When I was nine years old, my older brother taught me to program in BASIC on an ancient (even at the time) Atari 600XL. It had a tape drive (as in, audio cassette tapes) instead of a disk drive, and I still have nightmares about having to retype an entire program because saving to cassette failed about as often as it succeeded. But I did it, because I loved programming.
Later on, my mother got an IBM PC Compatible that she used for writing. I was allowed one hour per day on it, so I spent the rest of my time writing QuickBasic and C code in longhand in a notebook, except when my mother shooed me outside to play baseball or run around. I did it because I loved programming.
As time went on, my skills increased, and the scope of my programming projects grew. I loved writing code, I loved solving tricky programming problems, tracking down mysterious bugs, creating application architectures that were beautiful and practical at the same time.
So when it came time to choose a career, software was the obvious choice.
As with all “dream jobs,” the more it becomes a job the less it remains a dream. The job is more about appeasing capricious managers and clients; managing requirements; choosing between poorly made, poorly documented third-party libraries; planning releases; navigating the trade-offs between code quality and delivery speed; setting up a functional Xcode environment; sitting through weekly planning snores; and a million non-programming tasks. There are still moments, here and there, when I can sit down in front of a blank screen and craft a piece of software I’m proud of. Those moments are the glittering diamonds in the asphalt of my career.
And the AI people want to take that away.
Sure, AI can write code. Sometimes AI can write pretty decent code. AI can write code that’s “good enough,” if you keep prompting it for long enough. But writing code is the fun part of the job. It’s literally the only reason I got into this field in the first place. AI doesn’t yet help much with the other parts, the annoying parts, the “job” parts, at least not much more than a good search engine.
Why would I outsource the glittering diamonds and keep the asphalt?
As in every AI-critical piece of writing, I need to include the usual disclaimers. Yes, AI is amazing. It’s a transformational yada yada yada. Copilot Code Review has caught some bugs early on that would otherwise have slipped through the cracks, so I’m grateful for that. It helps with repetitive tasks like writing unit tests or scanning long log files. Okay? I said it: AI isn’t all bad.
But if I can write better code than AI, and it’s code I enjoy writing, code I can be proud of, code where the process of writing it teaches me things I couldn’t have learned any other way, then that’s what I’m going to do, even if it takes longer than prompting ChatPig or Cloude.
That’s what keeps me going, even if it only encompasses a small, dwindling, percentage of my work hours.
I don’t see that changing any time soon, because despite the profession’s insecurity, despite dire warnings of obsolescence, despite the continuous threat of replacement by something cheaper, I still love programming.
