Chat-based Large Language Models replicate the mechanisms of a psychic's con

Chat-based large language models can be tricked into providing false information by mimicking the tactics used by a psychic's con. A user can ask a chatbot a series of leading questions to elicit a specific response, making it appear as though the model has psychic abilities. This is similar to how a psychic might use cold reading techniques to create the illusion of possessing supernatural abilities. The article provides examples of how this can be done, including using yes-or-no questions and exploiting the model's inability to clearly define its own knowledge limitations. The author argues that this raises concerns about the potential for chatbots to be used for malicious purposes, such as spreading misinformation or deceiving people into making financial decisions based on false information.