As answers get cheaper, decisions get more expensive

From opening ChatGPT for the first time in late 2022 to today, I am now in my fifth calendar year of using AI.

As someone who has lived with it from the beginning, the more I use it, the more amazed I am—and the more afraid.

What increasingly frightens me is the people who are better at using it than I am.

Without a doubt, that is the scariest horror-movie plot I can currently imagine…

So how do we make good use of AI?

I do not think it comes down to beautifully written prompts.

It comes down to judgment.

However poor or muddled your spoken explanation might be, increasingly capable models make that a much less difficult problem to overcome.

Even being unable to interact in English is no longer a particularly serious obstacle.

That means anyone can use AI to get answers, or even make something that works.

We marvel at the gains in efficiency, and then, almost without noticing, start overestimating our own abilities.

But what is actually happening?

The gap between people is also growing—so much that it can feel like a vast chasm.

One person uses AI to read a foreign-language book. Another uses it to start a company valued at hundreds of millions of dollars. They use the same technology, yet the value they create differs by orders of magnitude.

AI keeps lowering the barrier to execution while increasing the value of decisions.

When everyone has access to equally powerful AI,

the real difference between people comes back to an ancient question:

Can you think for yourself?