Patents - May 14, 2019

Patenting artificial intelligence? Biotech might show us the way

As innovations involving artificial intelligence become more and more prevalent, some issues arise in patenting these innovations. These will become even more pronounced as AI advances to the level where the AI implementations might begin producing inventions themselves.

Why biotechnology? Because biotechnology once faced similar challenges in terms of its compatibility with the existing patent law system.

Similarly to AI, biotechnology deals with elements independent of human intervention (natural process product in biotech, and certain computations/calculations for AI such as AI neural networks), forming a sort of “black box”, which is an integral part of the technology even if the related content of it is, all or in part, unknown to the user.

This fact is at the core of the many issues stemming from the need to ensure not only adequacy of disclosure and assessment of non-obviousness/inventive step but also inventorship and patent eligibility, as the “black box’ relates to the question of eligibility of subject matter in many jurisdictions (e.g. laws of nature and mathematical relationships).

As a consequence, finding, for AI, solutions that are consistent with the solutions adopted for biotech would provide a consistency in the patent legal system. Providing a consistent approach to the protection of both technologies safeguards the legal principles regulating the patenting of innovations, which should be independent of the technology involved.

The challenge for the patent practitioner is that of treating similar features similarly and different features differently for each technology.

Read the full article on IPLaw360 for extended analysis from S+B founder and partner Enrica Bruno and patent attorney Brian Cash about why inventions produced from sentient AI could prove trickier to patent, how secure deposits could aid nonobviousness/inventive step analysis, and whether the practice regarding natural products could also be applied to AI.