Previously covered on LessWrong here, here, and here. 

Observation #1: The debate isn't really about the proposition — on a very literal reading of the proposition. (Max Tegmark: "We're just arguing that the risk is not zero percent." Yann LeCun: "...the risk is negligible...") 

Opinion #1: A better proposition might be something like, "We should slow down AI capabilities research for the sake of humanity." Or maybe, "Researchers should open source frontier AI models." 

Observation #2: Yann LeCun is claiming that AI safety is (or ought to be) an empirical and iterative process. He believes AI will incrementally progress from mouse-level to human-level and beyond (i.e. there will be no fast takeoff). He's not against AI safety per se; he's just advocating a different approach to AI safety. (Yoshua Bengio: "It's interesting that you've been proposing solutions to the safety problem, which means you believe we need to build safe AI. Which means that there is a problem that needs to be fixed.")

Opinion #2: LeCun catches a lot of flak among people who take AI x-risk very seriously. But I don't see why his proposed approach to AI safety is wrong or reckless or misinformed. In this vein, Sam Altman recently asserted that "safety and capabilities are not these two separate things". I think that's a claim that is worth seriously considering. I think it's worth thinking through the implications of that claim for AI safety. Maybe an empirical, iterative approach to AI safety is the most realistic path forward.

Observation #3: Melanie Mitchell's central view is that superhuman AGI is so far off that it's not worth taking seriously right now. ("We can acknowledge the incredible advances in AI without extrapolating to unfounded speculations of emerging superintelligent AI.") This doesn't seem to be LeCun's view. 

Opinion #3: It would have been more interesting (to me, at least) to get a second debater on the con side who agrees that AGI can easily be made safe and who could give arguments to that effect, rather than just expressing general skepticism about the near-term prospects of AGI.