AI told Reddit's biggest jerks they did nothing wrong 51% of the time Researchers at Stanford and Carnegie Mellon ran a test to measure how often chatbots flatter users. In a paper on social sycophancy in AI advice, they ran OpenAI, Anthropic, Google, Meta, DeepSeek, Qwen, and Mistral through thousands of personal ... boingboing.net boingboing.net / feeds boingboing-net-psychology .. boingboing-net-science / / #psychology / / 21 days 21d Share
Your brain guesses at reality instead of recording it A gray apple looks red because your brain invents reality, guessing ahead every 200 milliseconds. boingboing.net boingboing.net / feeds boingboing-net-science .. boingboing-net-psychology / / #science / / 26 days 26d Share