“AI for AI safety” by Joe_Carlsmith

This is a link post. (Audio version here (read by the author), or search for "Joe Carlsmith Audio" on your podcast app. This is the fourth essay in a series that I’m calling “How do we solve the alignment problem?”. I’m hoping that the individual essays can be read fairly well on their own, but see this introduction for a summary of the essays that have been released thus far, and for a bit more about the series as a whole.) 1. Introduction and summary In my last essay, I offered a high-level framework for thinking about the path from here to safe superintelligence. This framework emphasized the role of three key “security factors” – namely: Safety progress: our ability to develop new levels of AI capability safely, Risk evaluation: our ability to track and forecast the level of risk that a given sort of AI capability development [...] ---Outline:(00:29) 1. Introduction and summary(04:12) 2. What is AI for AI safety?(12:12) 2.1 A tale of two feedback loops(14:47) 2.2 Contrast with need human-labor-driven radical alignment progress views(17:24) 2.3 Contrast with a few other ideas in the literature(19:51) 3. Why is AI for AI safety so important?(23:14) 4. The AI for AI safety sweet spot(27:44) 4.1 The AI for AI safety spicy zone(29:51) 4.2 Can we benefit from a sweet spot?(31:59) 5. Objections to AI for AI safety(32:17) 5.1 Three core objections to AI for AI safety(34:03) 5.2 Other practical concernsThe original text contained 78 footnotes which were omitted from this narration. The original text contained 14 images which were described by AI. --- First published: March 14th, 2025 Source: https://forum.effectivealtruism.org/posts/pS7NDzfgydmyjEF3g/ai-for-ai-safety --- Narrated by TYPE III AUDIO. ---Images from the article:Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.