KEI at WIPO conversation on intellectual property (IP) and artificial intelligence (AI)

The second session of the WIPO Conversation on IP and AI is being held over three days from July 7 to 9, 2020. WIPO’s web page on the event is here.

I had submitted comments earlier to both WIPO and USPTO on this topic, here, and here. Today we were asked to speak no longer than two or three minutes.

Notes from my oral presentation at the July, 7, 2020 WIPO conversation about artificial intelligence and intellectual property.

I followed Anita Shaw, Intellectual Property Law Counsel for IBM United Kingdom, and began by saying that I agreed with IBM, that it was important to be cautious, and not set norms before understanding the ramifications. I also agreed with IBM that the current patent and copyright laws were designed for humans, and not machines.

I noted that proposals for AI protection often are justified as a protection for investment, and that these do not always end well. The EU regulation of data and broadcaster rights are examples of two unnecessary and badly implemented regimes that have harmed and not benefited innovators. US does not have a sui generis regime for databases, and does not have ROME style broadcaster rights, yet leads in the innovation for streaming of content over the Internet, and in creating and providing database related services.

I noted that exceptions are often left out of new sui generis regimes. Orphan Drug exclusivity and test data exclusivity are examples.

Policy makers need to consider the scale of AI generate output, which is unprecedented, and can quickly fill out a space, initially creating IP thickets.

AI is also likely to involve substantial increasing returns to scale. Access to data by third parties, as an essential facility, may be necessary, to prevent monopolization. Self driving cars and search engines are but two examples. We will likely need less protection of data, not more, to avoid excessive and innovation reducing concentration.

AI protection regimes also pose challenges to disclosure. Trade secrets for algorithms and data run counter to rationale for patents, which is to enhance the public domain.

Copyright terms don’t make sense for AI generated works. It is bad enough we give corporations more than half a century of protection for works for hire. We should not do the same for machines.