If your web app is doing something that's never been done before, which requires the application of scientific principles and so on, like... translating baby babbles into English, then it's still technology, but that would not be because it's a web app.
Vonnegut wrote a course for eng. lit creative writing on the 7 canonical plotlines of all stories. ever. I thought "goodness whats the point in writing now" but it hasn't stopped people making good entertaining fiction
All code is variations over abstraction, synthesis, loops, decision forks, recursion (tail or not) and all data structures iterate back over lists and arrays, over trees and DAGs.
"never been done before" is pretty rare anywhere.
But I do think it's a bad way of evaluating this. Personally, I would say that a more important distinction between tech and "tech" is the degree to which the technology makes or breaks the business.
For example, for Uber, the tech is an enabler, but the business is made or broken by the cabs, the vetting, the pricing, the number of cabs they can hire, the salaries and benefits they can avoid etc. Sure, the tech has to work, but there is no reason that Uber would be falling behind if their tech went into maintenance mode today - even if their competitors kept investing in new tech. So, Uber is a "tech" company in my book (even though they have spent inordinate amounts of money on the tech before).
On the other hand, a company like Cisco must make ever better routers or Arista or F5 or whoever will quickly steal their lunch. So Cisco is a tech company, at least on the NEM side.
An app doesn’t have to be cutting edge in order to clear some arbitrary bar of what constitutes “tech”.
That's a difficult position to engage with constructively.
Where is the tech then?
I do think there is tech and non-tech. But it gets very like pornography and the Meese commission: I know it when I see it.
What I find hard, is how many people writing for the REPL or compiled languages, dismiss what is done in CSS or in JS outside of webasm. or PHP. The arrogance is the exclusionary behaviour.
Carmack is very smart. Had he been born 40 years earlier he might well have been the one to work out how to optimise track to head delay on a drum memory, or hacked delay lines, or even have done something amazing in LSI before VLSI. But then some non-carmack doing bitblt optimisations to make games happen would have been wondering "am I a real tech or not"
(btw I think GPT-3 Adherents are smoking strong stuff. I don't expect anyone to write informally stated natural language of intent and have proscriptive, defined code come out the other side, but you know this is a strawman argument too)
I graduated in 82. Because I did not do the Mudge VLSI Course I was held in somewhat contempt by my peers who did. I did Pascal. I was held somewhat in contempt by the Fortran NAG library maintenance squad. We who ran the boxes, and wrote controllers for the plotter were a bit contemptuous of the ones coding for the BBC micro and I was completely distainful of the guy coding .BAT files. None of this is "good" behaviour because we were all doing "tech" But the guys in the engineering labs doing Hypersonic wind tunnels were sometimes a bit elitist about us in s/w ... Over in the school for the logical foundations in computer science they were already writing abstract meta-code which was as far removed from an instruction set as its possible to be: theories of computation, proof systems, you name it.
Physicists and Engineers have been arguing for centuries about the difference between their disciplines.
Well, the fundamental problem with this is that you just made the programmer even more powerful. A single programmer is now as productive as a thousand programmers. You still want to have a programmer because you want someone who understands what the system is actually doing right now and keeps it running. You also need someone who understands the limitations of the software generator. You can't just say "please create software that solves the halting problem". You also need someone who understands how to optimize software. You can easily ask for something that is possible but it could cost you millions of AWS credits.
There will always be someone who has intimate knowledge of the tool he is using and can charge appropriately for it. The more powerful the tool the more attractive the employee. Compare an engineer drafting on paper vs one doing it in CAD.
I can't say much about if and when we're worse at everything than AI, but we're not there yet. At least for now, there are still new opportunities just as open to aspiring techies now as web apps were 15 years ago or PC games were 30 years ago.
Web apps are sufficiently turing-complete that the arrogance of "thats not tech" is just wrong.
Carmack wouldn't write asm for a device now, in all cases. He might do Verilog for an FPGA but he might be doing something else.