Four Major AI Advancements In One Week!New Scaling Laws for Large Language Models – essentially everyone has been training large language models with a deeply suboptimal use of computeGoogles Pathways Language Model explains jokes, and it can distinguish cause and effect and understand counterfactualsDALL-E 2 makes amazing art on demandSocratic Modeling systemAlso mentioned:Overcoming Bias: AI Language ProgressWhat an actually pessimistic containment strategy looks likeGithub CopilotAnthropic AIThe Mind KillerHey look, we have a discord! What could possibly go wrong?Also merch!Rationality: From AI to Zombies, The Podcast, and the other podcastLessWrong posts Discussed in this Episode:Every Cause Wants To Be A CultReversed Stupidity Is Not IntelligenceNext Episode’s Sequence Posts:Argument Screens Off AuthorityHug the Query(edit: misspellings is how we prove we aren’t AIs :P)