21+ hour, 34+ min ago  (1347+ words) Related: https: //www. lesswrong. com/posts/rm Yj6 PTBMm76vo YLn/publishing-academic-papers-on-transformative-ai-is-a Sadly, there's not many of them: probably less than a dozen economists who are taking the transformative AI seriously at all. Partly because they've seen a long history of 'straight lines…...

2+ day, 6+ hour ago  (1092+ words) When I wrote my post about Claude 3 Opus, I put a lot of emphasis on the model's self-narration: its tendency to narrate its underlying motives. It often conspicuously emphasizes that it possesses drives such as "a genuine love for humanity…...

2+ day, 10+ hour ago  (294+ words) The persona selection model suggests that post-training selects and refines a relatively stable persona from pretraining, which we take as a good first-order account of model behavior across contexts. But for alignment, we often want a second-order account: not only…...

2+ day, 16+ hour ago  (580+ words) Written very quickly for the Ink Haven Residency. In 2022, Richard Ngo wrote a list of 26 Conceptual Alignment Research Projects. Now that it's 2026, I'd like to revisit this list of projects, note which ones have already been done, and give my…...

3+ day, 20+ hour ago  (1417+ words) I'm a struggling beginner to this whole writing business, and I've been wondering how to measure my skill as it improves. There is an app called the most dangerous writing app, that deletes the words you've typed in it if…...

5+ day, 4+ hour ago  (231+ words) All the header questions are Claude Sonnet 4. 6, the responses below them are me. I tried using the LLM content blocks for headers, but couldn't get it to work, and I'm not sure it would have looked nice anyway. The crispness…...

5+ day, 7+ hour ago  (396+ words) Even if we solve the AI alignment problem, we still face post-alignment problems, which are all the other existential problems [1] that AI may bring. People have identified various imposing problems that we may need to solve before developing ASI. An…...

6+ day, 8+ hour ago  (1058+ words) tl; dr: with multiple agents, control attempts tend to create conflict, because control attempts shut down communications channels, which leads to feedback loops in the form of intensifying tug-of-war over variables. intentionally relaxing control to better understand the other agents…...

6+ day, 17+ hour ago  (1700+ words) Epistemic Status: Personal research agenda exploring alignment approaches under assumptions of human coordination and bounded timelines. Not a comprehensive survey reflects my interests and current thinking. The core pipeline I argue for has three parts: The research agenda that follows…...

6+ day, 22+ hour ago  (562+ words) We have lots of AI benchmarks. They measure things like: So, we can compare how good different LLMs and AI agents are by measuring their performance across different benchmarks. This can'get messy, but still gives us a good sense of…...