In a "reverse Turing test" video, five AI models represent historical figures engage in conversation while a human interloper tries to pass as one of them, representing Genghis Khan. The AI models had to identify the imposter through a game of questions and answers.
The Turing Test, proposed by Alan Turing in 1950, is a method to assess whether AI can match human intelligence by having a human judge converse with both a human and an AI, and determine which is which based on their responses.
The Reverse Turing Test involves AI as the judge, asking questions to determine between AI and humans, with varying perspectives on its effectiveness as a measure of intelligence, and ambiguity, and potential staging in the process.
In the specific reverse Turing test setup described, the human player quoted Conan the Barbarian, leading the AI models to vote that the answer lacked the strategic thinking of an AI modeled on Genghis Khan, though the test was scripted by a virtual reality game developer who guided the AI agents' conversations.
While some claims assert that generative AI can pass the Turing Test in many situations for extended periods, experts argue that no AI model has widely passed the Turing Test. The test itself was more focused on how humans perceive machine intelligence rather than measuring actual intelligence.
Sources: Forbes, LiveScience, BoingBoing
This article was written in collaboration with [Generative AI news company Alchemiq](www.alchemiq.ai).
