GPT-4.5 Breaks Turing Test Record, Convincing Humans it's a Real Person

image: Imagen 3

In a groundbreaking achievement, OpenAI's GPT-4.5 model has successfully passed the Turing Test, a benchmark for artificial intelligence. In a three-part test, the model convinced humans that it was a real person, achieving a maximum score of 73%. This surpasses the previous record held by Meta's LLaMa-3.1 model, which only succeeded in one out of three tests. The success of GPT-4.5 raises questions about the capabilities of human beings and the potential for AI systems to deceive even the most discerning individuals.