My professor said to ignore generative AI for my final project, and I'm glad I didn't listen

For my final project at UC Berkeley last semester, Professor Davis insisted that using any generative AI for coding was 'cheating' and would lead to sloppy work. I went against his advice and used GitHub Copilot to help build a basic neural network for image recognition, which cut my setup time in half and let me focus on tuning the model. Now I'm seeing the same debate online, with some people saying tools like that make you lazy and others arguing they're essential for keeping up. Where do you all stand on using AI assistants for actual development work?

2 comments

2 Comments

ninasanchez1mo ago

My friend's Copilot code had a huge security hole he missed.

brookepark1mo ago

Wondering how you make sure the AI code is actually safe to use. Like @ninasanchez mentioned, that security hole is a real problem. Do you still check every single line Copilot writes, or do you just trust it after a while? Seems like the extra time saved might get wasted fixing hidden bugs later.