I wrote a science fiction novel because I love the genre, I love exploring how technology might evolve, and sometimes I go to sleep just making up stories like this in my head.
The novel is called "An Advance in Time." I'm not going to pitch it here. What I want to talk about is why this stuff fascinates me.
We're Bad at Predicting the Future
I heard it said once that humans are good at predicting linear progress and horrible at predicting exponential progress. That tracks with what I've observed.
We can imagine next year pretty well. Ten years from now? We either assume things will be mostly the same, or we imagine flying cars and miss the actual changes entirely.
Often things change slowly, then all at once. Electricity was one of those "all at once" moments. So were computers. And the internet. Each of these felt sudden when they arrived, even though the groundwork had been laid for years.
And yet, the speed of those transitions, while fast on the historical timeline, is glacial compared to what we're living through right now with AI.
The AI Moment
I feel like I'm probably ahead of about 95% of the population in my use of certain parts of AI. Used properly, it makes you 10x more effective. Used carelessly, it makes you lazy.
There's a fascinating study from METR that looked at experienced open-source developers using AI coding tools. Before the study, developers predicted AI would reduce their task completion time by 24%. After the study, they still believed it had helped, estimating a 20% improvement.
The actual result? AI increased completion time by 19%. It made them slower.
The researchers found that experienced developers, working in their own projects with lots of context the AI didn't have, spent significant time retrofitting AI outputs to fit their actual needs and debugging the results. The AI was generating plausible code that didn't quite fit the situation.
When the AI tools were removed, productivity dropped further and took time to recover. The developers had started relying on the tools even though the tools weren't actually helping.
AI can make you feel productive while actually making you less effective. The skill isn't using AI. The skill is knowing when to use it and when to think for yourself.
My Prediction for the Next 2-5 Years
Software (and firmware) will get very, very cheap to build. Competition will increase, and more features and functionality will be added to everything.
It's going to be both a rough and exciting time for engineers.
Big corporations will do as they always do: try to cut costs where they can. For them, that might actually be the right call. They're notoriously inefficient and bloated. Often they'd hire engineers just to make sure their competition didn't get them. Those days are probably over.
Companies that aren't bloated might be tempted to do the same. That would be a mistake.
If you think you get a 4x ROI on an engineer now, and you're happy with them, and you can make them several times more effective and productive, how many should you want to hire?
The answer in my head is "as many as you can possibly afford."
If margins are squeezed, or sales are down, or any number of other factors make that impossible, then sure, that makes sense. But for business owners like me: think of the headstart you could get on your competition. How many products you could ship if you do this right.
It won't be easy. It's a discovery process to figure out what workflow actually adds productivity. You've got to do your part to remove bureaucracy and processes that no longer make sense. You have to be okay with breaking things and not getting things perfect on the first try. And you have to be in an industry where that's okay.
You probably don't want your pacemaker AI-coded for the next couple decades.
But having AI do code reviews automatically? That's a no-brainer with no downside. (Check out Greptile. It's cheap and amazing.)
Did I Waste My Time?
Writing a novel was both fun and time consuming. On some level, it was probably a waste. But I don't regret it. And I think everyone should have a creative outlet that forces them to ask, "what if?"
If this technology exists, what else has to be true? How do people actually adapt? What stays the same even when everything changes?
I don't know what the future holds. But whatever the future holds, I'm convinced it will go much better for those who take the time to dream of the wonders that the future could hold, rather than living in the fears of the present day.