Submissions from lesswrong.com

		We're running out of benchmarks to upper bound AI capabilities (lesswrong.com)
		15 points by gmays 1 day ago \| past \| 9 comments
		AIs can now do easy-to-verify SWE tasks, I've shortened timelines (lesswrong.com)
		3 points by gmays 2 days ago \| past \| discuss
		The effects of caffeine consumption do not decay with a ~5 hour half-life (lesswrong.com)
		99 points by swah 2 days ago \| past \| 104 comments
		My Picture of the Present in AI (lesswrong.com)
		1 point by speckx 2 days ago \| past \| discuss
		Most people can't juggle one ball (lesswrong.com)
		106 points by surprisetalk 3 days ago \| past \| 37 comments
		"Alignment" and "Safety", Part One: What Is "AI Safety"? (lesswrong.com)
		1 point by joozio 5 days ago \| past \| discuss
		Paper Close Reading: "Why Language Models Hallucinate" (lesswrong.com)
		2 points by joozio 6 days ago \| past \| discuss
		Estimates of the expected utility gain of AI Safety Research (lesswrong.com)
		1 point by joozio 6 days ago \| past \| discuss
		What I like about MATS and Research Management (lesswrong.com)
		2 points by joozio 7 days ago \| past \| discuss
		Predicting When RL Training Breaks Chain-of-Thought Monitorability (lesswrong.com)
		2 points by gmays 7 days ago \| past \| discuss
		AI Safety at the Frontier: Paper Highlights of February and March 2026 (lesswrong.com)
		2 points by joozio 8 days ago \| past \| discuss
		How to emotionally grasp the risks of AI Safety (lesswrong.com)
		3 points by joozio 8 days ago \| past \| discuss
		You can't imitation-learn how to continual-learn (lesswrong.com)
		2 points by paulpauper 9 days ago \| past \| discuss
		A Mirror Test for LLMs (lesswrong.com)
		2 points by gmays 10 days ago \| past \| discuss
		I'm Suing Anthropic for Unauthorized Use of My Personality (lesswrong.com)
		5 points by usrme 10 days ago \| past \| 2 comments
		Why did everything take so long? (lesswrong.com)
		2 points by jstanley 11 days ago \| past \| discuss
		The state of AI safety in four fake graphs (lesswrong.com)
		3 points by allenleee 11 days ago \| past \| discuss
		Gyre (lesswrong.com)
		3 points by jstanley 11 days ago \| past \| discuss
		Less Dead (lesswrong.com)
		2 points by paulpauper 12 days ago \| past \| discuss
		Using complex polynomials to approximate arbitrary continuous functions (2025) (lesswrong.com)
		1 point by measurablefunc 12 days ago \| past \| discuss
		The Terrarium (lesswrong.com)
		1 point by johnfn 12 days ago \| past \| discuss
		AI's capability improvements haven't come from it getting less affordable (lesswrong.com)
		3 points by gmays 12 days ago \| past \| discuss
		I am definitely missing the pre-AI writing era (lesswrong.com)
		322 points by joozio 13 days ago \| past \| 240 comments
		Stanley Milgram wasn't pessimistic enough about human nature? (lesswrong.com)
		7 points by paulpauper 14 days ago \| past \| 1 comment
		Anthropic Donations: Guesses and Uncertainties (lesswrong.com)
		2 points by joozio 14 days ago \| past
		Folie à Machine: LLMs and Epistemic Capture (lesswrong.com)
		2 points by joozio 14 days ago \| past
		Tracking (Expert/Influential) Predictions about AI (lesswrong.com)
		3 points by joozio 14 days ago \| past
		You can't imitation-learn how to continual-learn (lesswrong.com)
		11 points by supermdguy 15 days ago \| past
		The Terrarium (lesswrong.com)
		2 points by cubefox 15 days ago \| past
		A Tom-Inspired Agenda for AI Safety Research (lesswrong.com)
		2 points by joozio 19 days ago \| past \| 1 comment
		More