zlacker

> Give criteria that 4 fails that a significant chunk of the human population doesn't also fail and we can talk.

The ability to respond "I don't know." to literally any question.

>>davegu+(OP)
I've had 4 respond i don't know to questions before.

and the base model was excellently calibrated. https://openai.com/research/gpt-4

"Interestingly, the base pre-trained model is highly calibrated (its predicted confidence in an answer generally matches the probability of being correct). However, through our current post-training process, the calibration is reduced."

Next ?