Almost everyone drives a car. Most of us use it for the same few things. Work, the grocery store, picking up the kids.
That is how most people use AI right now. They ask it a question. They have it clean up an email. That's fine. It's just the short trip to the store.
On that same road there are people doing a lot more with a vehicle. A contractor pulling a trailer. A trucker taking a full load through the mountains. An operator running an excavator a few inches from a gas line. A race car driver taking a corner at 150 mph.
All of them are driving. They are just doing it at a different level.
Key takeaways
Nobody asks which vehicle is the best. It depends on the job. You don't haul gravel in a sports car. You don't pick up a date in a dump truck.
AI works the same way. Some tools are good at writing. Some make images. Some write code. Some will run a whole task on their own while you do something else.
If you only know one tool, you only have one vehicle. If the job is hauling, go find the truck.
If I bought a race car tomorrow, I'd still be a guy who drives to the grocery store. The car wouldn't make me fast.
Put a pro in my daily driver and they'd still beat me around a track. They know when to brake. They know where to put the car. They have put in the hours.
AI is the same. Almost everyone can get the same tools for about $20 a month. The people getting great results have the same tool you do. They just know what they want before they ask. They give it the right information. They can tell when the answer is wrong.
Same road, different drivers
| On the road | With AI |
|---|---|
| Errands: work, the store, picking up the kids | Quick questions and cleaning up an email |
| Picking the truck for a hauling job | Picking the right tool for writing, images or code |
| A pro who knows when to brake | Knowing what you want, giving it the right information, catching the wrong answer |
| A car that drives itself while you watch the road | AI running a whole task while you pick the destination |
I drive a Tesla Model 3. The Tesla app says the car has driven itself for 11,482 of its 12,783 miles. That's 90%. Its longest streak is 486.8 miles.
I was still in the driver's seat for every one of those miles. I pick where we're going. I watch the road. I take over when it gets something wrong. And I drove the other 1,301 miles myself.
Tesla is working on taking the driver out. Its Robotaxi service is giving rides in a handful of US cities, some with nobody in the driver's seat. So far that only works inside small mapped areas. Elon Musk has been promising a car that drives itself anywhere since 2016. Tesla's latest target for customer cars is the second quarter of 2027.
AI is headed the same way. It can already do a lot of the driving. Someone still has to pick the destination and know when it took a wrong turn. Even in a Robotaxi, a person decides where the car goes.
At branded by, we let our accountant go. Claude now manages all of our QuickBooks and our tax preparation.
That is a long way from asking AI a quick question. It's the truck doing real work.
I'm still in the driver's seat. I'm the one who signs the tax return. If it's wrong, that's on me.
Nobody tells a trucker the truck did the work. Nobody tells a race car driver the car won. We know someone else built the machine. We still give the credit to the person who drove it well.
People who are good with AI should get the same. They didn't build it. They did steer it.
No more than driving a car you didn't build. Nobody tells a trucker the truck did the work. The credit goes to the person who drove it well, and so does the blame when it goes wrong.
It depends on the job, the same way the best vehicle does. Some tools are good at writing. Some make images. Some write code. Some run a whole task on their own.
It can already do a lot of the driving. Someone still has to pick the destination and know when it took a wrong turn. Even in a Robotaxi, a person decides where the car goes.
