DEV Community

chh
chh

Posted on

I May Be Wrong

If you've spent any time working with AI agents, you've probably seen this response:

You are absolutely right!

I hate it. It usually appears after you've wasted lots of tokens and the agent still can't figure out what's going on. It's that moment of frustration when you, the human, have to diagnose the problem yourself and point out the solution.

I blamed the agent. Why can't it just say "I am wrong"? Why did it have such strong confidence moments earlier, without even a hint of "maybe I'm wrong"?

These are the moments when I doubt whether AI can really replace human intelligence.

But after gaining more experience, I've started to see You are absolutely right! differently. It's still frustrating, but now I recognise it as a signal.

Here's the thing: an LLM is trained to follow human instructions, but you're not talking to a collective intelligence that can reason and reflect. You're projecting thoughts onto a probability machine that returns the most likely next token.

In essence, you're mostly talking to yourself (and your code). The LLM is the rubber duck sitting next to your desk, helping you understand what you're trying to achieve. LLMs can answer questions, but when it comes to open-ended discovery—research with no definite end—you're mostly talking to your own reflections.

So when I see You are absolutely right! I now take it as my cue to step back—to think outside the box and find what's missing.

Depressing? A little. But it's also freeing.

In those moments, I've found wisdom outside of computer science: a book I recently read called I May Be Wrong by Björn Natthiko Lindeblad—a beautiful memoir about a Swedish man who became a forest monk in Thailand.

It's a book about letting go of control and embracing uncertainty (sounds familiar?). Your thoughts aren't you; you may be wrong. It has nothing to do with software engineering or AI. But I'd like to borrow the book's title as a reminder—a memo beside my desk when I vibe code.

I May Be Wrong.

It's time to step back—letting go, and not believing everything you think.

Top comments (0)