Tried GPT-4o for coding and it failed harder than I expected

I spent last weekend using GPT-4o to help me debug a Python script for work, thinking it would save me time like everyone says. Instead, it kept suggesting the same broken loop fix three times even after I told it the error. Learned that for specific logic bugs, I still have to trace through line by line myself. Has anyone else found a specific type of problem where AI just can't help at all?

2 comments

2 Comments

brian_murray29d ago

Fair point, but it did help me find syntax errors faster than usual.

sanchez.robin29d ago

Yeah "faster than usual" is the part I keep coming back to. I get that it saved you some time but for me it felt like I was fighting the suggestions more than letting them help. The AI kept guessing wrong about what I wanted and it ended up slowing me down instead. Plus I like finding my own mistakes sometimes, it helps me learn the language better in the long run.