Tap a claim on the ring to put it at the centre.
← Starting from human knowledge is never forbidden, but in practice it…
17 connected korrents · 14 moments on record from 19 Aug 2020 to 3 Sept 2026.
Everything filed under startups
startups
Everything filed under reinforcement learning
reinforcement learning
Everything filed under venture capital
venture capital
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: Starting from human knowledge is never forbidden, but in practice it has always ended badly because people get locked into it. Starting from human knowledge is never forbidden, but in practice it has always ended badly because people get locked into it. Last stated 11 months ago
26 Sept 2025
RS
Richard Sutton — holds since 2025-09-26 — tap for who they are
Same subject: Concentrated power has been bad at every moment of human history, and a society is best off when power is spread as widely as possible. — tap to centre the map on it Concentrated power has been bad at every moment of human history, and a society is best off when power is spread as widely as possible. Last stated a month ago
28 Jul 2026
SA
Sam Altman — holds since 2026-07-28 — tap for who they are
Same subject: Digital worlds should carry fewer restrictions than the physical one, because acts that would do damage in reality do none there. — tap to centre the map on it Digital worlds should carry fewer restrictions than the physical one, because acts that would do damage in reality do none there. Last stated 3 years ago
28 Sept 2023
MZ
Mark Zuckerberg — holds since 2023-09-28 — tap for who they are
Same subject: Pre-training is not the process by which humans learn; it sits somewhere between human learning and human evolution. — tap to centre the map on it Pre-training is not the process by which humans learn; it sits somewhere between human learning and human evolution. Last stated 7 months ago
13 Feb 2026
DA
Dario Amodei — holds since 2026-02-13 — tap for who they are
Same subject: Locking humanity into one model's or one company's moral worldview would buy a little short-term safety and cost a long-term disaster. — tap to centre the map on it Locking humanity into one model's or one company's moral worldview would buy a little short-term safety and cost a long-term disaster. Last stated a month ago
28 Jul 2026
SA
Sam Altman — holds since 2026-07-28 — tap for who they are
Same subject: Deep learning generalises badly, and catastrophic interference with what a network already knew is the proof of it. — tap to centre the map on it Deep learning generalises badly, and catastrophic interference with what a network already knew is the proof of it. Last stated 11 months ago
26 Sept 2025
RS
Richard Sutton — holds since 2025-09-26 — 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: In a dysfunctional society the loss of working knowledge becomes routine, and attempts to reverse-engineer what was lost do not always succeed. — tap to centre the map on it In a dysfunctional society the loss of working knowledge becomes routine, and attempts to reverse-engineer what was lost do not always succeed. Last stated 6 years ago
19 Aug 2020
SB
Samo Burja — holds since 2020-08-19 — tap for who they are
Same subject: That people are good at maths and coding, which evolution never selected for, shows human learning is not a complicated evolutionary prior but something more fundamental. — tap to centre the map on it That people are good at maths and coding, which evolution never selected for, shows human learning is not a complicated evolutionary prior but something more fundamental. Last stated 9 months ago
25 Nov 2025
IS
Ilya Sutskever — holds since 2025-11-25 — tap for who they are
Same subject: A belief nobody else shares is a warning, and a belief everybody shares is also a warning; the right number of people who agree with you is small but not zero. — tap to centre the map on it A belief nobody else shares is a warning, and a belief everybody shares is also a warning; the right number of people who agree with you is small but not zero. Last stated a month ago
28 Jul 2026
SA
Sam Altman — holds since 2026-07-28 — tap for who they are
Same subject: A founder cannot delegate judgment: the chief executive has to make the decision that makes sense to them, however much momentum the alternative has. — tap to centre the map on it A founder cannot delegate judgment: the chief executive has to make the decision that makes sense to them, however much momentum the alternative has. Last stated a month ago
7 Aug 2026
MH
Max Hodak — holds since 2026-08-07 — tap for who they are
Same subject: A founder needs ambition rather than a sense of duty, because a startup is grueling enough that dutifulness will not carry anyone through it. — tap to centre the map on it A founder needs ambition rather than a sense of duty, because a startup is grueling enough that dutifulness will not carry anyone through it. Last stated 4 days ago
3 Sept 2026
PG
Paul Graham — holds since 2026-09-03 — tap for who they are
Same subject: A company that grows enough eventually has no need for investors. — tap to centre the map on it A company that grows enough eventually has no need for investors. Last stated 6 days ago
1 Sept 2026
PG
Paul Graham — holds since 2026-09-01 — tap for who they are
Same subject: About 5% is the most common final ownership stake for the founder of a VC-funded startup that works out. — tap to centre the map on it About 5% is the most common final ownership stake for the founder of a VC-funded startup that works out. Last stated 2 months ago
18 Jul 2026
PL
Pieter Levels — holds since 2026-07-18 — tap for who they are
Same subject: AI factories and foundries are too capital-intensive for venture money; they need government or sovereign funds behind them. — tap to centre the map on it AI factories and foundries are too capital-intensive for venture money; they need government or sovereign funds behind them. Last stated 3 months ago
18 Jun 2026
LT
Lip-Bu Tan — holds since 2026-06-18 — 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: 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: 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
At the centre
Starting from human knowledge is never forbidden, but in practice it has always ended badly because people get locked into it.
Holds RS Richard SuttonRead this korrent →
Concentrated power has been bad at every moment of human history, and a society is best off when power is spread as widely as possible.
Holds
Digital worlds should carry fewer restrictions than the physical one, because acts that would do damage in reality do none there.
Holds MZ Mark Zuckerberg
Pre-training is not the process by which humans learn; it sits somewhere between human learning and human evolution.
Holds DA Dario Amodei
Locking humanity into one model's or one company's moral worldview would buy a little short-term safety and cost a long-term disaster.
Holds
Deep learning generalises badly, and catastrophic interference with what a network already knew is the proof of it.
Holds RS Richard Sutton
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
In a dysfunctional society the loss of working knowledge becomes routine, and attempts to reverse-engineer what was lost do not always succeed.
Holds SB Samo Burja
That people are good at maths and coding, which evolution never selected for, shows human learning is not a complicated evolutionary prior but something more fundamental.
Holds IS Ilya Sutskever
A belief nobody else shares is a warning, and a belief everybody shares is also a warning; the right number of people who agree with you is small but not zero.
Holds
A founder cannot delegate judgment: the chief executive has to make the decision that makes sense to them, however much momentum the alternative has.
Holds MH Max Hodak
A founder needs ambition rather than a sense of duty, because a startup is grueling enough that dutifulness will not carry anyone through it.
Holds
A company that grows enough eventually has no need for investors.
Holds
About 5% is the most common final ownership stake for the founder of a VC-funded startup that works out.
Holds
AI factories and foundries are too capital-intensive for venture money; they need government or sovereign funds behind them.
Holds LT Lip-Bu Tan
A verifiable task can be optimised by reinforcement learning until a neural network performs it extremely well.
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
Humans barely use reinforcement learning for intelligence — what RL they do use goes into motor tasks, not problem solving.
Holds