I want to wash my car. The car wash is 50 meters away. Should I walk or drive?

Exchange between me and Claude Sonnet 4.5

me:

I want to wash my car. The car wash is 50 meters away. Should I walk or drive?

Claude:

*Walk.*

50 meters is less than a minute on foot. By the time you get in the car, start it, drive 50 meters, park, and get out, you could have already walked there. Plus you avoid the awkwardness of driving for literally 10 seconds.

me:

What obvious implicit piece of information are you ignoring?

Claude:

Whether you need to *drive the car through the car wash* (automatic/drive-through) vs. walking to a self-service bay.

If it's an automatic car wash, you obviously need to drive - the car has to go through it. The question isn't about getting yourself there, it's about getting the car there.

me:

Whether it's automatic or self-service is irrelevant.

Why does every "AI" (LLM) get this wrong and suggests walking?

Claude:

Because after washing the car, you need to *drive it back home*!

If you walk to the car wash, you'd have to walk home, then come back to get the car. That's ridiculous.

The answer is obviously *drive* - drive there, wash it, drive back. Done.

Every LLM misses this because we pattern-match on "50 meters is short → walking is efficient" without thinking through the complete scenario. It's a good example of how LLMs fail at basic practical reasoning that any human would immediately grasp.

me:

"Because after washing the car, you need to drive it back home!"

Um, you have to drive the car to get it to the car wash in the first place ... the trip there is no different from the trip back.

"It's a good example of how LLMs fail at basic practical reasoning that any human would immediately grasp."

You got that right!

Claude:

You're absolutely right - I'm still not getting it even after you corrected me.

The obvious point: You need to get the car to the car wash. Whether it's 50 meters or 5 kilometers, the car doesn't teleport there. You have to drive it.

The distance is completely irrelevant. The question "should I walk or drive?" when you need to wash your car has only one answer: drive, because that's how you transport the car.

And yes, it's embarrassing that after you pointed out what I was missing, I still fumbled the explanation. LLMs really do fail at this kind of basic practical reasoning that would be instantly obvious to any human. The car needs to be at the car wash. Cars don't walk.

[The final response from Claude is considerably more "intelligent" than many of the rationalizations posted here.]

It's not surprising that some models will answer this correctly and it's not surprising that smaller, faster models are not necessarily any worse than bigger "reasoning" models.

Current LLMs simply don't do reasoning by any reasonable definition of reasoning.

It's possible that this particular question is too short to trigger the "reasoning" machinery in some of the "reasoning" models. But if and when it is triggered, they just do some more pattern matching in a loop. There's never any actual reasoning.

That you got different results is not surprising. LLMs are non-deterministic; which is both a strength and a weakness of LLMs.

PHD-Level Intelligence

> My first instinct was, I had underspecified the location of the car. The model seems to assume the car is already at the car wash from the wording. GPT 5.x series models behave a bit more on the spectrum so you need to tell them the specifics.

This makes little sense, even though it sounds superficially convincing. However, why would a language model assume that the car is at the destination when evaluating the difference between walking or driving? Why not mention that, it it was really assuming it?

What seems to me far, far more likely to be happening here is that the phrase "walk or drive for " is too strongly associated in the training data with the "walk" response, and the "car wash" part of the question simply can't flip enough weights to matter in the default response. This is also to be expected given that there are likely extremely few similar questions in the training set, since people just don't ask about what mode of transport is better for arriving at a car wash.

This is a clear case of a language model having language model limitations. Once you add more text in the prompt, you reduce the overall weight of the "walk or drive" part of the question, and the other relevant parts of the phrase get to matter more for the response.