared that as long as a natural person makes significant contributions to an AI-assisted invention’s conception, they can be considered an inventor. However, AI systems themselves cannot be listed as inventors, even jointly.

The guidance explains factors for evaluating inventorship and determines that AI-assisted inventions are not inherently unpatentable due to improper inventorship if they had significant human contribution. It also covers topics like disclosure duties, correcting inventorship errors, and priority claims.

This directive provides important clarity on inventorship issues involving AI, but it is not the final word. The USPTO is seeking public comment on this guidance and may refine it based on feedback. After the comment period closes, we can expect further examination guidelines as we enter unfamiliar territory at the intersection of AI and intellectual property.

Key Points

• The guidance explains that AI systems cannot be named as inventors on patent applications. Only natural persons who contributed significantly to the conception of the invention can be named as inventors.
• However, the use of AI does not preclude an invention from being patentable. As long as a natural person made significant contributions to the conception of the invention, they can be named as the inventor.
• The guidance provides factors to determine if a natural person’s contributions rise to the level of inventorship, such as contributing to the conception of the invention and not merely reducing it to practice.
• Applicants have a duty to inquire into proper inventorship, including inquiring whether and how AI was used, to ensure the inventors named on the application significantly contributed.
• If an examiner has questions about inventorship, they can require applicants to submit information about inventorship even if it may not be material to patentability.
• Inventorship errors can be corrected pre-issue by filing a request for correction of inventorship. Post-issue, a reissue application is required.
• The guidance aims to incentivize human ingenuity in AI-assisted inventions while also promoting responsible AI use. Further guidance on other patentability issues related to AI is expected.

Examples

Here are some examples of how AI could potentially contribute to the conception of an invention:

• A researcher trains an AI system on a large dataset to identify novel drug candidates. The AI system analyzes the dataset and proposes a new chemical compound predicted to be effective against a disease. If the researcher only provided the data and prompt to the AI system, without contributing to the conception of the specific output compound, they may not qualify as an inventor. However, if the researcher designed the AI system and training process to produce this type of output, they may have significantly contributed to the conception.
• An engineer is trying to design a new airplane wing shape that improves lift and efficiency. They use a generative design AI system to iterate through countless design variations. The engineer selects one of the AI-generated designs, modifies it based on their aerospace expertise, and validates the design in simulation tests. Here, the engineer’s modifications and simulations likely comprise significant contributions to the conception of the final wing design.
• A computer scientist builds an AI system for generating novel logo designs that embody specific concepts. They provide the AI system with a set of keywords describing a new brand. The AI system outputs a logo design representing the brand traits. While the computer scientist designed the AI system, they did not contribute to the conception of the specific logo design and may not qualify as an inventor.
• A chemist is searching for new high-strength polymer materials. They train an AI system on polymer datasets to identify potential atomic arrangements with desired properties. The AI system proposes a new molecular configuration expected to yield a very resilient polymer. The chemist synthesizes the polymer and confirms the properties. Here, the chemist’s synthesis and validation work may allow them to be considered an inventor, even if they did not conceive of the polymer structure.

The key is evaluating whether the natural person made significant contributions to the conception or reduction to practice of the specific invention being claimed, beyond just operating the AI system. 

What is a Significant Contribution?

Here are some key factors to consider when determining if a person’s contributions to an AI-assisted invention qualify them as an inventor:

• Did they identify the problem to be solved and provide creative direction to the AI to guide it towards that specific solution? Simply presenting a general problem may be insufficient.
• Did they design, select, customize, or train the AI model used in a novel way that was critical for deriving the invention? Or was a pre-existing model just operated as a black box?
• Did they choose the specific AI output/result and recognize it as a worthwhile invention at the time, as opposed to just appreciating it after the fact?
• Did they iteratively evaluate and refine the raw AI output through their own expertise to realize the invention?
• Did they contribute additional concepts or modifications that were critical to arriving at the claimed invention versus what the AI originally produced?
• If reducing the invention to practice, did they conduct meaningful validation, analysis, or physical experimentation on the AI output to prove the real-world invention?

Essentially, did the person just operate the AI, or did they also meaningfully contribute to the conception of the specific invention being claimed? That is key to evaluating inventorship.

Conclusion

This new guidance provides important clarity on inventorship issues for AI-assisted inventions. Applicants are encouraged to review the guidance and ensure proper inventorship on patent applications, particularly when AI was involved in the inventive process.