5 min read

The Engine Is Not the Car

The Engine Is Not the Car

Walk outside. Look at the street. Count the cars.

There are probably dozens within view—sedans, trucks, SUVs—parked, idling, threading through traffic. You probably don't think about them. Barely anybody does. They're background, furniture. In a way, among the most unremarkable objects in the modern world. Ubiquitous.

Now hold that thought while I tell you what you're actually looking at.

In the late 1800s, a handful of people scattered across Europe and America figured out that if you ignited gasoline vapor in a confined space, you could capture the explosion and turn it into rotational motion. Most were working in barns, sheds, small workshops. Their machines shook violently, leaked oil, terrified horses, and broke down constantly. The idea that these clattering contraptions would reshape civilization would have seemed absurd to most.

A century and a half later, the automobile has restructured how humans live in ways that are difficult to fully comprehend: where cities are built, how far people live from where they work, how goods move across continents, how wars are fought, how teenagers become adults. The same object, the same capability, reshaping geography, economics, family structure, urban planning, supply chains, geopolitics. Not metaphorically but literally. The same force operating at every scale of human organization, globally.

We call it a "car" like it's a simple thing, a noun. An object you purchase and own. But the word obscures what you're actually dealing with. A car is not so much an object as it is a solution. A solution to a problem that constrained human civilization for its entire existence: how do you move mass through space faster than muscle allows? How do you compress distance? How do you make geography less tyrannical?

The car is a time machine; a destination that would have taken a day on foot now takes minutes. Goods that were impractical to transport before become routine. The constraint dissolved.

The Machine Beneath the Mundane

Open the hood of any one of those unremarkable objects parked on your street. What's underneath is staggering.

The engine: cylinders, pistons, crankshaft, valves, camshaft—the system that converts fuel explosion into rotational motion. The fuel system: tank, pump, injectors—delivering precisely metered combustion material. The cooling system: radiator, pump, thermostat—managing the heat that would otherwise destroy the engine. The lubrication system: oil pan, pump, galleries—reducing friction between metal surfaces moving at thousands of cycles per minute. The ignition system: battery, alternator, spark plugs—providing precisely timed electrical discharge. The exhaust system: manifold, catalytic converter—managing combustion byproducts. All within an astonishingly space-efficient configuration.

That's just the engine and its support systems. We haven't touched the transmission, differential, driveshaft, axles, wheels, tires, suspension, steering, brakes. We haven't talked about the chassis, body, electrical system, climate control, instrumentation.

Very little of this is optional; each system exists because the utility demands it. The engine needs cooling because it generates heat. Cooling needs circulation. Circulation needs power. The systems don't merely connect—they generate the need for one another, the car bootstraps its own complexity. You could hardly simplify it by removing pieces.

And all of this—the interlocking, mutually dependent machinery—produces something remarkable: a multi-ton vehicle that most people can operate competently. Two pedals, a wheel, a few gauges. That's the interface for several thousand coordinated components hurtling down the highway, under control at 80mph.

This is what a solved problem looks like.

Now Look at What We're Calling AI

The claims are extraordinary. Transformative. "The new electricity." "The new internet." Artificial general intelligence on the horizon. Personal assistants that will reshape how we live and work.

The implicit proposition is simple: scale the language model. Bigger parameter counts, better benchmarks, more compute. Push the capability curve upward and the transformative utility will emerge—just spill out somehow.

Here's what we actually have.

A model takes text as input and produces text as output. You inject the fuel—your prompt, your context—and the mechanism fires. An explosion of computation occurs, and out comes the response. It's genuinely impressive, the power is real.

This is merely a cylinder.

A single cylinder igniting fuel, producing a burst of energy, the fundamental unit of power generation in an internal combustion engine. Necessary, but radically insufficient versus the practical goal.

Making the cylinder bigger doesn't get you a car. Making it more efficient doesn't get you a car. Polishing it, measuring it, writing futurist poetry about it—none of this produces transportation.

From Cylinder to Engine

To be fair, some people are starting to build engines.

We're seeing systems that move beyond the single firing: orchestration layers, tool use, “memory”, feedback loops. Attempts to determine when to inject fuel, how to sequence firings, how to route the output of one explosion into the next.

An agentic coding framework and a car engine's fuel injection system seem like completely different things—one is software, the other is mechanical precision. But they're solving the same problem: how do you deliver the power source to the combustion event in a controlled, timed, sustainable way? Similar pattern. Different substrate.

These are proto-engines. Attempts to turn raw power into coordinated, sustained operation. Valuable, necessary, sputtering, and backfiring.

But the engine is not the car.

Power Generation Is Not Utility

Imagine you've built a perfect engine. Cylinders firing in sequence, fuel delivery optimized, cooling managed. It runs beautifully on a workbench, spinning a shaft with tremendous continuous power.

Now get in it and drive to the grocery store.

You can't, the engine doesn't touch the ground. It has no wheels, no steering, no seat, no way to tell it where to go or know what it's doing. The engine generates power. The car delivers utility. The gap between them is everything else.

Framing this as "model versus system" misses the point. A car is not an engine with stuff around it, it's a unified solution to a problem. At the level of common utility, the distinction between ‘cylinder’, ‘engine’, and ‘car’ dissolves. What matters is whether the thing gets you there.

The Shape of the Solution

So what would the AI equivalent of the car actually look like—not in features, but in structure?

The automobile expressed a solution in metal, rubber, glass, and gasoline. The cognitive partner—the thing people actually imagine when they hear "AI assistant"—would express that solution in software, models, interfaces, and data structures. Different substrates. Same pattern: integrated systems delivering transformative utility through simple interfaces and reliable operation.

The engine is the power layer—models plus orchestration plus tools. The machinery that produces cognitive capability.

But then you need everything else.

The drivetrain: the system that takes power and puts it to use. How outputs connect to anything—systems, environments, consequences. Without this, capability goes nowhere.

The wheels and tires: the point of contact with reality. What keeps the system grounded in what's actually happening, what's actually true.

The suspension: what absorbs the roughness of real conditions. Error handling. Resilience. Graceful degradation.

The steering and pedals: the input interface. How a human directs high-dimensional machinery with low-dimensional intent. This must be radically simpler than the system it controls.

The dashboard: the output interface. The right information, legibly presented at the right time. Enough to operate, not so much that it overwhelms.

The seat and cabin: the livability layer. Whether people can actually use the system in the context of their lives or business. Whether it fits human bodies and human minds.

Each of these is necessary, not features for later. The car does not work without them.

The Frame

This is not a pessimistic argument. The cylinder works. The power is real. Proto-engines are emerging.

What's missing is the framing of the task.

The discourse is so fixated on the cylinder—scaling it, measuring it, prompting it—that it has yet to recognize what we're assembling. The proposition that scaling the model will eventually produce transformative utility is, at minimum, incomplete. At worst, it's a category error. You can’t drive to the store by staring into the cylinder. The cylinder does not become the car by getting bigger.

The automobile reshaped civilization not only because someone built a great engine, but because someone finally built a car—the full system, every piece in place because the utility dictated it.

Walk outside. Look at the cars. Consider what it took.

Then consider what we're building—and what we're not.