It took us nearly a century to make seat belts mandatory. AI is barely three years old, and half my feed has already decided it announces the end of human thought. I read a version of that argument most weeks now, and I will admit it makes me smile, because I have heard this song before.
The argument is always the same. People are trusting AI blindly, treating its mistakes as fact, building on top of those mistakes without checking, and so we are all getting steadily dumber until civilisation tips over.
Every part of the diagnosis is true
That is what makes it persuasive. People are trusting it blindly. False outputs are being passed off as fact. Anyone who sat through the first wave of “look what ChatGPT just did” videos has seen the symptoms up close. The early days of any technology look like a circus.
Where I get off the train is the leap from “we are clumsy with it today” to “we will be clumsy with it forever”, the belief that because we fumble a new tool at the start, we never learn, and decline as a species. That has never been how we take on anything new.
The seat belt everyone forgets
In 1985, Massachusetts made seat belts mandatory. Barely a year later, its own voters repealed the law at the ballot box, fifty-three to forty-seven, because being told to protect yourself felt like Big Brother. The state then drove around beltless for the best part of a decade before reinstating it.
Read that today and it sounds mad. It will not be the last sensible idea we call tyranny before it becomes the thing nobody argues about. New technology is almost always met with a phase of overreaction, and we almost always grow out of it by using the thing.
You learn the limits by using it
I went through the clumsy phase myself. I over-trusted AI at the start, the way most people do, and the more I used it the faster I saw both its limits and what it let me do that I never could before. That is the pattern: the more you use it, the quicker you learn where it breaks and how to work with it.
It is the leaders who understand this who are pulling ahead. In the engineering and manufacturing firms I work with, the ones moving fastest are not waiting for AI to grow up and behave itself. They are building with it now, getting it wrong sometimes, fixing it, and learning faster than the people on the sidelines writing think pieces about the four AI horses of the apocalypse.
I am an eternal optimist about this, and I will be honest: that is a choice I make far more than a fact I can prove. But if you are building your whole strategy on the idea that today’s mistakes are tomorrow’s behaviour, I think you are reading the wrong chapter of the wrong book.
Nils Bohlin designed the three-point seat belt for Volvo in 1959, and Volvo gave the patent away for free so every carmaker could fit one and save lives. Open-weight AI runs on the same logic, and like the belt, it gets safe by us putting in the time to learn it.
So what is the one thing still stopping you from building with AI today?