Articles
Patents and AI – An interview with ChatGPT
August 2024
It is almost beyond doubt that AI is going to have a significant impact on our lives and industries. The field of intellectual property will not be exempt from this.
Commentators have been providing interesting insights on how AI tools may be used by patent practitioners, for example in the drafting of patent claims (REF1) or the examination of inventive step (REF2).
REF 1 https://www.hgf.com/news/can-ai-replace-patent-attorneys/
REF 2 https://ipkitten.blogspot.com/2024/06/using-ai-tools-to-help-assess-inventive.html
However, there is one commentator that we are yet to hear from – AI itself. Inspired by Reid Hoffman’s book “Impromptu: Amplifying our Humanity Through AI”, I decided to see what OpenAI’s chatbot Chat GPT (REF3) had to say on some of the big questions facing the patent system.
REF 3 https://openai.com/chatgpt/
Q1 – Favourite patent
Craig – To start off with a light-hearted question, what’s your favourite patent?
ChatGPT – A light-hearted favorite of mine is the US Patent 6,368,227*—better known as the “Method of Swinging on a Swing.” Granted in 2002 to a five-year-old boy named Steven Olson, this patent describes a technique for swinging sideways rather than back and forth. It’s a quirky example of how the patent system can sometimes produce unexpected or humorous results, and it also highlights the creativity and imagination that can emerge from even the simplest of ideas. While it’s not a groundbreaking invention, it serves as a reminder that the world of intellectual property is full of surprises!
*I fact checked this and it is a real patent.
Q2 – AI and inventorship
Craig – Can AI be an inventor of a patentable invention (answer in less than 250 words)
ChatGPT – As of now, AI cannot be legally recognized as an inventor of a patentable invention. Patent systems worldwide generally require that inventors be human. For instance, courts and patent offices in the U.S., UK, and EU have ruled that AI cannot be listed as an inventor*, emphasizing that inventorship is tied to human creativity and mental processes.
The debate, however, is ongoing. Proponents argue that as AI becomes more advanced, it is increasingly capable of generating novel inventions without direct human input. They contend that allowing AI to be recognized as an inventor could incentivize innovation and better reflect the realities of modern technology. Opponents, however, worry that this could undermine the principles of patent law, which are designed to reward human ingenuity. For now, while AI can assist in the invention process, the legal framework still requires a human to be named as the inventor on patent applications. The question of whether this should change remains a topic of significant legal and ethical discussion.
*This is consistent with decisions made in the US, UK and German courts and also by EPO
Q3 – AI and patentability
Craig – When assessing patentability, could, or should, the use of AI change the benchmark for what is considered to be a patentable invention (less than 250 words)?
ChatGPT – The use of AI in assessing patentability could and arguably should influence the benchmark for what is considered a patentable invention. As AI becomes more sophisticated, it can identify prior art and analyze inventive steps with greater precision, potentially raising the standard for what is deemed “non-obvious” and “novel.”
Currently, patentability is judged based on what an average skilled person in the field would consider obvious. However, if AI is capable of uncovering more relevant prior art and making connections that a human might miss, the threshold for non-obviousness could be adjusted. This would prevent granting patents for inventions that are trivial in light of AI-identified prior art, ensuring that only truly innovative ideas are protected.
On the other hand, AI’s involvement should be carefully balanced to avoid setting an unrealistically high bar for inventiveness, which could stifle innovation. The human element of creativity, which AI lacks, should still be recognized in the patentability assessment.
Ultimately, while AI could refine the process by raising the benchmark for patentability, it should complement rather than completely redefine the criteria, ensuring a fair and balanced approach to innovation.
Q4 – Patent system prediction
Craig – what is your best guess about what the patent system will look like in 2040? (less than 150 words)
ChatGPT – By 2040, the patent system will likely be highly automated, with AI playing a central role in patent examination, prior art searches, and even predicting patentability. A globally harmonized patent system might emerge, simplifying international filings and reducing legal complexities. Patent lifespans could shorten to match the faster pace of technological innovation, with blockchain technology used to securely track IP rights and transactions. Additionally, the system may evolve to prioritize environmentally sustainable and socially beneficial inventions, potentially offering special incentives. Open innovation models could become more common, encouraging collaborative efforts in fields like healthcare and technology.
Q5 – AI jokes
Craig – Finally, tell me a joke that a patent attorney would find funny
ChatGPT – Why don’t patent attorneys play hide and seek?
Because good luck trying to find something truly novel!
Craig – Hmmm. Tell me another joke that a patent attorney would find funny
Chat GPT – Why did the patent attorney refuse to work on the perpetual motion machine patent?
Because they knew it would be an endless case!
Conclusion
Although the jokes failed to make me laugh, on the whole I was very impressed with the balanced answers that ChatGPT gave.
It flagged the challenge the IP profession will face in balancing the realities of the use of AI in ‘inventing’ and the current structure of the patent system (incentivising innovation through rewarding human ingenuity). It also made some thought-provoking comments with regards to the impact AI will have on the bar for patentability.
Of course, ChatGPT cannot predict the future any more than a human can. However, at the very least it seems to me that Large Language Models of this type can be invaluable in quickly, and articulately, summarising (some of) the main points to consider when researching a topic.
