ChatGPT-4.5 And The Turing Test: 73% Success Signals A New Era—And New Risks

ChatGPT-4.5, OpenAI’s latest innovation, has reached a decisive milestone by passing the Turing test with an impressive success rate of 73%. This breakthrough demonstrates the ability of modern AIs to convincingly mimic human conversation, but also raises concerns about risks of manipulation and fraud.

Historic Success of ChatGPT-4.5 in the Turing Test

After raising 40 billion dollars in a funding round, OpenAI has reached a major step in the development of artificial intelligence with the launch of ChatGPT-4.5! An AI model that managed to convince the majority of participants in a Turing-type test that it was human. This feat was confirmed by a recent study conducted by the University of California at San Diego, which assessed the ability of large language models to convincingly imitate human conversation.

In this study, ChatGPT-4.5 passed the test in 73% of text conversations, outperforming predecessors such as GPT-4.0 and other AI models like ELIZA and Llama 3.1 405B. This impressive performance highlights ChatGPT-4.5’s ability to detect subtle cues in language and to adopt responses that seem more human.

Between AI Progress and Ethical Concerns

The Turing test, designed by British mathematician Alan Turing in 1950, evaluates whether a machine can imitate human conversation in a way that deceives a human judge. If the judge fails to distinguish the machine from a person, the test is considered passed. According to Cameron Jones, a postdoctoral researcher at the University of California, GPT-4.5 excels in aspects such as the use of emotional language but struggles with real-time information or current events.

As artificial intelligence, with technologies like ChatGPT, continues to develop, it finds applications in many fields, including emerging sectors like Web3. However, it also raises social and economic concerns. Indeed, risks exist, such as misinformation and fraud, where AIs like ChatGPT could be used to manipulate public opinion or commit malicious acts.

While OpenAI already announces the launch of GPT-4.1 for summer 2025, this success of ChatGPT-4.5 in the Turing test demonstrates significant advances made by AI and suggests new possibilities for the future. Hopefully, the excitement around this new model won’t overload the servers like the Ghibli-style images.