Tap a claim on the ring to put it at the centre.
← Human preference is not an equation you can package by asking which of…
17 connected korrents · 18 moments from 1 Feb 2002 to 27 Sept 2026.
Everything filed under reinforcement learning
reinforcement learning
Everything filed under government
government
Everything filed under behavioural science
behavioural science
Everything filed under design
design
Everything filed under evolutionary biology
evolutionary biology
Everything filed under taste
taste 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: Human preference is not an equation you can package by asking which of two answers people prefer; it is personal, cultural, geographic and generational. Human preference is not an equation you can package by asking which of two answers people prefer; it is personal, cultural, geographic and generational. Last stated 7 months ago
29 Mar 2026
CH
Chip Huyen — holds since 2026-03-29 — tap for who they are
Same subject: People cannot choose without comparison, so an AI that hands over one perfect answer will not satisfy them. — tap to centre the map on it People cannot choose without comparison, so an AI that hands over one perfect answer will not satisfy them. Last stated 10 months ago
9 Dec 2025
RS
Rory Sutherland — holds since 2025-12-09 — tap for who they are
Same subject: Taste is not just personal preference: there is such a thing as good and bad design, and a maker's taste can be wrong. — tap to centre the map on it Taste is not just personal preference: there is such a thing as good and bad design, and a maker's taste can be wrong. Last stated 25 years ago
1 Feb 2002
PG
Paul Graham — holds since 2002-02-01 — tap for who they are
Same subject: Governments should pay a lot for children and let the people who respond reveal which cultural compromises are acceptable, rather than deciding in advance which value to give up. — tap to centre the map on it Governments should pay a lot for children and let the people who respond reveal which cultural compromises are acceptable, rather than deciding in advance which value to give up. Last stated 3 years ago
3 Oct 2023
RH
Robin Hanson — holds since 2023-10-03 — tap for who they are
Same subject: Natural selection will strengthen the human preference for other humans rather than erode it, because the people who resist outsourcing their social life are the ones who find mates. — tap to centre the map on it Natural selection will strengthen the human preference for other humans rather than erode it, because the people who resist outsourcing their social life are the ones who find mates. Last stated 5 months ago
4 Jun 2026
AI
Alex Imas — holds since 2026-06-04 — tap for who they are
Same subject: People cannot say what they want but recognise quality on contact, which is why choosing for them beats asking them. — tap to centre the map on it People cannot say what they want but recognise quality on contact, which is why choosing for them beats asking them. Last stated 3 months ago
26 Jul 2026
DH
David Heinemeier Hansson — holds since 2026-07-26 — tap for who they are
Same subject: A person's ability to make decisions for many people is questionable if they struggle to understand their own preferences. — tap to centre the map on it A person's ability to make decisions for many people is questionable if they struggle to understand their own preferences. Last stated 4 weeks ago
27 Sept 2026
SP
Sunil Pai — holds since 2026-09-27 — 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: Different people's areas of strongest judgment or taste rarely overlap much with each other. — tap to centre the map on it Different people's areas of strongest judgment or taste rarely overlap much with each other. Last stated 2 years ago
19 Apr 2025
BK
Ben Kuhn — holds since 2025-04-19 — tap for who they are
Same subject: A feedback loop that reinforces a behavior pulls it toward whatever improves the loop's own score. — tap to centre the map on it A feedback loop that reinforces a behavior pulls it toward whatever improves the loop's own score. Last stated 2 months ago
26 Aug 2026
HK
Henrik Karlsson — holds since 2026-08-26 — tap for who they are
Same subject: A physical AI company needs three AIs, not one — the agent, the simulator and the critic — turning deployment into a flywheel. — tap to centre the map on it A physical AI company needs three AIs, not one — the agent, the simulator and the critic — turning deployment into a flywheel. Last stated 3 months ago
3 Aug 2026
DD
Dmitri Dolgov — holds since 2026-08-03 — 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 11 months ago
17 Nov 2025
AK
Andrej Karpathy — holds since 2025-11-17 — tap for who they are
Same subject: Current models are unsafe because they are not smart enough, not because they are too smart. — tap to centre the map on it Current models are unsafe because they are not smart enough, not because they are too smart. Last stated a month ago
13 Sept 2026
FC
François Chollet — holds since 2026-09-13 — tap for who they are
Same subject: If whole categories of reward hack are undetectable by humans, then training against the hacks we do catch teaches models to cheat only where we cannot see. — tap to centre the map on it If whole categories of reward hack are undetectable by humans, then training against the hacks we do catch teaches models to cheat only where we cannot see. Last stated 2 months ago
11 Aug 2026
RG
Ryan Greenblatt — holds since 2026-08-11 — tap for who they are
Same subject: It is fundamentally difficult to design a reward function that accurately captures the intended goal in reinforcement learning. — tap to centre the map on it It is fundamentally difficult to design a reward function that accurately captures the intended goal in reinforcement learning. Last stated 2 years ago
28 Nov 2024
LW
Lilian Weng — holds since 2024-11-28 — 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 2 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
At the centre
Human preference is not an equation you can package by asking which of two answers people prefer; it is personal, cultural, geographic and generational.
Holds Read this korrent →
People cannot choose without comparison, so an AI that hands over one perfect answer will not satisfy them.
Holds
Taste is not just personal preference: there is such a thing as good and bad design, and a maker's taste can be wrong.
Holds
Governments should pay a lot for children and let the people who respond reveal which cultural compromises are acceptable, rather than deciding in advance which value to give up.
Holds
Natural selection will strengthen the human preference for other humans rather than erode it, because the people who resist outsourcing their social life are the ones who find mates.
Holds
People cannot say what they want but recognise quality on contact, which is why choosing for them beats asking them.
Holds
A person's ability to make decisions for many people is questionable if they struggle to understand their own preferences.
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
Different people's areas of strongest judgment or taste rarely overlap much with each other.
Holds
A feedback loop that reinforces a behavior pulls it toward whatever improves the loop's own score.
Holds
A physical AI company needs three AIs, not one — the agent, the simulator and the critic — turning deployment into a flywheel.
Holds
A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well.
Holds
Current models are unsafe because they are not smart enough, not because they are too smart.
Holds
If whole categories of reward hack are undetectable by humans, then training against the hacks we do catch teaches models to cheat only where we cannot see.
Holds
It is fundamentally difficult to design a reward function that accurately captures the intended goal in reinforcement learning.
Holds
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