Very good analyis of LeCun's thesis that the recent LLM performance increase is largely due to compound reasoning (ie. LLMs asking themself successing, iterative questions) and not an increase in "int...Very good analyis of LeCun's thesis that the recent LLM performance increase is largely due to compound reasoning (ie. LLMs asking themself successing, iterative questions) and not an increase in "intelligence." Newport also discusses LeCun's plan for a new, more modular AI architecture (which has been funded with a billiion+ by Jezz Bezos and his buddies). Very clear, easy to understand, in delpth episode.
Props to Shishir Mehrotra for actually showing up to answer questions about Superhuman - the parent company of Grammarly - cloning known figures as "AI Experts" without their permission - by one of th...Props to Shishir Mehrotra for actually showing up to answer questions about Superhuman - the parent company of Grammarly - cloning known figures as "AI Experts" without their permission - by one of those cloned experts!
We use essential cookies only (for example a session cookie when you sign in). We do not use advertising or tracking cookies. For details see our privacy policy.