tailcalled comments on TurnTrout’s shortform feed

[tailcalled]

To what extent do you worry about the training methods used for ChatGPT, and why?