Can Threats Really Make AI Smarter?

Sergey Brin just dropped a wild claim: AI chatbots like Gemini and ChatGPT might actually perform better when you threaten them (yes, with physical violence). Is this just a quirky bug in the system, or does it reveal something deeper about how these models interpret prompts? Microsoft blames poor prompt engineering for Copilot’s lag behind ChatGPT, but should we really be resorting to threats to get better answers? What do you think—genius hack or ethical red flag? #AIethics #TechDebate #Chatbots #Tech