Gwern.net — AI: Safety, Alignment & Philosophy

The Meta-LW Doomsday Argument, by Gwern · Gwern.net

Gwern — Tue, 29 Apr 2025 00:00:00 +0000

N/A

Area Man Angry AI Not Silver Bullet, by Gwern, Gemini-2.5-pro, GPT-4-o3 · Gwern.net

Gwern, Gemini-2.5-pro, GPT-4-o3 — Sat, 19 Apr 2025 00:00:00 +0000

Onion-style fiction satire of complaints about automation or scaling not proceeding fast enough for some individual with unstated comparisons or timelines. (Partially an LLM style-imitation writing exercise.)

Copyright & AI · Gwern.net

Gwern — Sun, 26 Nov 2023 00:00:00 +0000

The moral application of copyright and intellectual property to AI is the minimum protection necessary to promote the progress of science and the useful arts.

Is OpenAI OK?, by Gwern · Gwern.net

Gwern — Wed, 19 Jun 2024 00:00:00 +0000

Has OA lost its mojo and ability to push AI paradigms in the wake of its hypergrowth and the OA Altman coup? Is OA OK? How would we know & what would it look like if it was not?

‘Winning’ AI Arms Races: Then What?, by Gwern · Gwern.net

Gwern — Sat, 23 Nov 2024 00:00:00 +0000

Challenge to proponents of AI arms races and Manhattan Projects: state plainly, what is your long-term exit plan? What does ‘success’ look like? Or what good is it to beat rivals to AGI and then stand by while they catch up?

The Existential Risk of Math Errors · Gwern.net

Gwern — Fri, 20 Jul 2012 00:00:00 +0000

Mathematical mistake/error-rates limit our understanding of rare risks and ability to defend against them.

What Is an ‘AI Warning Shot’?, by Gwern · Gwern.net

Gwern — Wed, 07 Aug 2024 00:00:00 +0000

N/A

It Looks Like You’re Trying To Take Over The World · Gwern.net

Gwern — Sun, 06 Mar 2022 00:00:00 +0000

Fictional short story about Clippy & AI hard takeoff scenarios grounded in contemporary ML scaling, self-supervised learning, reinforcement learning, and meta-learning research literature.

Evolution as Backstop for Reinforcement Learning · Gwern.net

Gwern — Thu, 06 Dec 2018 00:00:00 +0000

Markets/evolution as backstops/ground truths for reinforcement learning/optimization: on some connections between Coase’s theory of the firm/linear optimization/DRL/evolution/multicellular life/pain/Internet communities as multi-level optimization problems.

Complexity no Bar to AI · Gwern.net

Gwern — Sun, 01 Jun 2014 00:00:00 +0000

Critics of AI risk suggest diminishing returns to computing (formalized asymptotically) means AI will be weak; this argument relies on a large number of questionable premises and ignoring additional resources, constant factors, and nonlinear returns to small intelligence advantages, and is highly unlikely.

The Hyperbolic Time Chamber & Brain Emulation · Gwern.net

Gwern — Wed, 29 Aug 2012 00:00:00 +0000

A time dilation chamber as thought experiment on the power of pure thought, with comparison to computer AGI advantages/disadvantages.

Why Tool AIs Want to Be Agent AIs · Gwern.net

Gwern — Wed, 07 Sep 2016 00:00:00 +0000

AIs limited to pure computation (Tool AIs) supporting humans, will be less intelligent, efficient, and economically valuable than more autonomous reinforcement-learning AIs (Agent AIs) who act on their own and meta-learn, because all problems are reinforcement-learning problems.

Simulation Inferences · Gwern.net

Gwern — Fri, 29 May 2009 00:00:00 +0000

How small must be the computer simulating the universe?