I’m reviewing developer portfolios and need to figure out which GitHub projects are authentic versus AI-generated. Many job applicants include their GitHub profiles, but I suspect some repositories might be completely made by tools like ChatGPT or similar AI assistants. I want to screen these before scheduling interviews since it’s easier to catch during face-to-face conversations by asking about their coding choices. Are there any red flags or patterns I should look for when browsing through someone’s code repositories? What techniques work best for spotting artificial intelligence generated projects during the initial review process?
Look at the documentation first. AI repos usually have these ridiculously polished README files - perfect grammar, every section filled out, way too clean for most personal projects. Real developers? Their docs are messier, with typos and half-finished sections. Check the coding patterns too. AI code is weirdly consistent - same formatting and spacing everywhere. Humans aren’t that perfect. We change our style between files, especially stuff we wrote months apart. Error handling’s a dead giveaway. AI loves implementing textbook exception handling everywhere, but real projects are inconsistent. Some parts have solid error handling, others barely have any.
the dead giveaway for me? commits that are way too clean. every message follows the exact same format, and you never see those “fix typo” or “damn forgot semicolon” commits that we all make. also watch for variable names - ai loves making everything super descriptive and verbose when real devs usually keep it shorter.
use git blame - it’s a dead giveaway. ai repos usually show one person committing massive chunks of code all on the same day. real devs leave messy commit histories where files get touched multiple times, refactored, and improved bit by bit over months.
Check the commit history first - that’s your biggest tell. AI repos pop up overnight with huge initial commits that dump entire, finished apps. Real devs build piece by piece. You’ll see projects start simple and add features over weeks or months. Look at the comments too. AI comments everything - every function, every block. Most devs barely comment unless they have to. The dependencies are another giveaway. AI loves throwing in popular libraries even when you don’t need them. You’ll get bloated package files full of unnecessary stuff. Lastly, check how the code’s organized. AI makes these perfect textbook folder structures and follows every design pattern to the letter. Real developers cut corners and do weird stuff based on what they’re actually dealing with.
Here’s something nobody’s mentioned yet - check the testing patterns. AI repos are weird about tests. They either have zero tests or these insanely perfect test suites that cover literally every edge case. Real devs? Our testing is messy. We write tests for the hard stuff but skip the obvious parts. Or we start strong then get lazy halfway through. Also look at issues and branches. Real repos have open issues sitting around, dead feature branches nobody cleaned up, and merge conflicts that got fixed with duct tape. AI-generated repos are too clean - they don’t show the actual struggle of building something.
This topic was automatically closed 24 hours after the last reply. New replies are no longer allowed.