The AI Trap Early Adopters Keep Falling Into. Here's How to Escape It | Anthropic, Drew Bent

Drew Bent, head of education at Anthropic, argues that early AI adopters are trapped by outdated mental models of AI capabilities, treating modern models like the limited tools of 2022. He advocates shifting from transactional, step-by-step prompting to treating AI as a collaborator by providing rich context and open-ended problems. He references an Anthropic study showing students who used AI transactionally learned 17% less than those who didn't use it, while inquiry-based AI users performed comparably to non-AI users. Bent envisions AI in education as a personalized learning companion that enhances human-to-human connection rather than replacing it, and predicts that building AI agents will become a foundational professional skill over the next 40 years.