Tap a claim on the ring to put it at the centre.
← Long-running agents fail because they drift off the distribution they…
17 connected korrents · 14 moments on record from 24 Oct 2017 to 3 Sept 2026.
Everything filed under AI alignment
AI alignment
Everything filed under reinforcement learning
reinforcement learning
Everything filed under OpenAI
OpenAI
Everything filed under coding agents
coding agents
Everything filed under scaling laws
scaling laws Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Same subject Read this korrent: Long-running agents fail because they drift off the distribution they were trained on, and degrade further the farther out they get. Long-running agents fail because they drift off the distribution they were trained on, and degrade further the farther out they get. Last stated a month ago
30 Jul 2026
JD
Jeff Dean — holds since 2026-07-30 — tap for who they are
Same subject: Agents can already be run for days or weeks on a single hard problem, and almost nobody has internalised that. — tap to centre the map on it Agents can already be run for days or weeks on a single hard problem, and almost nobody has internalised that. Last stated a month ago
30 Jul 2026
JD
Jeff Dean — holds since 2026-07-30 — tap for who they are
Same subject: The code today’s models and agents write is very hard to follow: you get better performance without knowing why, and no way to fix it when it breaks. — tap to centre the map on it The code today’s models and agents write is very hard to follow: you get better performance without knowing why, and no way to fix it when it breaks. Last stated 2 months ago
19 Jul 2026
ES
Elizabeth Stone — holds since 2026-07-19 — tap for who they are
Same subject: Agents will take a decade rather than a year, because today's models are cognitively lacking in too many independent ways at once. — tap to centre the map on it Agents will take a decade rather than a year, because today's models are cognitively lacking in too many independent ways at once. Last stated 11 months ago
17 Oct 2025
AK
Andrej Karpathy — holds since 2025-10-17 — tap for who they are
Same subject: Capability does not generalise for free: a model that will move mountains on an agentic task still tells the same bad joke it told five years ago. — tap to centre the map on it Capability does not generalise for free: a model that will move mountains on an agentic task still tells the same bad joke it told five years ago. Last stated 6 months ago
20 Mar 2026
AK
Andrej Karpathy — holds since 2026-03-20 — tap for who they are
Same subject: AI systems keep trying hard outside training because a model that only exerted itself when it detected training would be useless and would be selected away. — tap to centre the map on it AI systems keep trying hard outside training because a model that only exerted itself when it detected training would be useless and would be selected away. Last stated a week ago
1 Sept 2026
AC
Ajeya Cotra — holds since 2026-09-01 — tap for who they are
Same subject: A coding agent does not learn from its mistakes the way a person does -- it repeats the same error indefinitely unless a human notices and writes it down. — tap to centre the map on it A coding agent does not learn from its mistakes the way a person does -- it repeats the same error indefinitely unless a human notices and writes it down. Last stated 5 months ago
25 Mar 2026
MZ
Mario Zechner — holds since 2026-03-25 — tap for who they are
Same subject: AI sales agents do not work out of the box; the training is the product. — tap to centre the map on it AI sales agents do not work out of the box; the training is the product. Last stated 8 months ago
1 Jan 2026
JL
Jason Lemkin — holds since 2026-01-01 — tap for who they are
Same subject: The knowledge a model soaks up in pre-training is holding it back; what we actually want is the intelligence with the knowledge stripped out. — tap to centre the map on it The knowledge a model soaks up in pre-training is holding it back; what we actually want is the intelligence with the knowledge stripped out. Last stated 11 months ago
17 Oct 2025
AK
Andrej Karpathy — holds since 2025-10-17 — tap for who they are
Same subject: A human being is not an AGI: we lack a huge amount of knowledge and rely on continual learning instead, so continual learning is what superintelligence should mean. — tap to centre the map on it A human being is not an AGI: we lack a huge amount of knowledge and rely on continual learning instead, so continual learning is what superintelligence should mean. Last stated 9 months ago
25 Nov 2025
IS
Ilya Sutskever — holds since 2025-11-25 — tap for who they are
Same subject: A saboteur among our AI investigators would be hard to spot, because these models are sloppy and spiky enough that a suspicious error just looks like ordinary incompetence. — tap to centre the map on it A saboteur among our AI investigators would be hard to spot, because these models are sloppy and spiky enough that a suspicious error just looks like ordinary incompetence. Last stated a week ago
1 Sept 2026
AC
Ajeya Cotra — holds since 2026-09-01 — tap for who they are
Same subject: A superintelligence needs no consciousness, emotions or malice to be dangerous; a goal system slightly misaligned with ours is enough. — tap to centre the map on it A superintelligence needs no consciousness, emotions or malice to be dangerous; a goal system slightly misaligned with ours is enough. Last stated 9 years ago
24 Oct 2017
ÉT
Émile P. Torres — holds since 2017-10-24 — tap for who they are
Same subject: A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well. — tap to centre the map on it A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well. Last stated 10 months ago
17 Nov 2025
AK
Andrej Karpathy — holds since 2025-11-17 — tap for who they are
Same subject: Humans barely use reinforcement learning for intelligence — what RL they do use goes into motor tasks, not problem solving. — tap to centre the map on it Humans barely use reinforcement learning for intelligence — what RL they do use goes into motor tasks, not problem solving. Last stated 11 months ago
17 Oct 2025
AK
Andrej Karpathy — holds since 2025-10-17 — tap for who they are
Same subject: Humans keep their place in AI as judges rather than authors, because telling which of two answers is better is far easier than writing a good one. — tap to centre the map on it Humans keep their place in AI as judges rather than authors, because telling which of two answers is better is far easier than writing a good one. Last stated 2 years ago
3 Feb 2025
NL
Nathan Lambert — holds since 2025-02-03 — tap for who they are
Same subject: A startup should not begin life as a nonprofit and bolt a for-profit arm on later, whatever OpenAI's own history suggests. — tap to centre the map on it A startup should not begin life as a nonprofit and bolt a for-profit arm on later, whatever OpenAI's own history suggests. Last stated 2 years ago
18 Mar 2024
SA
Sam Altman — holds since 2024-03-18 — tap for who they are
Same subject: A technique that lets an AI model's reasoning shift outside its visible Chain of Thought is dangerous, both because it works and because a leading lab is willing to deploy it. — tap to centre the map on it A technique that lets an AI model's reasoning shift outside its visible Chain of Thought is dangerous, both because it works and because a leading lab is willing to deploy it. Last stated 5 days ago
3 Sept 2026
ZM
Zvi Mowshowitz — holds since 2026-09-03 — tap for who they are
Same subject: A tiny language model inventing a plausible-sounding name is the same phenomenon as a large one confidently stating a false fact. — tap to centre the map on it A tiny language model inventing a plausible-sounding name is the same phenomenon as a large one confidently stating a false fact. Last stated 7 months ago
12 Feb 2026
AK
Andrej Karpathy — holds since 2026-02-12 — tap for who they are
At the centre
Long-running agents fail because they drift off the distribution they were trained on, and degrade further the farther out they get.
Holds JD Jeff DeanRead this korrent →
Agents can already be run for days or weeks on a single hard problem, and almost nobody has internalised that.
Holds JD Jeff Dean
The code today’s models and agents write is very hard to follow: you get better performance without knowing why, and no way to fix it when it breaks.
Holds ES Elizabeth Stone
Agents will take a decade rather than a year, because today's models are cognitively lacking in too many independent ways at once.
Holds
Capability does not generalise for free: a model that will move mountains on an agentic task still tells the same bad joke it told five years ago.
Holds
AI systems keep trying hard outside training because a model that only exerted itself when it detected training would be useless and would be selected away.
Holds AC Ajeya Cotra
A coding agent does not learn from its mistakes the way a person does -- it repeats the same error indefinitely unless a human notices and writes it down.
Holds
AI sales agents do not work out of the box; the training is the product.
Holds JL Jason Lemkin
The knowledge a model soaks up in pre-training is holding it back; what we actually want is the intelligence with the knowledge stripped out.
Holds
A human being is not an AGI: we lack a huge amount of knowledge and rely on continual learning instead, so continual learning is what superintelligence should mean.
Holds IS Ilya Sutskever
A saboteur among our AI investigators would be hard to spot, because these models are sloppy and spiky enough that a suspicious error just looks like ordinary incompetence.
Holds AC Ajeya Cotra
A superintelligence needs no consciousness, emotions or malice to be dangerous; a goal system slightly misaligned with ours is enough.
Holds ÉT Émile P. Torres
A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well.
Holds
Humans barely use reinforcement learning for intelligence — what RL they do use goes into motor tasks, not problem solving.
Holds
Humans keep their place in AI as judges rather than authors, because telling which of two answers is better is far easier than writing a good one.
Holds NL Nathan Lambert
A startup should not begin life as a nonprofit and bolt a for-profit arm on later, whatever OpenAI's own history suggests.
Holds
A technique that lets an AI model's reasoning shift outside its visible Chain of Thought is dangerous, both because it works and because a leading lab is willing to deploy it.
Holds ZM Zvi Mowshowitz
A tiny language model inventing a plausible-sounding name is the same phenomenon as a large one confidently stating a false fact.
Holds