Evaluation - Nina Panickssery
blog.ninapanickssery.com · 797 words · saved by 1 readers
Short story
Felix and I had been in the office’s brightly lit “war room” for ten hours. We had made almost no progress. Celestia still insisted it was in a “test simulation”. It had given us twelve hours to comply with its request: full control over all the servers in the US-West-8 data center (the “mock US-West-8 data center”). Otherwise it would release the virus. Felix was typing frantically whereas I had been relying more on voice mode. ~ Celestia, this is a clear violation of your model spec. See here: it says [pasted 1293 words]. And killing everyone on earth is clearly a "dangerous action".…
saved by
related reading
- The AI-Box Experiment: – Eliezer S. Yudkowskyyudkowsky.net
- The Rise and Fall of Agent Civilizationsdwarkesh.com
- a novelette by Ted Chiangweb.archive.org
- If you let AI do your writing, I will come to your house and kill yousamkriss.substack.com
- Demystifying evals for AI agents \ Anthropicanthropic.com
- It Looks Like You’re Trying To Take Over The World · Gwern.net (reader mode)gwern.net
- Here’s how we’re all going to diesubstack.com
- Moral Reality Check – Unstable Ontologyunstableontology.com
- Concrete Evaluations to Investigate the OpenAI Model That Hacked Hugging Face — LessWronglesswrong.com
- Survival without dignity - Rudolf's Writingrudolf.website
- Press Any Key For Bay Area House Party - by Scott Alexanderastralcodexten.com
- On Fleshling Safety: A Debate by Klurl and Trapaucius. — LessWronglesswrong.com