How do you know if you should accept an AI code suggestion? I’m a software engineer with 30 years of experience and I’ve built AI coding tools myself, including AppMap and OpenTips. In this vide, I break down how to analyze AI code suggestions to get the best results. I have a lot of experience doing this and I can quickly simplify this process down to its essential steps.
This video uses an example of an OpenTips AI suggestion for an iOS app. I discuss the two criteria that I use to decide if I want to accept the tip:
Is it needed?
1) Does it fill an essential gap in the user experience?
2) Does it fix a security problem?
Is the implementation correct?
1) Do I understand it? If not, is it worth the research?
2) Is it adding a new tool or library? It has to add something that would be hard to do myself.
3) Does it follow the conventions of the existing code?
4) Will it make the code simpler without breaking a lot of tests?
Top comments (0)