> They’re writing TypeScript that compiles to JavaScript that runs in a V8 engine written in C++ that’s making system calls to an OS kernel that’s scheduling threads across cores they’ve never thought about, hitting RAM through a memory controller with caching layers they couldn’t diagram, all while npm pulls in 400 packages they’ve never read a line of.

But sure. AI is the moment they lost track of what’s happening.

I feel this is conflating different things. Yes, the abstraction tower was massive already before, but at least the abstractions were mostly well-defined and understandable through interfaces: even if you don't understand the intricacies of your storage device, driver and kernel, you can usually get a quite reliable and predictable mental representation how files work. Same goes for network protocols, higher-level programming languages or the web platform.

Sure, there are edge cases where the abstraction breaks down and you have to get into the lower levels, but those situations are the exception, not the norm.

With AI, there is no clearly defined interface, and no one really knows what (precise) input a given output will produce. Or maybe to put it better, the interface is human language and your mental representation is the one you have talking to a human - which is far more vague than previous technical abstractions.

On the bright side, at least we (still) have the intermediate layer of generated code to reason about, which offsets the unpredictability a bit.