Taking Artificial Intelligence Beyond the Turing Test

34 Pages Posted: 13 May 2020 Last revised: 3 Nov 2020

See all articles by Patric Reinbold

Patric Reinbold

University of Wisconsin - Madison, Law School

Date Written: April 15, 2020

Abstract

Patentability under the 35 U.S.C. § 103 obviousness standard relies on
the ability of a person having ordinary skill in the art; however, the
involvement of artificial intelligence in the inventive process challenges that
standard. A misconception of the current capabilities of AI leads to arguments
of universal obviousness where an independent and creative AI dominates the
inventive process. Fortunately, AI of such a caliber does not yet exist. Thus,
the difficulty of defining the inventive entity threatens patent protection’s
incentive to innovate but need not extinguish the right to patentability
outright. Redefining the standard of obviousness and distinguishing the user’s
contributions enable the patentability of inventions resulting from the use of
AI as an innovative tool.

The proposed standard for the obviousness inquiry—a person having
ordinary skill in AI—accounts for the inventor’s objectives, access to big and
deep data, and knowledge of the existing datasets to control the form and
operation of the machine learning resulting in AI-assisted inventions.
Preexisting common law for patenting chemical compounds and the evolution
of the obviousness test permit shifting the standard to overcome the statutory
hurdles facing the patentability of AI-assisted inventions. However, shifting
the level of ordinary skill in the art to the user does not permit the patentability
of independent and autonomous inventions by inventive AI systems.

The standard proposed here turns on the effort exerted by the inventor
in comparison to the AI system. The inventor’s control and design of the
inventive process help to resolve the level of ordinary skill in the art for AI-assisted inventions by looking at the user’s starting point, motivation based
on the prior art and data, reasonable expectation of success, and control over
the inventive process. Thus, the user’s control defines the level of ordinary
skill in the art and enables the trier of fact to refer to its established inquiries
in determining obviousness.

Keywords: Artificial intelligence, machine learning, patentability, nonobviousness, non-obviousness, AI, AI-assisted inventions, inventive AI

Suggested Citation

Reinbold, Patric, Taking Artificial Intelligence Beyond the Turing Test (April 15, 2020). Wisconsin Law Review, Vol. 2020, Available at SSRN: https://ssrn.com/abstract=3576695 or http://dx.doi.org/10.2139/ssrn.3576695

Patric Reinbold (Contact Author)

University of Wisconsin - Madison, Law School ( email )

Madison, WI
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
291
Abstract Views
1,103
Rank
192,249
PlumX Metrics