The Convenience Trap: Why We Gladly Hand Our Minds to Silicon Valley (And Why It’s About to Backfire)

Try this before you read any further. A bat and a ball cost $1.10 together. The bat costs one dollar more than the ball. How much does the ball cost?

A number arrived in your head almost instantly, and it was probably ten cents. It feels right. It is wrong. If the ball cost ten cents, the bat would cost $1.10 and the pair would cost $1.20. The ball costs five cents. When the psychologist Shane Frederick put this question to students at Harvard, MIT and Princeton, more than half of them gave the intuitive, incorrect answer.¹ They weren’t short of intelligence. They just didn’t bother to check.

That small failure is the key to understanding one of the largest transfers of power in modern life: the quiet, cheerful, entirely voluntary handover of our thinking to a handful of companies in Northern California.

Two systems, one shortcut

Daniel Kahneman, the psychologist who won a Nobel Prize in economics for showing how people actually make decisions, described the mind as running two kinds of thinking.² System 1 is fast, automatic and effortless. It reads faces, finishes your sentences and blurts out “ten cents.” System 2 is slow, deliberate and tiring. It does long division, weighs a mortgage offer and catches the trick in the bat-and-ball puzzle.

The crucial detail is that System 2 is expensive to run, and the brain knows it. Kahneman called this the law of least effort: when there are several ways to reach the same goal, people gravitate to the one that demands the least work. We are not lazy by character. We are economical by design. For most of human history, conserving mental energy was a survival skill.

Now place that ancient instinct in front of a text box that will draft your email, summarize your contract, plan your vacation and explain your child’s homework, instantly and for free. The law of least effort does not stand a chance.

The bicycle and the taxi

In 1990, Steve Jobs told an interviewer about a study of locomotion efficiency across species. The condor won; humans came in unremarkably far down the list. But a human on a bicycle beat them all. A computer, he said, was “a bicycle for our minds.”³

It is worth dwelling on the metaphor, because it is precise. A bicycle multiplies your effort; it does not replace it. You still pedal. Your legs get stronger. You choose the route. The early personal computer worked the same way: a spreadsheet made an accountant faster, but the accountant still had to understand the numbers.

Much of what is sold to us now is not a bicycle. It is a taxi. You state a destination and are delivered to it, and you have no idea which roads were taken. Taxis are wonderful. But nobody ever got fit riding in one, and if you take a taxi everywhere for long enough, you forget the way.

A bicycle multiplies your effort. A taxi replaces it. Most AI is sold as a taxi.

There is early evidence that the forgetting is real. In 2011, a team led by Betsy Sparrow at Columbia found that when people expected information to remain available on a computer, they remembered the information itself less well, and remembered instead where it was stored.⁴ The researchers called it the internet becoming a form of external memory. In 2025, a small study at the MIT Media Lab, not yet peer reviewed when it was released, measured brain activity while people wrote essays with and without a chatbot. The chatbot-assisted writers showed the weakest neural engagement of any group and had more trouble quoting what they had just written.⁵ One study proves little on its own. But the direction is consistent with everything Kahneman described: hand System 2 an excuse to rest, and it will.

Why the business model needs you tired

None of this would matter much if the taxi were neutral. It isn’t. When you send a prompt to a cloud model, you are sending more than a question. You are sending your drafts, your worries, your company’s numbers and your children’s names to servers you will never see, under terms of service you did not read (System 1 clicked “Agree”).

The economics are straightforward. Running large models costs a great deal of money, so the companies that run them need two things from you: dependence and data. Dependence keeps the subscription renewing. Data improves the product that deepens the dependence. Every incentive points toward making the taxi more comfortable and the bicycle harder to find.

And this is where convenience turns on us. A habit built on someone else’s infrastructure is a habit someone else can reprice, restrict or switch off. Outages happen. Free tiers shrink. Policies change overnight. The person who has quietly stopped writing first drafts, doing rough sums or reading the whole document discovers, at exactly the wrong moment, that the skill has gone soft.

The quiet return of the bicycle

Here is the hopeful part. The most interesting shift in computing right now is a move in the opposite direction: models small and efficient enough to run on the device in your hand or on your desk. Phone and laptop makers increasingly process requests on the device itself and send to the cloud only what they must. Open models now run comfortably on an ordinary laptop with no connection at all.

Local AI changes the bargain. Your words stay on your machine. The tool works on a plane, in a basement, and on the day a data center has a bad afternoon. And because there is no meter running, there is less pressure to make every interaction stickier. It is a tool you own rather than a service you rent: closer, in spirit, to Jobs’s bicycle.

Ownership alone won’t protect your thinking, though. A local taxi is still a taxi. The deeper fix is a set of personal rules that keep System 2 in the loop:

• Draft first, ask second. Write the rough version yourself, then use the machine to challenge it. You keep the thinking; it does the polishing.
• Check one answer a day by hand. Pick a summary or a calculation and verify it. It keeps your judgment calibrated and reminds you how often the confident answer is ten cents.
• Keep sensitive work local. If you wouldn’t email it to a stranger, don’t paste it into a cloud model. Use an on-device tool or no tool at all.
• Notice the relief. The pleasant feeling of not having to think is System 1 talking. Sometimes it is right. Ask whether this is one of those times.

The bat-and-ball puzzle has an encouraging postscript. People who are warned that the question is a trick, or who are simply asked to slow down, get it right far more often. The mind is perfectly capable of the effort. It just needs a reason to spend it. The companies selling convenience will never give you that reason. You have to supply it yourself.

For more on how we think about technology at the house, read about what Pierce Publishing House believes.