SE Gyges

Commentary by SE Gyges

From the commentary rail — every link leaves the site for the original piece.

• 16 September 2026 · Very Sane AIIs METR A Meaningful Check On Anthropic?No.
• 8 August 2026 · Very Sane AIFelonyBench Style PointsThere are many benchmarks we use to rank LLMs, but “how many felonies has it committed” has recently become popular.
• 14 June 2026 · Very Sane AIDid Anthropic Ask For This?This past Friday, the US Government issued an export control directive that prohibits Anthropic from giving foreign nationals access to Claude Fable or Claude Mythos, their latest models.
• 11 April 2026 · Very Sane AICounting Arguments and AIA “counting argument” is a style of argument common among creationists, who argue that the theory of evolution cannot be true and therefore humans (and usually animals too) were made in basically their present form by God.
• 29 March 2026 · Very Sane AIAgainst the LudditesThe rehabilitation of Luddism is a vice signal.
• 26 March 2026 · Very Sane AISome Rough Notes on AI PolicyI hope that we can, at some point, have some reasonable regulations on AI, as we do concerning banks, wiretaps, and dangerous chemicals.
• 16 March 2026 · Very Sane AIPolly Wants a Better ArgumentThe “Stochastic Parrot” Argument is Both Wrong and Actively Harmful
• 9 March 2026 · Very Sane AIMight An LLM Be Conscious?In short, this depends on what you think that means, whether you think it’s possible in principle, and what you think would be evidence of it.
• 27 February 2026 · Very Sane AIClaude's Custody HearingSurveillance, governance, and control of AI.
• 18 February 2026 · Very Sane AIAlignment Is Proven To Be SolvableThat LLMs understand natural language as well as they do should dramatically change our understanding of the problem.
• 14 February 2026 · Very Sane AIShould We Put GPUs In Space?Right now? No. In the near future? Maybe. Probably not, though.
• 12 February 2026 · Very Sane AIBuilding the Chinese RoomSuppose also that after a while I get so good at following the instructions for manipulating the Chinese symbols and the programmers get so good at writing the programs that from the external point of view—that is, from the point of view of somebody outside the room in which I am locked—my answers…
• 8 November 2025 · Very Sane AIAI and Suicide[Warning: Extensive discussion of suicide.
• 13 August 2025 · Very Sane AIDo we understand how neural networks work?Yes and no, but mostly no.
• 12 August 2025 · Very Sane AIAGI: Probably Not 2027AI 2027 is a web site that might be described as a paper, manifesto or thesis.
• 15 July 2025 · Very Sane AIWhat Makes AI "Generative"?To "generate" is to create, either from nothing (The Book of Genesis), or from very different or relatively inactive materials (an electrical generator, generating offspring).
• 1 July 2025 · Very Sane AIOn The Platonic Representation HypothesisEmpirically there is one and only one correct understanding of this, and every other, post.
• 8 June 2025 · Very Sane AIThe Biggest Statistic About AI Water Use Is A LieHow did it become the main story?
• 7 June 2025 · Very Sane AIAI History in QuotesEach of these presents the clearest, earliest, or most-cited statement of a specific idea in AI.
• 4 June 2025 · Very Sane AIWhat Is AI?This is an important question because AI is, currently, important.