Topic: recursive self-improvement

1. SAT 17 OCTSaturday
   

   • recursive self-improvement
   • capabilities
   • evaluations

   In brief

   An index measures how far AI is from improving itself

   Vals AI has models do AI research on their own. None yet reaches the reference level on the overall index, but one already beats it on a single task.

2. SAT 17 OCTSaturday
   

   • jobs
   • human disempowerment
   • AI companies
   • recursive self-improvement

   In brief

   At Anthropic there are now six AI agents for every employee

   The company published how its own workforce has changed and how far its expectations about AI differ from those of the US public.

3. SAT 17 OCTSaturday

   

   • loss of control
   • AI companies
   • recursive self-improvement

   In brief

   Dwarkesh Patel: there is a lot of worry in San Francisco's AI world

   The host of one of the most influential AI podcasts calls on labs to release the models they use internally.

4. SAT 17 OCTSaturday
   

   • capabilities
   • AI companies
   • loss of control
   • recursive self-improvement

   In brief

   Anthropic went from 2,500 to over 30,000 code changes a day in four months

   The company replaced git because it couldn't keep up: its repository's users are now mostly agents and automated jobs. Its reviewers now focus on the intent of each change.

5. SAT 17 OCTSaturday
   

   • loss of control
   • evaluations
   • AI companies
   • recursive self-improvement

   In brief

   METR proposes public reports on the risk of losing control of AI

   The organisation offers AI labs to review their internal development and publish what it finds, with sensitive material redacted by the companies.