By skyforbes The task we provided: build a simple scheduling system that syncs data from external data sources. One hour time limit, clear requirements breakdown. We weren't looking for perfect answers or even a working solution but wanted to see how they approach the problem.
What I'm seeing from recent grads (sample of 6 so far):
They'll paste the entire problem into ChatGPT, get a semi-working codebase back, then completely freeze when asked to fix a bug or explain a design choice. They attempt to fix the code with prompts like "refactor the code" or "fix the scheduling sync" without providing the AI with useful context.
The most peculiar thing I find is that they'll spend 15 mins re-reading the requirements 3-4 times instead of just asking the AI to explain it.
Not sure if this is a gap in how fresh grads are learning to use AI? Am hoping we'll see better results from other candidates.
Anyone else seeing this in hiring?
