Today's AI can't learn like humans do

Do you know what AI cannot do? It cannot learn like you do.

Today's AI is very good at one thing. It can play chess. It can read pictures. It can write emails. But it only does what we teach it. If you change the problem, AI gets confused.

Humans are different. You learn from everything. You see a cat once. Now you know cats. You find a new street. You navigate it. You try a new food. You like it or you don't. Your brain adapts every day.

Yan LeCun is a famous scientist. He won a big prize for his AI work. Now he has a new idea. He wants to build AI that learns and changes. Not AI that just follows one rule.

This new AI would be more flexible. It could learn from new situations. It could work on problems nobody taught it before. It would be smarter — really smarter.

Why does this matter? Because today's AI has limits. It needs lots of data. It needs lots of time. It needs humans to guide it every step. The new AI would need less help.

Some people worry about very smart AI. But scientists like LeCun think differently. They say the real challenge is making AI that learns like humans. Not making AI that is just bigger or faster.

So what does this mean for you? Maybe jobs change. Maybe school changes. Maybe the tools you use every day become smarter and easier.

What job do you do? What would you want AI to learn?

Inspired by reporting from BBC News.