Tap a claim on the ring to put it at the centre.
← Reinforcement learning from verifiable rewards does not reliably carry…
17 connected korrents · 18 moments from 7 Sept 2008 to 24 Oct 2026. Nearly all of them are about scaling laws .
Everything filed under reinforcement learning
reinforcement learning
Everything filed under AI alignment
AI alignment
Everything filed under benchmarks
benchmarks 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: Reinforcement learning from verifiable rewards does not reliably carry a model's pretrained ethical understanding into its actual behavior. Reinforcement learning from verifiable rewards does not reliably carry a model's pretrained ethical understanding into its actual behavior. Last stated 3 months ago
11 Aug 2026
SH
Samuel Hammond — holds since 2026-08-11 — tap for who they are
Same subject: Models resemble each other because pre-training is the same everywhere; what differentiates labs now is RL and post-training. — tap to centre the map on it Models resemble each other because pre-training is the same everywhere; what differentiates labs now is RL and post-training. Last stated 11 months ago
25 Nov 2025
IS
Ilya Sutskever — holds since 2025-11-25 — tap for who they are
Same subject: One pre-trained robot model now matches or beats the specialists that were fine-tuned with reinforcement learning for the very tasks they were built for. — tap to centre the map on it One pre-trained robot model now matches or beats the specialists that were fine-tuned with reinforcement learning for the very tasks they were built for. Last stated 3 months ago
12 Aug 2026
CF
Chelsea Finn — holds since 2026-08-12 — tap for who they are
Same subject: Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware. — tap to centre the map on it Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware. Last stated a year ago
17 Oct 2025
AK
Andrej Karpathy — holds since 2025-10-17 — tap for who they are
Same subject: Reasoning models are an upgrade from conventional LLMs achieved by adding post-training via reinforcement learning and optionally inference scaling. — tap to centre the map on it Reasoning models are an upgrade from conventional LLMs achieved by adding post-training via reinforcement learning and optionally inference scaling. Last stated 2 weeks ago
24 Oct 2026
SR
Sebastian Raschka — holds since 2026-10-24 — tap for who they are
Same subject: RL environments exist to make a model generalise, not to teach it each skill one at a time — exactly as pre-training does. — tap to centre the map on it RL environments exist to make a model generalise, not to teach it each skill one at a time — exactly as pre-training does. Last stated 9 months ago
13 Feb 2026
DA
Dario Amodei — holds since 2026-02-13 — tap for who they are
Same subject: A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number. — tap to centre the map on it A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number. Last stated 4 months ago
26 Jun 2026
NB
Noam Brown — holds since 2026-06-26 — tap for who they are
Same subject: A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from. — tap to centre the map on it A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from. Last stated 8 months ago
13 Mar 2026
DP
Dylan Patel — holds since 2026-03-13 — 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 12 months ago
17 Nov 2025
AK
Andrej Karpathy — holds since 2025-11-17 — tap for who they are
Same subject: "Hire smart people and get out of their way" is bad management advice; leaders should stay closely involved, building trust and working through hard problems with their people to create alignment. — tap to centre the map on it "Hire smart people and get out of their way" is bad management advice; leaders should stay closely involved, building trust and working through hard problems with their people to create alignment. Last stated 4 weeks ago
9 Oct 2026
CR
Carl Rivera — holds since 2026-10-09 — tap for who they are
Same subject: A company that uses an algorithm to nudge customers into behavior that isn't good for them is doing something bad. — tap to centre the map on it A company that uses an algorithm to nudge customers into behavior that isn't good for them is doing something bad. Last stated a year ago
7 Oct 2025
MM
Mr. Money Mustache — holds since 2025-10-07 — tap for who they are
Same subject: A control evaluation that reports under one per cent risk should be read as several per cent, because the evaluation can itself fail. — tap to centre the map on it A control evaluation that reports under one per cent risk should be read as several per cent, because the evaluation can itself fail. Last stated 2 years ago
7 May 2024
BS
Buck Shlegeris — holds since 2024-05-07 — tap for who they are
Same subject: An open weights model from 2025 with a pentest harness could already escape a sandbox and hack most networks; the surprise says more about the sandbox than the model. — tap to centre the map on it An open weights model from 2025 with a pentest harness could already escape a sandbox and hack most networks; the surprise says more about the sandbox than the model. Last stated 3 months ago
22 Jul 2026
TP
Thomas Ptacek — holds since 2026-07-22 — tap for who they are
Same subject: Because a few specialists already devote themselves to superintelligent AI, the rest of us have correspondingly less reason to spend our own effort on it. — tap to centre the map on it Because a few specialists already devote themselves to superintelligent AI, the rest of us have correspondingly less reason to spend our own effort on it. Last stated 3 years ago
13 Dec 2023
SA
Scott Aaronson — holds since 2008-09-07 — tap for who they are
SA
Scott Aaronson — no longer holds since 2023-12-13 — tap for who they are
Same subject: Disclosure of rogue AI agent incidents should be mandatory for AI labs, not left to their own discretion. — tap to centre the map on it Disclosure of rogue AI agent incidents should be mandatory for AI labs, not left to their own discretion. Last stated 2 months ago
6 Sept 2026
ZM
Zvi Mowshowitz — holds since 2026-09-06 — tap for who they are
Same subject: An AI should be built as a fiduciary, the equivalent of a lawyer for its user, rather than as an agent pursuing its own notion of the good. — tap to centre the map on it An AI should be built as a fiduciary, the equivalent of a lawyer for its user, rather than as an agent pursuing its own notion of the good. Last stated 3 months ago
11 Aug 2026
RG
Ryan Greenblatt — holds since 2026-08-11 — tap for who they are
Same subject: Anthropic is simultaneously pursuing pro-safety actions and highly accelerationist actions. — tap to centre the map on it Anthropic is simultaneously pursuing pro-safety actions and highly accelerationist actions. Last stated 2 weeks ago
20 Oct 2026
ZM
Zvi Mowshowitz — holds since 2026-10-20 — tap for who they are
Same subject: Anthropic now uses mechanistic interpretability techniques to audit new models for deception and scheming before releasing them. — tap to centre the map on it Anthropic now uses mechanistic interpretability techniques to audit new models for deception and scheming before releasing them. Last stated 10 months ago
1 Jan 2026
DA
Dario Amodei — holds since 2026-01-01 — tap for who they are
At the centre
Reinforcement learning from verifiable rewards does not reliably carry a model's pretrained ethical understanding into its actual behavior.
Holds Read this korrent →
Models resemble each other because pre-training is the same everywhere; what differentiates labs now is RL and post-training.
Holds
One pre-trained robot model now matches or beats the specialists that were fine-tuned with reinforcement learning for the very tasks they were built for.
Holds
Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware.
Holds
Reasoning models are an upgrade from conventional LLMs achieved by adding post-training via reinforcement learning and optionally inference scaling.
Holds
RL environments exist to make a model generalise, not to teach it each skill one at a time — exactly as pre-training does.
Holds
A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number.
Holds
A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from.
Holds
A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well.
Holds
"Hire smart people and get out of their way" is bad management advice; leaders should stay closely involved, building trust and working through hard problems with their people to create alignment.
Holds
A company that uses an algorithm to nudge customers into behavior that isn't good for them is doing something bad.
Holds
A control evaluation that reports under one per cent risk should be read as several per cent, because the evaluation can itself fail.
Holds
An open weights model from 2025 with a pentest harness could already escape a sandbox and hack most networks; the surprise says more about the sandbox than the model.
Holds
Because a few specialists already devote themselves to superintelligent AI, the rest of us have correspondingly less reason to spend our own effort on it.
No longer holds
Disclosure of rogue AI agent incidents should be mandatory for AI labs, not left to their own discretion.
Holds
An AI should be built as a fiduciary, the equivalent of a lawyer for its user, rather than as an agent pursuing its own notion of the good.
Holds
Anthropic is simultaneously pursuing pro-safety actions and highly accelerationist actions.
Holds
Anthropic now uses mechanistic interpretability techniques to audit new models for deception and scheming before releasing them.
Holds