What is Nostr?
geeknik
npub1fk8…c5sk
2023-12-03 23:20:23

geeknik on Nostr: Researchers at UC San Diego conducted an experiment to determine whether OpenAI's ...

Researchers at UC San Diego conducted an experiment to determine whether OpenAI's GPT-4 could pass the Turing Test, in comparison to GPT-3.5, the 1960s program ELIZA, and human participants. GPT-4 achieved a success rate of 41%. The study, released as a preprint on arXiv and not yet peer-reviewed, revealed that variables such as education level and familiarity with AI did not markedly affect the participants' capacity to identify AI presence. The study emphasized the enduring significance of the Turing Test in evaluating the abilities of AI in social interaction and deception.

https://arstechnica.com/information-technology/2023/12/real-humans-appeared-human-63-of-the-time-in-recent-turing-test-ai-study/
Author Public Key
npub1fk8rya2ra7lp8m60f8jrjg4yqfv2cc8dah8wqc49drccs3dqngzqtgc5sk