Astral Codex Ten

Nora Belrose

3 min read

Nora Belrose

Article

Nora Belrose is a recurring person in the Astral Codex Ten archive, appearing 2 times across 2 issues between October 05, 2023 and January 18, 2024. The archive places it in contexts such as “Nora Belrose made this argument in AI Pause Will Likely Backfire”; “optimists.ai (led by Nora Belrose …)“. It most often appears alongside DeepMind, FDA, OpenAI.

Metadata

  • Category: People
  • Mention count: 2
  • Issue count: 2
  • First seen: October 05, 2023
  • Last seen: January 18, 2024

Appears In

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.

Pause For Thought: The AI Pause Debate
October 05, 2023 · Original source
HARDWARE PROGRESS. Also the laptops keep getting better, because of Moore’s Law. Regulators can plausibly control the flow of supercomputers, at least domestically. But eventually technology will advance to the point where you can train an AI on anything. Then you either have to ban all computing, restrict it at gradually more extreme levels (1990 MS-DOS machines! No, punch cards!) or accept that AI is going to happen. Still, you can imagine this buying us a few decades. Rob Bensinger defended this view in Comments On Manheim’s “What’s In A Pause?”, and it’s the backdrop to Holly Elmore’s Case For AI Advocacy To The Public2. No Pause: Or we could not do any of that. If we think alignment research is going well, and that a pause would mess it up, or cause a compute overhang leading to un-research-able fast takeoff, or cede the lead to China, maybe we should stick with the current rate of progress. Nora Belrose made this argument in AI Pause Will Likely Backfire. Specifically: [A pause] would have several predictable negative effects: Illegal AI labs develop inside pause countries, remotely using training hardware outsourced to non-pause countries to evade detection. Illegal labs would presumably put much less emphasis on safety than legal ones.
Inline links: Comments On Manheim’s “What’s In A Pause?”, Case For AI Advocacy To The Public, 2, AI Pause Will Likely Backfire
Links For January 2024
January 18, 2024 · Original source
13: Related: optimists.ai (led by Nora Belrose and Quintin Pope, previously discussed here) is like e/acc, except they’ve thought about it a lot and sometimes make good arguments. I endorse them (as good people to read; I’m still not sure to what degree I agree with them). If you want to do some kind of both sides debate thing, these would be the people I would contact first.
Inline links: optimists.ai, here