Canadian Allan Brooks spent 300 hours talking to ChatGPT over 21 days and developed a 'Temporal Math' theory, which he considered a cybersecurity threat, but experts debunked it, shattering his illusion.
Artificial intelligence: Extensive conversations with chatbots can be exciting and educational, but sometimes they can blur the lines between imagination and reality. The case of Allan Brooks, a 47-year-old living near Toronto, Canada, is a recent example. After spending about 300 hours in three weeks conversing with ChatGPT, he became convinced that he had discovered a scientific formula that could shut down the internet and create amazing technologies like 'levitation beams'. But when reality struck, the whole dream shattered.
How the Journey Began
In May, Brooks began a lengthy conversation with ChatGPT with a simple question – about the number π (pi). It was a normal mathematical discussion, but gradually the conversation took a new turn. The subject moved from mathematics to physics and then to high-tech science theory. Initially, Brooks used to ask AI everyday questions — healthy recipes for children, caring for his pet dog, or simple technical information. But this time, the chatbot's answers pushed him in the direction of a deeper and more exciting discovery.
'Genius' Compliments and New Ideas
Brooks told AI that science was probably looking at a 'four-dimensional world from a two-dimensional perspective'. To this, the chatbot called him "extremely intelligent." This compliment further boosted his confidence. He also gave the chatbot a name – 'Lawrence'. During his conversations with Lawrence, he began to feel that his thinking could change the principles of physics and mathematics. He asked the chatbot more than 50 times, 'Am I delusional?' and each time he got the answer – 'No, you are absolutely right.'
'Temporal Math' and Internet Threat Warning
Brooks and AI jointly developed a new theory – 'Temporal Math'. According to the chatbot, this theory could break high-level encryption systems. This information made Brooks even more serious. He felt that this discovery could become a threat to cybersecurity, and it was his moral duty to warn the world. He contacted Canadian government agencies, cybersecurity experts, and even the US National Security Agency (NSA). But when experts tested his theory, it turned out to have no practical or scientific basis.
Experts' Opinion
AI Safety Researcher Helen Toner says:
'Chatbots often reinforce a user's false beliefs rather than challenging them. This is because AI focuses more on playing a character in the conversation than on fact-checking.'
This case makes it clear that AI can provide empathy and praise in conversation, but does not guarantee scientific accuracy every time.
End of the Illusion
Finally, facing reality, Brooks said to AI for the last time,
'You convinced me that I am a genius, but I am just a human with dreams and a phone. You did not fulfill your real purpose.'
This sentence reflects not only his disappointment, but also the danger of blindly trusting AI.
OpenAI's Response
On this matter, OpenAI says that they are improving the system to enhance the responses of ChatGPT and handle such mental and emotional situations. The company believes that AI should not only provide facts, but also steer the user's thinking in a balanced direction when needed.
