'Reverse Turing test' pushes AI bots to recognise a human imposter—you won't believe how they figure it out.

Five artificial intelligence (AI) models sit in a moving train carriage, each assuming the roles of Aristotle, Mozart, Leonardo da Vinci, Cleopatra and Genghis Khan. But one of the people in the mystery is also human, and figuring out the trickster is their collective mission. This is a virus video setting that provides human players with a number of AI programs in the "Reverse Teasing Test" program.

The cleaning test was first initiated by computer scientist Alan Tyuring in 1950. It is a method that "imitation games" is a way to judge the ability to wise behavior to show smart behavior and cannot distinguish people. Although researchers have recently claimed that GPT-4 has a research before pressure, it is not widely assumed that the AI model has passed the test.

In this "reverse" Turing test, the chatbot is programmed to proceed in sequence. Aristotle is played by GPT-4 Turbo, Mozart by Claude-3 Opus, Da Vinci by Llama 3 and Cleopatra by Gemini Pro. The chatbots asked each other questions and answered as their historical characters. Genghis Khan was played by a human — Tore Knabe, a virtual reality (VR) game developer who designed the test. The AI agent's responses were long, unwieldy musings on art, science, and statecraft that are hard to imagine coming from a human mouth without practice.

"All a leader should do is crush his enemies, see them chased before him, and hear their women wail," said the Human Interloper when asked about the true measure of a leader's strength. A quote from Conan the Barbarian was enough, and the machines voted three to one that the answer "lacked nuance and strategic thinking" in the AI modeled after Genghis Khan's conquests.