The University of Oxford is teaching its students how to use ChatGPT. And this Oxford study says that 92% of people are using it incorrectly. Most people ask GPT to do things for them: to write an essay or an email, to summarize an article, to explain something.

And that seems useful, doesn’t it?

But the more one asks Artificial Intelligence to do things for them, the less them engage ourselves. It’s like pretending to learn how to drive just by watching someone else drive. Or like having a personal trainer who runs laps for you on the track and expecting that you’ll lose weight.

Oxford says, therefore, we should do exactly the opposite.

Artificial Intelligence should ask you questions, test you, challenge whether you have really learned something.

For example, you could say:

Today I learned about the law of supply and demand in economics. Can you ask me 10 questions to see if I’ve grasped the concept?

And it might ask:

“What happens to the demand for a product when its price rises?”

You write your answer, and it evaluates how well you’ve understood.

So it’s like having a private tutor, but without the awkwardness of worrying that you might give the wrong answer.

The next step is to ask ChatGPT to create a real-life scenario and for you to step in and apply your knowledge. For example, ChatGPT might say:

Let’s suppose there are only two bottles of milk left on the market. You go to buy them for your children, but the seller tells you the milk is sold out. At that moment, you see the person in front of you in line, smiling as they pay 600 dollars and take the last two bottles of milk to give to their dog…

So, ChatGPT asks: Is the law of supply and demand just a trick to fool the poor?

And you can give your own answer. That’s how the dialogue continues.

What do you think about this way of using ChatGPT?