It's a logical fallacy that just because some technology experienced some period of exponential growth, all technology will always experience constant exponential growth.
There are plenty of counter-examples to the scaling of computers that occurred from the 1970s-2010s.
We thought that humans would be traveling the stars, or at least the solar system, after the space race of the 1960s, but we ended up stuck orbiting the earth.
Going back further, little has changed daily life more than technologies like indoor plumbing and electric lighting did in the late 19th century.
The ancient Romans came up with technologies like concrete that were then lost for hundreds of years.
"Progress" moves in fits and starts. It is the furthest thing from inevitable.
Most growth is actually logistic. An S shaped curve that starts exponential but slows down rapidly as it reaches some asymptote. In fact basically everything we see as exponential in the real world is logistic.
* Code using sympy to generate math problems testing different skills for students, with difficulty values affecting what kinds of things are selected, and various transforms to problems possible (e.g. having to solve for z+4 of 4a+b instead of x) to test different subskills
(On this part, the LLM did pretty well. The code was correct after a couple of quick iterations, and the base classes and end-use interfaces are correct. There's a few things in the middle that are unnecessarily "superstitious" and check for conditions that can't happen, and so I need to work with the LLM to clean it up.
* Code to use IRT to estimate the probability that students have each skill and to request problems with appropriate combinations of skills and difficulties for each student.
(This was somewhat garbage. Good database & backend, but the interface to use it was not nice and it kind of contaminated things).
* Code to recognize QR codes in the corners of worksheet, find answer boxes, and feed the image to ChatGPT to determine whether the scribble in the box is the answer in the correct form.
(This was 100%, first time. I adjusted the prompt it chose to better clarify my intent in borderline cases).
The output was, overall, pretty similar to what I'd get from a junior engineer under my supervision-- a bit wacky in places that aren't quite worth fixing, a little bit of technical debt, a couple of things more clever that I didn't expect myself, etc. But I did all of this in three hours and $12 expended.
The total time supervising it was probably similar to the amount of time spent supervising the junior engineer... but the LLM turns things around quick enough that I don't need to context switch.