Criticisms of Popular AI X-Risk Models

​Bad takes abound in accelerationist and AI-sceptic circles, but seemingly sensible people are critical of popular AI risk models too. Within EA and AI Safety p(doom)s vary wildly depending on who you talk to. What are the strongest good-faith critiques of the standard x-risk models, and can we learn from them? Even if we're right that AI is a serious risk to humanity, which parts of our current strategies and approaches might be self-defeating?

Please RSVP 24 hours before the event if you're planning to attend, so we can book a table for the right number of people.