💡 The biggest risk of AI in education isn’t that kids will cheat. It’s that they’ll never learn to think.
Priya Lakhani, an AI education entrepreneur, explains in this TED Talk (TEDNext 2025) why we need to redesign how AI interacts with students.
The problem with “easy” AI:
When a student asks AI “what’s the answer?”, the AI gives it. The student copies it. No learning happened.
Cognitive science is clear: effort and difficulty are necessary for knowledge to consolidate in memory. If AI eliminates that effort, it eliminates learning.
💡 How should AI work in education?
Instead of giving answers, it should:
- Ask questions that guide the student to discover the answer
- Detect comprehension level and adjust difficulty
- Reinforce active learning, not passive reception
An example:
Student: “What is 15 × 8?” Wrong AI: “120” Educational AI: “Can you calculate 15 × 8 by breaking it into simpler parts?”
💡 Explanation in a nutshell#
The “desirable difficulty” concept in learning sciences says we learn better when we have to struggle. An AI that simply gives answers acts like a calculator: useful, but doesn’t build critical thinking. The challenge is designing AI that teaches without doing all the work for you.
More information at the link 👇
