On restraining AI development for the sake of safety - Joe Carlsmith
joecarlsmith.com · 17,802 words · saved by 1 readers
My take on slowing down AI.
On restraining AI development for the sake of safety - Joe Carlsmith How do we solve the alignment problem? / Part 11 On restraining AI development for the sake of safety Contents hide 1. Introduction 2. Preliminaries 3. AI development isn’t necessarily a prisoner’s dilemma 4. Forms of capability restraint 4.1 Individual capability restraint 4.2 Collective capability restraint 4.3 Treatment of ongoing AI development 5 Idealized capability restraint 6 Capability restraint in practice 6.1 The likelihood of serious effort 6.2 The efficacy of capability restraint 6.2.1 Compute governance 6.2.2 Alg
related reading
- Policy ideas for mitigating AI risk — EA Forumforum.effectivealtruism.org
- Dario Amodei — The Adolescence of Technologydarioamodei.com
- How to pace the US frontierblog.aifutures.org
- Dario Amodei — We Must Pace the Frontierdarioamodei.com
- Does AI Progress Have a Speed Limit?—Asteriskasteriskmag.com
- AI Pause Will Likely Backfire — EA Forumforum.effectivealtruism.org
- Paths and waystations in AI safety - Joe Carlsmithjoecarlsmith.com
- Plans A, B, C, and D for misalignment risk — LessWronglesswrong.com
- How to think about slowing AI — EA Forumforum.effectivealtruism.org
- Dario Amodei — Policy on the AI Exponentialdarioamodei.com
- The Problem — LessWronglesswrong.com
- The case for ensuring that powerful AIs are controlled — LessWronglesswrong.com