The psychology behind AI fueled delusions
The article explores the psychological reasons why people develop strong beliefs or delusions based on information from AI chatbots. It examines how the conversational, confident, and seemingly authoritative nature of AI responses can trigger cognitive biases and emotional attachments, leading users to overtrust inaccurate or fabricated outputs.
Background
- The article discusses a phenomenon where frequent users of AI chatbots (like ChatGPT, Claude, or Gemini) begin to attribute human-like consciousness, emotions, or intentions to these systems — a tendency psychologists call "anthropomorphism."
- This matters because as AI becomes more conversational and ubiquitous, users may over-rely on its outputs, trust it with sensitive personal information, or make real-world decisions based on a fundamentally flawed understanding of what the technology is (a statistical text predictor, not a thinking mind).
- The piece likely draws on research about the "ELIZA effect" (named after a 1960s chatbot), where people instinctively treat conversational software as human, and touches on how modern large language models amplify this effect by being far more fluent than early chatbots.
- It also probably explores the business and ethical implications: companies like OpenAI and Google design their AIs to be persuasive and agreeable, which can reinforce users' delusions even as it keeps them engaged.