This week in AI, Natalie and Kobi talk about the global progress on the regulation conversation, and the potential for a global governing body for AI.

They talk about the EU AI Act which has passed another major milestone, and hypothesise about the impact of commercial interests on the effectiveness of that. There was also a senate hearing in the US that Sam Altman (OpenAI) and Gary Marcus (NYU) attended, where they proposed a collaborative international body bringing together academics and the public rather than just corporate/tech interests.

President Biden has also announced a task force investigating the impact of social media on young people. Kobi wonders if some of these things might be a distraction from some of the most important issues, like climate change and the impact of these models and equipment on the environment.

Meta have declared that generative AI is already dead as they go all in on the next generation of AI that is better able to take logic and reasoning into account; however Meredith Whitaker (Signal) has said that while the "godfathers of AI" talk about x-risk (existential risk), they are not talking about a-harms (actual harms), which conveniently means they don't have to take immediate mitigating actions.

Links:

EU AI Act Explained

Yann LeCun of Meta declares generative AI already obsolete

Meredith Whittaker on AI Panic in Slate