Canada: Can AI Be An Inventor? Redux

Advanced artificial intelligence (AI) systems, such as ChatGPT and DABUS, have sparked debates about their potential role in patenting new inventions. Some argue that these large language models (LLMs) should even be considered as named inventors. While previous discussions have mostly disagreed with this view, it brings up further questions about the impact of generative AI on the patent system. In this post, we explore these questions and suggest possible solutions.

Before ChatGPT's launch, multiple patent offices and courts dealt with cases questioning whether DABUS, another AI system, should be considered as an inventor¹. The consensus has been a firm "No," and the topic seems to be following a similar trajectory as celebrity gossip news. However, large language models (LLMs) and other generative AI are likely to play a growing role in the innovation process. This leads to questions about their potential impact on the patent system, even if the idea of naming AI as an inventor is becoming less relevant.

My firm belief that generative AI is neither human nor sentient can be reinforced by examining numerous instances of creativity in human history that emerged from lived experiences. For example, blues music² appears to have originated from the lived experiences of freed Black Americans during the Reconstruction and Jim Crow, leading to a transformative musical genre that has shaped jazz, rock and roll, and most modern pop music³. Until we can confirm a computer's capacity for lived experiences, including the gamut of joy, love, loss and suffering, I suggest that we are far from experiencing a computer having any powerful artistic breakthrough. Major religions like Buddhism and Christianity also focus on the human condition and the kinds of transformations in can bring about in personal and spiritual growth. Much human creativity is inspired this way. There is much to explore in various fields of human endeavor to better comprehend the nature of creativity before attributing it to AI.

However, it is important to acknowledge that generative AI will undoubtedly impact the patent system. Just as internet search engines had a significant influence, generative AI, being even more impressive, cannot be disregarded. It would be unhelpful to claim that AI will have no effect on the patent system.

Wider Impacts to the Patent System

As we seek answers, we will require time, patience, and judicial guidance. For now, I identify three aspects of generative AI that warrant consideration and propose approaches for each: 1) anticipatory AI responses; 2) inventiveness of prompts; and 3) inventive AI responses.

1) Anticipatory AI responses

This concept can be compared to results generated by modern search engines or other resources. If an AI, as of the priority date, produces a response based on search results in its database that matches the claimed invention, it would naturally imply that the invention is unpatentable. However, generative AI often does not cite its sources, so it might require adjustments to provide such citations. Nonetheless, simply generating the response may not be enough to disqualify patentability, as the prompt itself could be inventive, as discussed below.

2) Inventiveness of prompts

When I was trained in patent drafting and "spotting the invention", it was pointed out that oftentimes the inventiveness is in the formulation of the technical problem itself. In other words, sometimes the point of the invention arises at the point of the posing the problem itself.⁴ If accepted within the patent system as a valid inquiry, then part of the prior art analysis may require inventors to keep track of their AI prompts and to submit them with their information disclosure statements (IDS) or otherwise make them available during litigation. If the prompt itself is considered inventive, then the fact that the solution was generated by an AI may be deemed no different than using a microscope or other machinery to solve a well constructed technical problem. Certainly, much in the way of excellent science depends on carefully constructed hypotheses to which the scientific method can be used to test. So a legal test for "what is an inventive prompt" may be necessary.

3) Inventive AI responses

One could consider that even if the prompt is non-inventive, if the AI was unable to generate the inventive solution before the priority date, the inventor's contribution may be sufficiently inventive to uphold patentability. In other words, if the AI could only provide a partial solution based on the prompt, and the inventor built upon that response, the added material from the inventor could be evaluated against the AI response to determine if the inventor contributed "something more" that elevated the invention to a patentable level.

AI as a Tool for Invention

It's important to clarify that, under these tests, I advocate for treating generative AI as a machine or tool that is itself new, much like scientific instruments and computers that have significantly aided the innovation process. However, we should not anthropomorphize these tools as human inventors. Instruments like the slide rule and digital calculator greatly accelerated innovation, but they are tools created by humans. Generative AI is an incredibly clever tool, but it remains just that—a tool. It is now part of the state of the art for innovation, similar to other instruments.

There will be numerous evidentiary issues to address over time. For instance, will it be necessary to maintain dated snapshots of various generative AI engines so that patent offices and courts can examine what the AI was capable of generating at a particular time? During the search process, a patent examiner might input a prompt like, "If I provided the prompt <> before <>, what response would you generate?" This is akin to the Wayback Machine, though generative AI is much more advanced. In this context, generative AI's relationship to inventive step is similar to the Wayback Machine's relationship to anticipation. Ensuring due process and trustworthy results within the legal system will require work.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.