Dev.to AI 🤖 Ai 👁 0 📖 2 min read

The AI's answer was correct. Why did it still take 10 comments to fix his problem?

I've been answering Windows activation questions on Microsoft's Q&A forum lately, mostly because I deal with this stuff all day anyway. Every question there now gets an AI answer automatically. And honestly? They're good

I've been answering Windows activation questions on Microsoft's Q&A forum lately, mostly because I deal with this stuff all day anyway. Every question there now gets an AI answer automatically. And honestly? They're good. Well written, accurate, links to the right docs.

Today a guy posted that his Windows key stopped working after he swapped in a new hard drive. The AI answer was textbook: install the same edition, sign in with your Microsoft account, run the activation troubleshooter. All correct.

It didn't fix anything.

What actually happened took about ten comments to dig out. The "key" he'd saved before swapping drives was a placeholder, not his real one, so a key checker called it invalid and he thought his license was gone. The troubleshooter showed zero PCs on his account, even though he had three others signed in, because none were ever linked to the license. Then we got one error code, then a different one, and only then did the real picture show up.

None of that was in his original question. He didn't know which details mattered, so he couldn't have written them down. The AI answered the question he asked. The humans in the thread answered the question he should have asked, by asking him things back.

That's the part I keep thinking about. AI is great at "here's the answer." It's much worse at "wait, where did that key come from?" Asking a follow-up and noticing something sounds off is most of what real troubleshooting is. Every good support person I know spends more time asking questions than giving answers.

And the thread itself is now the useful thing. The next person who swaps a drive and hits the same wall will find the whole conversation, the wrong turns included, which a single polished AI answer can't give them.

So I'm curious what you've seen:

  • Have you had an AI give you a perfectly correct answer that didn't help, because the real problem was something you didn't think to mention?
  • Do you think AI assistants will get good at asking the follow-up questions, or is that always going to be the human part?
  • If forums fill up with instant AI answers, do fewer humans bother replying, and does that kill the back-and-forth?

Genuinely want to hear from people who do support work, the AI answers and the human ones.

📰 Read the original article on Dev.to AI

Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.