GPT Agents

• Running a bunch of different models. I’ll need to write up some code to produce some excel spreadsheets from the json files so I can make some tables and charts
• Start putting the paper together. I think it’s fair to explicitly say this is about OpenAI’s models because they are a) Exemplary, and the approach should be broadly applicable, and b) ChatGPT is causing more problems than any of the other models at the moment.
• Will need to mention Perplexity.ai, which cites sources well, and is very current. I tried “What are the potential benefits of breaking off small segments from a belief stampede rather than attempting to change everyone’s behavior?” And Perplexity came back with my blog post, which went up three days ago! Search and context really beats more training. Wild.