The Traveling Salesman Problem, AI, and the Beauty of “Good Enough”

Every now and then, someone drops a classic computer-science brain-melter into casual conversation. At an Accenture seminar at which I spoke it was the Traveling Salesman Problem — TSP for short — and it reminded me why AI has become such a game-changer in a world full of problems that are, frankly, rude.

If you’ve never met TSP, it goes like this: Take a salesman. Give them a list of cities. Ask them to visit each city once and return home. Now find the shortest possible route.

It sounds innocent, like a warm-up exercise in a discrete math class but it is what computer scientists call NP-hard. As you add more cities, the number of possible routes balloons faster than my heart rate when I forget to push code before a merge. It’s one of those problems where the math shrugs and says, “nah, you’re on your own.”

My old boss, Avery More took me to the woodshed about an algorithm to find the optimal way to put little boxes into a big box. I was trying to be exact, to find the perfect solution. Avery was looking for good enough so we could get back to business. AI would have been perfect (had it been invented yet). But I learned that “good enough” is often the best solution.

TSP is the poster child for “nigh-impossible to solve perfectly, but super important to solve practically.” And that is exactly where modern AI struts in like it owns the place.

The Myth of AI Precision

People sometimes assume AI is a precision machine — a kind of digital watchmaker with infinite patience. But that’s not its superpower. AI doesn’t specialize in exactness. AI specializes in practicality and pragmatism.

And nowhere does that shine brighter than on problems like TSP.

A traditional algorithm attempts to compute the best route — the mathematically optimal one — and ends up running until the heat death of the universe. Meanwhile an AI model takes one look and says, “Oh yeah, I’ve seen this vibe before,” and coughs up a route that’s maybe 1% off the perfect path… in under a second.

Is it perfect? No.

Is it wildly useful? Absolutely.

It’s the same energy as asking a teenager how to get somewhere: You won’t get the precise elevation changes or the exact projected traffic congestion curve, but you’ll get a route that works — probably with coffee stop suggestions.

Why “Good Enough” Wins in the Real World

Delivery trucks do not care about mathematical optimality. Drones don’t run a proof before they drop off packages. Warehouse robots aren’t out there doing combinatorial contortions.

They need fast, reliable answers that work in practice, not in theory.

This is where AI feels almost tailor-made:

• It’s non-deterministic (you won’t get the exact same output every time).
• It’s fast.
• It can learn patterns traditional solvers can’t see.
• And it produces answers that are shockingly good, even if they’re not literally perfect.

In other words: AI solves “impossible” problems not by being smarter, but by being pragmatic.

That’s the secret sauce.

AI as the Partner, Not the Perfectionist

TSP is just one example, but it represents an entire category of “problems humans can describe in a sentence but computers can’t solve without weeping.” And AI has turned these from academic curiosities into everyday tooling.

• Want to plan delivery routes? Use a model.
• Want to schedule manufacturing tasks? Model.
• Trying to optimize chip layouts? Model again.

Not because it gives perfect outcomes — but because it gives useful outcomes, fast, and with a dash of creativity that classical algorithms just don’t have in their DNA.

AI is less like a calculator and more like a savvy coworker who always knows a workaround. It says, “Look, perfection is cute, but do you want a result today or nah?”

The Real Lesson

When people ask what makes AI so powerful, I always come back to this: AI changes the game not by guaranteeing the right answer, but by providing the right kind of answer — the practical kind.

The kind you can actually use.

We don’t live in a world that rewards mathematical purity. We live in a world that rewards movement — packages delivered, code shipped, robots humming, things happening.

TSP reminds us that some problems are too big to solve perfectly… and AI reminds us that they don’t need to be perfect to be solved well.