Practise the language developers use when discussing AI-assisted code review: flagging issues, surfacing optimizations, and writing clear PR comments that reference AI tool output.
0 / 5 completed
1 / 5
A PR comment reads: 'The ___ a potential bug on line 42 — worth checking the null path.' What did the AI tool do?
'AI flagged a potential bug' means the AI review tool marked a line or block as suspicious. It does not mean the bug is confirmed — the developer must investigate and decide whether it is a real issue or a false positive.
2 / 5
In a PR comment you write: 'Copilot ___ an optimization here — caching the result would save a DB call per request.'
'Copilot suggested an optimization' is standard PR language for sharing an AI-proposed improvement. Including the source (Copilot) keeps the review transparent — reviewers know the idea came from AI, not the author.
3 / 5
The AI review tool raised ten warnings, but after investigation you mark eight as ___. What does this term mean?
False positives are AI warnings that turn out to be incorrect. A high false-positive rate reduces trust in the tool and causes alert fatigue — developers start ignoring warnings, including real ones.
4 / 5
A team member asks: 'Should we ___ the AI suggestion or follow it?' They want to know whether to disregard the tool's advice.
A human override means a developer explicitly decides the AI suggestion is wrong or not applicable in this context. Overrides should be documented (e.g., in a PR comment) so future reviewers understand why the AI advice was not followed.
5 / 5
Your PR comment states: 'Leaving this note for transparency — ___ this pattern as a potential race condition, but I've reviewed the execution model and it cannot occur here.'
Attributing a flag to the AI tool in a PR comment is good practice — it explains the origin of the concern and shows that the developer reviewed it rather than dismissing it without thought.