-
Alex Bunardzic
Digital Exprt • 6K followers
I've examined numerous code bases throughout my career. I've seen some horrendous code, bowl of spaghetti code, Big Ball of Mud systems architecture, repos with few or even zero tests, etc. And yet, all those code bases came to be via some code review process. Someone looked at that code and approved it. No problem. But when it comes to code generated by AI, suddenly the bar is raised super high and even the slightest sloppiness gets triumphantly declared as, "see, I told you -- AI sucks!" I find such double standards disingenuous. Let's call a spade a spade. Most of the code found out in the wild today is subpar. And yet, somehow we hobble with all those half-baked systems. From what I've seen so far, AI-generated code is in no way inferior to the typical code found in many repos. Therefore, AI is not going to create catastrophes no more than sloppy human design and coding has created thus far. If anything, AI has the potential to improve the batting average of the good, reliable code that is safe to improve. Contrary to the alarmist reports, the sky is not falling.