Debugging With AI Help

The most common debugging move with AI is: paste the error message, paste the code around it, describe what you expected, and ask why. This works shockingly often because stack traces contain rich clues that the AI has seen a thousand times.

What to paste, in order

1. 1The exact error message, including the last few lines of the stack trace
2. 2The function that triggered it, with line numbers if possible
3. 3The input you gave, if any
4. 4A one-sentence summary of what you expected

A real-looking debug conversation

Me:
I'm getting this error in my Node.js script:
  TypeError: Cannot read properties of undefined (reading 'map')
      at loadUsers (index.js:14:22)

My code at line 14 is:
  const names = response.data.users.map(u => u.name);

I'm calling a REST API and expected an array of users.

AI:
The error means response.data.users is undefined at the moment you
call .map. Two likely causes:
1. The API returned an error shape, like { error: "unauthorized" },
   instead of { data: { users: [...] } }.
2. The response structure is different — maybe it's just { users: [...] }.

Add this log before line 14 to find out:
  console.log(JSON.stringify(response.data, null, 2));

When AI is great at debugging

• Syntax errors and typos
• Common library errors with famous stack traces
• Explaining what an unfamiliar error message means
• Turning a vague bug into a list of hypotheses

When AI is bad at debugging

• Bugs in logic that look correct line by line
• Issues caused by data it cannot see
• Bugs that only appear in production
• Timing and concurrency problems

The rubber-duck bonus

Sometimes explaining the bug to the AI in full sentences is enough to make you spot it yourself. That is called rubber-duck debugging and it is one of the best side effects of working with an AI partner.

“If you cannot explain the bug to the AI, you do not understand it yet.”

The big idea: AI is a great debugging partner when you paste the right information and ask why before fix. Shrink the problem, share the evidence, and guide the conversation toward the root cause.