Who will be our "mechanics" in 2099?

I saw a post this morning talking about Google's Project IDX (https://idx.dev/) and it got me thinking. So, now you get to suffer...(enjoy?). Here goes:

There's an intractable truth that we need to come to terms with: the demand for software engineers, prior to the pandemic, outpaced human population 4-to-1 (CAGR). With the eruption of AI, it's not unreasonable to expect that delta to grow. There are only so many engineers out there, or even people interested in performing the work of an engineer - perhaps even capable of doing it (assuming that IQ is real and there is a certain level of abstraction capabilities required that maps sufficiently to IQ to state that there is a minimum level required).

Tools like this are a great idea for the following 2 reasons: 1) they reduce the level of effort for each project/task for existing developers, opening up their bandwidth; and 2) they lower the barrier to entry to software engineering to potentially deal with the delta mentioned before. Both of these benefits are theoretical as we really haven't seen the impact, yet. And it's unclear how well these tools will adapt to a changing digital world/landscape, nor how engineers themselves will adapt to them and adopt them.

But let's assume that things work out as expected. It makes me think of cars. Cars used to be highly mechanical and their maintenance was something that a lot of people could do themselves, maybe even taking pride in it and considering it a fundamental pillar of what it meant to be a [insert gender here]. But as modern life started pulling more and more people away from manual labor and into knowledge work, professions, and office-dwelling, we also saw a shift from manual transmissions to automatic transmissions. Automatic transmissions are simpler for the driver and more comfortable in stop-and-go traffic (I'd argue manuals are nicer to drive on open highways, but that's personal preference). With this, we saw an alienation of drivers from the automobile as a machine. No longer was the car an extension of the driver in a very true and corporeal sense, it was a machine that the driver only needed to operate.

Now, I am a big fan of the backside warmers during Midwest winters and reverse cameras/assisted parking have made my life much easier living in a busy metropolis. But we've gotten to the point now that the car owner/operator can no longer perform any maintenance nor can they fix the machine when something goes wrong. I'd posit that most of us, myself included, couldn't even name the parts of an engine nor accurately describe what they do - much less with electric vehicles.

Even at the level of mechanics, actually servicing a car today starts with a diagnostic test, a check of the car's own computer to identify issues. Yes, the actual work of resolving the issue is still performed by the mechanic, at least I think it is and they claim it is on the bill. Each diagnostic is calibrated/configured for the make and model in question and some shops specialize in one specific carmaker.

With every now tool or technology, there is an implicit trade-off. The question is: what is that trade-off and what is the impact, iterated across time, of accepting that trade-off? In this case, we're reducing the cognitive load and level of effort in creating applications. We're moving from a manual transmission to an automatic transmission.

It's perfectly logical, or potentially ridiculous, to assume that the sequence of alienation and the loss of explicit, tangible knowledge that has occurred with the automobile will replicate itself in the digital space. As we make it easier to operate the tools to make an application, the knowledge of the underlying pieces of the puzzle that make the applications work or can make the applications fail will gradually fade away.

So, in 2099, who will be our "mechanics"? Who will be the people who still understand the working moving pieces and can fix them when things go awry? How will we recognize that we're running out of mechanics? How will we go about training more, assuming there will be a much higher population and a much more pronounced delta between demand and human population growth, which will, presumably, be supplemented or offset by increasingly complex abstractions designed to further increase each person's capacity and reduce the barrier to entry for more and more people?