AI BriefingGPT-4.5 Passes Turing Test with 73% Score
Key point
GPT-4.5 passed a controlled Turing test, being judged human by evaluators 73% of the time.
Details
The first peer-reviewed and pre-registered study demonstrating that an LLM passed the Turing test under controlled conditions has been published.
GPT-4.5 passed the Turing test with a 73% human evaluation rate, while GPT-4o scored below chance level, showing a stark performance gap.
The study suggests that fraud detection, content moderation, and identity verification tools built on the assumption that 'AI-generated text is detectable' may fail at the level of GPT-4.5.
Notably, persona instruction design was revealed to be a key variable determining the risk of deceptive use, meaning that operators cannot reliably predict AI's misuse and deception risks from simple model benchmarks alone.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Stanford and NVIDIA Release 'Contrastive Language Model' CLM with 9x Faster InferenceTechnical Rebuttal of TypeSafe Jev Model's 'Calibrated Probabilities' Claim
Amazon Launches AI, Energy, and Healthcare Research Initiative with Stanford