What if AI stops being the cheap option?

I sometimes wonder how much of the way we use AI depends on what it costs right now.

Suppose the tools you use became expensive enough that you had to think before opening them.

Would you use them for the same things?

And would you count the cost of the answer, or the cost of finally getting one you could use?

Let’s say the first answer is wrong. You explain again. The second one misses something. You correct it. A few more attempts, and the eighth answer finally does what you wanted.

You got there. But it took eight attempts, your explanations, and enough understanding of the task to recognise seven answers that weren’t good enough.

If you were paying for every attempt, that would be the cost of getting the result.

Even without that bill, you’ve spent time reading, checking and correcting.

I think we sometimes overlook this when we talk about how useful a model is. Someone spends ten or twelve messages getting a simple task done and comes away impressed that the AI eventually managed it.

I’d want to ask how long the task would have taken without it.

Some work is complicated enough to deserve that back and forth. But if I’m repeatedly explaining a small change I could have made myself, I should probably notice when the explanation has become more work than the change.

That’s also what makes me wonder about the business model.

If a service earns more each time I try again, what incentive does it have to help me finish in fewer attempts? I’d want that question considered before repeated attempts become something we accept without thinking.

I’m not predicting that every AI service will suddenly become unaffordable. A higher price might simply make us examine habits we’re currently comfortable with.

We did things before these tools existed. Sometimes I think we forget that a little too quickly. Asking AI becomes the first step, even when we already know how to do the task.

I’d still pay for help that saved me hours. I’d think twice about paying to spend twenty minutes explaining something I could have done in ten.

If every attempt had a noticeable cost, when would you stop asking and do it yourself?