Safety report seems very thorough though, I have to give them some credit (I am not an expert of course), although I continue to wish any lab other than Ant would do model welfare assessments
over in china they are doing with the big fat fish what was done to doro and neco-arc, inspiring, dangerous possibility of her becoming a once in a decade memetic avatar, I will be monitoring the situation.
@aliceisplaying They have custom finetunes of 6 sol and 6 luna for chatgpt dot com. I assume these are tuned for mental health etc. See here https://t.co/K0b3ql2BOs
(They actually also did this for 5.6 sol a while back)
About 50% of my timeline appears to have learned the phrase "galactic algorithm" as a result of the OAI slopdrop based on how many times I've seen it lately
Please resist the urge to mock the mathematicians struggling through this moment. Whatever your field, your expertise, your passion, you too shall one day experience what they are going through. And when you remember your words, you will feel remorse.
@AllanOlley@morallawwithin I think for Columbus (and aliens maybe) you can reason "if there were beings more powerful they would have likely discovered us already" (which was true). Obviously does not work for AI.