We are getting better at building things that almost nobody can explain.
That is the whole problem. Not that the tools are complicated. The problem is the gap between our ability to create a system and our ability to sustain it. We keep widening that gap, then acting surprised when the thing we shipped becomes a black box with a product wrapped around it.
I have started calling it the Incompetence Apocalypse. The name is a little loud. The pattern is not.
You, with a phone you cannot explain
The phone in your pocket is a fair test. It is a library, a map, a camera, and a second memory. Most of us could not describe how any of that actually works. We are users. That is fine for a phone.
It is less fine as a habit.
We have outsourced directions to GPS, recall to search, and idle time to the feed. Some research even suggests that the mere presence of a phone reduces available cognitive capacity — "brain drain." Whether or not that study is the last word, the lived version is familiar: we can access almost anything and understand almost nothing.
A person who only knows how to tap is easy to strand. The tool dies, the map disappears, the model is confidently wrong, and there is no second skill underneath.
The company that can ship but cannot stay
The same pattern shows up at work. We cut the corner to hit the date. We skip the docs. We let the design that "we'll clean up later" become the design.
Then the people who understood the system leave. The rest of us inherit a pile of behavior nobody can justify. Innovation turns into archaeology. You spend the quarter patching last year's velocity.
This is the company version of the same gap: excellent at creating, careless about keeping.
The few who still understand
Zoom out and the gap becomes political. When the systems that move money, news, and care are opaque, power collects around the small group that can still read them. A handful of companies are building the models that will sit under everyone else's work. Most of us will only ever be customers of that layer.
A world no one fully understands is not just inconvenient. A single bad deploy, a security hole, a cascading failure in something we treated as magic — those are now societal events. We are operating machines we could not rebuild from scratch.
What to do instead of panic
Keep the tools. Learn the mechanism.
If you use a thing every day, learn enough of it that you cannot be easily fooled by it. Read the docs. Build something small without the assistant. Fix one thing you used to file away as "the platform." Choose the slower path of understanding over the faster path of tapping.
On a team, that looks like the same rule we should apply to generated code: do not ship what you cannot repair. Write down how the system works. Spread the knowledge past the two people who "just know." Pay down the debt before it becomes the product.
The useful question is not whether you use the tools. It is whether you would still know what to do if they went away for a week.