I first heard this quote years ago and it has stuck with me ever since. It sounds like a joke at first, but the more I sat with it, the more I realized it explains half of tech history.
If we think about it, nobody ran out of stones. Stones are still lying around everywhere. People just figured out bronze, and then iron, and suddenly stone was no longer the best tool for the job. It didn't disappear. It just got left behind.
Once you see that pattern, you see it everywhere. Film cameras didn't die because we ran out of film. We didn't move to the cloud because the world ran out of physical servers. And gasoline cars won't wait around for the last drop of oil before something better takes over.
The one I keep thinking about, because it's my job, is software.
We're not going to change the way we build software because we run out of programmers. But AI might change it anyway, simply because some of the habits we've built up over the last 20 or 30 years stop being the most efficient way to get things done. Not wrong. Not broken. Just no longer the best tool for the job.
I already see it in my own day to day. I used to code with a dozen tabs open: Stack Overflow threads, Reddit discussions, Substack posts, GitHub repos I was reading for reference, API docs for whatever service I was integrating. Now I describe what I want in plain English and an assistant drafts the function, and my job shifts to reading it critically, pushing back, and shaping it. The typing was never the hard part, but we built a whole workflow around it anyway.
Agents take it a step further. Instead of a single autocomplete suggestion, I can hand one a task like rename this concept across the codebase, update the tests, and tell me what broke, and it works through the steps on its own: reading files, running the test suite, fixing its own mistakes. The first draft of a migration, a test suite for a legacy module, a prototype of an idea I would have parked for a quieter week. These used to cost days. Now they cost a code review.
Even debugging is changing shape. Pasting a stack trace into a chat and getting a plausible explanation in seconds doesn't replace understanding the system, but it replaces the hour of narrowing down where to look. Same with boilerplate, glue code, config files, one-off scripts. Nobody is going to miss writing those by hand, the same way nobody missed chiseling stone.
None of this means the old way stopped working. Hand-written code still compiles.
The old thing can still work fine. There can be plenty of it left. It becomes obsolete anyway.
So now, whenever someone tells me why change it? The current thing works perfectly fine, I think about the Stone Age. It has a pretty good comeback:
We still have stones too.