Search
Study Asserts GPT-4.5 Large Model Successfully Navigates Turing Test
2025-04-03 /
Read about 0 minute
Author：小编
For the first time, researchers at the University of California, San Diego, have released findings that empirically demonstrate an AI system surpassing the conventional three-party Turing test. In this experiment, four AI systems—GPT-4.5, LLaMa-3.1-405B, GPT-4o, and ELIZA—engaged in conversations with a panel of judges and two human participants. The judges' task was to identify which participants were human. Notably, GPT-4.5 was mistaken for a human in 73% of the interactions, significantly outperforming the accuracy rate for real humans.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
