Paul Gagnon is Assistant General Counsel with Taiga, a manufacturer of electric off-road vehicles and pioneer of the electrification of powersports. Member of the Quebec Bar, Paul also holds a Masters in biotechnology and law (Université de Sherbrooke) and a Masters in intellectual property law and competition law (MIPLC - Max-Planck). In 2020, Paul was listed in the IAM300 as one of the top 300 IP strategists in the world. Co-author of the Montreal Data License, Paul regularly gives conferences dealing with legal issues arising from AI, data licensing and intellectual property.
Misha Benjamin is the incoming General Counsel at Sama, overseeing all legal aspects of their software and services for annotating training and production data for AI. Prior to that, he was Associate General Counsel at McKinsey & Company, where he helped develop risk guardrails and oversaw contractual discussions for advanced analytics and AI engagements. In particular, he helped structure their risk approach to high-risk AI engagements and informed how new AI regulations would shape their business and those of their clients. He first developed his AI expertise at Element AI, where he was responsible for shepherding AI software from research to commercial deployment, formed the company’s position on issues such as the reuse of trained models and monitored and pushed for regulatory oversight of AI. He also pushed for more legal clarity on rights related to training data for AI, resulting in the creation of the Montreal Data License Framework.
Conference : Are lawyers and regulators too focused on AI models?
Thursday 5 may 2022, 16h45 - 17h30 — Amphi mauve
The pace of regulation and legal commentary on AI has been accelerating non-stop for the last few years, but the bulk of that attention has focused on the front end of the value chain - namely models and outputs. We will discuss areas that are still overlooked, the consequences this could have and what practitioners can do to fill those gaps. For instance, the fact that open source licenses don’t properly address important issues when making data (or data labels) available to the public, that data prep/engineering is overlooked in AI contracting and that regulatory oversight has largely failed to focus on the key phases of the AI pipeline.