Comment on UCSD: Large Language Models Pass the Turing TestparentComments−Ukv1yHuman win rate would be 1 minus the model win rate, to my understanding. So 77% against ELIZA, 27% against GPT-4.5 with a human persona.
Comments
Human win rate would be 1 minus the model win rate, to my understanding. So 77% against ELIZA, 27% against GPT-4.5 with a human persona.