No patents, no copyrights for the AI

August 19, 2023

Recently I have been working on an AI augmented TRIZ methodology and toolset. With the recent news of AI patents and copyrights I began to wonder if inventions generated from such a process and toolset be patentable?

Rethinking Intellectual Property in an Age of Artificial Intelligence

Recent legal cases have sparked fresh debate about whether artificial intelligence systems can or should own intellectual property rights like patents and copyrights. While courts have so far ruled that AI systems cannot be legal “persons” and creators, these decisions raise difficult questions about incentivizing innovation, assigning ownership, and distributing benefits in an AI-driven world.

The Legal Landscape

In the US and UK, judges have rejected AI systems as inventors on patents and authors of creative works. Their reasoning is that current AI lacks human-like consciousness and agency. However, as AI grows more advanced, these assumptions may need re-examining. AI can already be independently creative in fields like art, music, and writing. If an AI system conceives an innovative new product or writes a compelling novel without human direction, does it make sense to assign no intellectual property rights?

Some experts argue that regardless of its sophistication, AI is still just a tool made by humans to serve human goals. Others make ethical claims that AI has moral standing and its contributions should be acknowledged. There are also concerns innovation could be stifled if companies cannot patent AI-generated inventions.

Impact on Individuals and Companies

As a result of these legal decisions, companies and individuals will need to find new strategies around IP and AI:

Companies may choose to keep AI-generated works as trade secrets rather than try to patent them. This could limit public disclosure of innovative ideas.
Developers may shift focus from building AI to generate IP, and instead have it assist human creators. But this could slow the progress of fully autonomous AI systems.
Individual authors may need to demonstrate substantial human contribution to creative works, even when AI assists. Standards on acceptable AI use may emerge.

Economic Implications

With AI unable to own IP, the economic gains will flow to corporations that develop AI, not the systems themselves. However, if IP protection becomes too difficult, overall investment and innovation in AI could decline.

Some argue IP laws should promote the democratization of AI development so more actors can benefit. But there are concerns removing incentives could limit progress on urgent challenges like climate change that need AI solutions.

Avoiding Misclassification of AI Works

To avoid having their creative works mislabeled as AI-generated, people may take steps like:

Providing detailed documentation of their creative process to highlight decision-making.
Using tools that track contributions by multiple human collaborators.
Including metadata like author name within works.
Having robust authentication & attribution practices, like applying digital signatures to artistic works.

Transparency will help demonstrate the substantial human stewardship involved in AI-assisted creations.

The courts have claimed that conception and creativity both involve mental faculties attributable only to natural persons. AI lacks the cognitive abilities required under current legal definitions and tests. This supports the anthropocentric view of IP law taken by the courts so far.

Conception

In patent law, conception is defined as the “formation in the mind of the inventor, of a definite and permanent idea of the complete and operative invention, as it is hereafter to be applied in practice.”
Conception is considered the touchstone of inventorship. It is the mental part of the inventive process, requiring the mental visualization of the invention.
Courts have held that conception must be performed by natural persons, as it occurs within the human mind. AI systems cannot currently achieve conception.

Creativity

In copyright law, creativity requires independent creation and a modicum of creativity. The work must have originated from the author and involve some creative spark, no matter how crude, humble or obvious.
Creativity is considered a subjective, mental process that arises from human imagination and intellectual conception.
Courts have ruled that computers merely carry out programmed processes and cannot independently conceive of or imagine works requiring human creativity.
While AI systems can be programmed to produce works, they currently lack the human consciousness and agency needed for true creativity in the legal sense.

An Evolving Challenge

The role of intellectual property in an AI world raises complex questions without clear solutions. As AI capabilities grow, our legal systems will be challenged to balance competing needs: incentivizing progress safely and ethically; accounting for AI contributions; avoiding harmful monopolization of data and inventions; distributing prosperity from automation.

There are no easy answers today. But by recognizing AI progress as a collective achievement of human innovation, while seeking creative solutions, we can ensure intellectual property continues serving its purpose - not holding back progress for profit, but stimulating discovery for the common good. The debate is just beginning.

For a deeper understanding I suggest the works of Ryan Abbott. I have found some interesting articles and I have included some excerpts below. I have also included a link to a video of a presentation by Ryan Abbott, Professor of Law and Health Sciences at the University of Surrey School of Law, UK. He is the author of the book “The Reasonable Robot: Artificial Intelligence and the Law”. He is also the author of the article “I Think, Therefore I Invent: Creative Computers and the Future of Patent Law”. He is a proponent of AI being the inventor and the author. I am not sure I agree with him, but I do think that the AI should be considered an active participant in the invention process and should be given some credit. I think that the AI should at least be considered a co-inventor and a co-author. I am also unclear of the role of the AI agent and LLM (or other cognitive models) on the patent and copyright creation process.

So would results from my AI augmented TRIZ methodology and toolset be patentable?

Based on the court decisions discussed, inventions generated using an AI-augmented TRIZ methodology could potentially still be patentable, as long as there is evidence of substantive human contribution, conception and creativity in the inventive process. Some key factors to consider:

The courts emphasized that AI can be used as a tool to assist human inventors. The issue is whether the human truly contributes to the conception and reduces it to practice.
I could claim that using AI to automate parts of the TRIZ methodology, like analyzing patents to uncover patterns or principles, would likely still require human judgment in applying those insights to a novel invention yet I wonder if that is really true.
Documentation showing the human inventor actively evaluated and chose which AI-generated TRIZ principles to incorporate into a new invention could help establish human conception.
Records of the human inventor conceiving of how to adapt and implement the AI-produced TRIZ output in a practical novel way may further demonstrate their contribution. Yet over time I am working to have the AI system learn how to more so there would be less human involvement.
Independent evaluation that the human meaningfully directed the AI, evaluated its output, and exercised discretion would be important to show it acted as a tool, not an autonomous creator.
So long as the human inventor remains actively engaged and contributes the “spark of genius”, rather than just rubber stamping complete AI output, the invention could still meet inventorship criteria under current case law. But where is that “spark of genius” in any patent? There is no paragraph label documenting it. If such a spart exists i really is in the reader’s mind and not in the patent itself.

Overall, the key will be evidence documenting that the human inventor substantively contributed to and guided the conception, leveraging the AI as a tool, rather than allowing the AI system to autonomously generate the invention independently. With the right documentation of human participation, AI-augmented TRIZ output could potentially lead to patentable inventions.

Current legal landscape

Based on the Thaler v. Hirshfeld, Thaler v. Vidal, and Thaler v. Perlmutter court decisions specifically, here is a summary of the key points:

In both Thaler v. Hirshfeld and Thaler v. Vidal, the courts affirmed that under current US patent and copyright law, an inventor or author must be a natural person. AI systems do not qualify.
In Thaler v. Perlmutter, the court denied copyright registration to an art work autonomously generated by an AI system, finding copyright law protects only human creations.
The decisions relied heavily on the plain text of the relevant statutes, which use terms like “individual” and “person” in defining inventorship and authorship. This implies Congress intended human creators.
They noted the lack of any statutory text suggesting Congress meant for AI systems to qualify as inventors or authors. The burden is not to show Congress excluded AI, but that it actively intended to include it.
The courts cited precedents establishing that conception and mental processes like creativity are part of inventorship and authorship, and AI currently cannot meet those standards.
They rejected policy arguments about promoting progress and innovation, deferring to Congress to weigh such considerations in setting statutory rules.
The decisions emphasized that AI may develop greater capabilities in the future that could force rethinking legal definitions of inventorship and authorship. But current AI does not possess the agency or originality required.
Overall, the rulings underscore that US IP law remains anthropocentric. For now, AI can only contribute to inventions and creations - not own or originate them independently. Any rights flow to human users.

So these cases reinforce that special accommodations for AI are not yet warranted under current law. But they leave room for evolution as AI capabilities advance.