When the AI goes in circles
Recognise when the AI is stuck in a loop of failed fixes, stop it early, go back to the last good point, and restart with a fresh chat and a better description.
Why this matters
Here's a pattern you'll meet sooner or later. Something's wrong. You ask the AI to fix it. It says "I've fixed the issue!" It isn't fixed. You say so. It apologises, tries something else, and now something different is wrong. You tell it. It goes back to its first idea. Forty minutes later, the chat is enormous, six files have changed, and you're further from working than when you started.
This is called going in circles, or a loop. Every developer who builds with AI meets it, including professionals. It isn't your fault, and it isn't a sign the problem is too hard for you. It's a known weakness of AI tools: once a chat fills up with failed attempts, the AI keeps getting pulled back towards them.
The fix is surprisingly simple, and it's never "try harder in the same chat". It's: stop, go back, start fresh with a better description. This lesson turns that into a routine you can run in five minutes.
In the full lesson
- The idea in plain English
- Do it: a loop escape drill
- Check it worked
- Quick check
- What you can do now
The step-by-step, the prompt to copy, the video and the check-it-worked list, for every lesson in all 15 modules. GHS 399 ($40), paid once by mobile money or card, yours to keep.