Tu on the Limits of Using Artificial Intelligence and GPT-3 in Patent Prosecution

S. Sean Tu (West Virginia University College of Law) has posted “Limits of Using Artificial Intelligence and GPT-3 in Patent Prosecution” on SSRN. Here is the abstract:

The underlying basis for patent law is a quid pro quo: the inventor discloses how to make and use the invention and in exchange for that disclosure, the public gives the inventor an exclusive right to practice that invention, for a limited time. Pervasive use of Artificial Intelligence (AI) technologies such as Generative Pre-trained Transformer 3 (GPT-3) and other similar AI tools may shift the balance of power envisioned in the patent system away from the public and towards the inventor. Specifically, GPT-3 and other AI products may allow inventors to claim significantly more in their patent than they originally invented, thus inequitably enlarging their exclusive rights without benefiting the public. Patent law has built in tools such as the enablement, utility and definiteness requirements to help limit the possible unjustified expansion of patent rights caused by pervasive use of AI. Additionally, if AI wholly disrupts the quid pro quo established by Congress, then patent law may have to adjust by moving from a peripheral claiming approach back to central claiming approach.