AIFP
Article
AIFP is a recurring organization in the Astral Codex Ten archive, appearing 2 times across 2 issues between December 10, 2025 and February 05, 2026. The archive places it in contexts such as “I sometimes do work for AIFP, but I wasn’t involved in this particular effort”; “AIFP’s going longer again. AIFP has also responded to titotal’s critique”. It most often appears alongside A16Z, Anthropic, Hitler.
Metadata
- Category: Organizations
- Mention count: 2
- Issue count: 2
- First seen: December 10, 2025
- Last seen: February 05, 2026
Appears In
Related Pages
-
- A16Z (2 shared issues)
-
- Anthropic (2 shared issues)
-
- Hitler (2 shared issues)
-
- neoreaction (2 shared issues)
-
- Oliver (2 shared issues)
-
- OpenAI (2 shared issues)
-
- Richard Hanania (2 shared issues)
-
- Trump (2 shared issues)
-
- Trump administration (2 shared issues)
-
- Twitter (2 shared issues)
-
- US (2 shared issues)
-
- 100 Above The Park (1 shared issues)
External Links
Source Context
Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.
I sometimes do work for AIFP, but I wasn’t involved in this particular effort. Still, I agree with everything they say - except point 7, “AIs must not make important decisions or control critical systems”. Every time you take a Waymo, you’re letting an AI control a critical system; every time it chooses to stop at a red light but not a green one, it’s making an “important decision” (if you don’t think this decision is important, consider the consequences of failure). This isn’t a gotcha: it’s fine for near-term AI systems to make important decisions in cases where they’ve been well-tested and there’s good reason to think that they outperform humans on net. Getting rid of the last 0.001% of hallucinations and inexplicable behavior would be nice, but shouldn’t delay rollout if there are compensatory advantages. [EDIT: See author response, they don’t disagree]
Inline links: See author response, they don’t disagree
I don’t think this is quite right - I think they’re actually following their math and so when they redid the math and got different results they said so - but I agree it’s ironic that when everyone else had long timelines, AIFP went short, and now that everyone else is starting to come around, AIFP’s going longer again. AIFP has also responded to titotal’s critique of their timeline model here.