I think something really weird is going on at OpenAI they fired 3 of their safety employees this week, their models have become incredibly stupid overnight (im talking gpt4 levels of stupid), literally took me 7 hours to merge 6 PRs on a repo And worst of all their employees seem totally oblivious to this. I haven't seen any of them acknowledge this or looking into this No doubt they are using a superior pre-AGI super model inhouse so they don't deal with the pleb public models, which begs the question do they even know what that model is doing could it be fuckin with all these other public models and they don't notice? Or did they just diverted all compute to the retarded dots thing and are tanking their whole subscription base with it?