I turned 59 this year, which in my mind is almost old (but younger minds might disagree). I’ve been fortunate to witness many big shifts in CompSci, including the rise of PCs, the advent of the web, the switch to object oriented programming, and the mobile explosion.

But all of these pale against what I’ve witnessed with AI this year.

I had a three-month sabbatical from September to December. Prior to my sabbatical, I could use AI for basic programming tasks, but it wasn’t capable of heavy lifting without a lot of assistance. When I returned to work in December, everything had changed.

In those three months, AI became capable of carrying out large requests, reviewing PRs, and planning complex tasks. My job changed from writing code to reviewing code created by AI. I switched from using an IDE to using Conductor to manage multiple AI tasks in parallel.

I’ve read some concerns from fellow aging developers that AI will replace us, but at least for now it seems more likely that AI will replace the need for junior developers. We need experienced developers to review AI generated code (again, at least for now). There have been numerous times I’ve had to correct AI code that contained serious flaws.

But increasingly these flaws aren’t directly found by me, but instead found by AI after I raise questions. I ask AI if its approach is valid, and it comes back and basically says, “Oh, good question, you’re right that there are unwanted side effects that need to be addressed.” And then it fixes them.

At some point I won’t need to raise these concerns – AI will do it by itself – but I’m crossing my fingers that my experience will be needed for several years before I retire.