"haphazard in their design"? There is no 'design', more like 'haphazard in their nature" or rather "haphazard as a result of the process of evolution" but yeah understanding how it work requires a lot of such experiments.
However, the AI better be less messy. I hope we can do better here with the 'design'. But in any case it's going to be an Alien Intelligence rather than Artificial (human) Intelligence.

Though one thing I really miss from Perl (in Python) is regex being part of the language rather than a library. This was very useful in Perl.

Ok, "killed" is too strong, fine, so rather "displaced"?

But "rightfully" was a tribute the Perl's syntax being heavily punctuation based and with rampant plurality - the same thing can be expressed multiple ways. This did not help withe ease of use and maintainability, hence "rightfully". (At may old age I am starting to favor "verbosity" over "expressiveness". Perl always looked like "assembly" of scripting languages.)

TL;DW; but did it explain why particularly Python become so popular? It killed Perl (rightfully) and almost killed Javascript (Node remains) but did not become a front-end scripting.

me thinking the same, up vote parent!

I am glad the *president* is already on it! /s

How about running the answer through another model for sanity checks?

And... corn grows fine around windmill towers.

Without reading a paper, I'd assume both teacher and student are the same models just tuned with prompts?
Still, if the information transferred between teacher and student somehow, without direct references to T conveyed T, it's interesting.

This may have been true 3 years ago. But now, observing agentic AI that helps me produce software I am less sure it isn't so. When it goes through the problem solving, its steps, that is describes to me, are similar to what'd go through. It is a very 'reasonable' approach to solving problems I gave it.

LLMs "learn" (ie tune multidimensional function's parameters to produce correlated output fro given stimulus). There is no hardcoding anything as in adding another "if my action does not cause harm to human, then do". You could try to run multiple stimulus data into learning data that would amount to "do not kill humans" to slant/tune this correlation but there will not be a 100% proof hard coding.

Indeed, however comunism never existed anywhere, even in China. At best, they are totalitarian, and hopfully leaning benevolent today. But I see that since USofA is leaving the stage, there is indeed a vaccum.

Last 3 are questionable. Really I'd expect FORTRAN (weather modeling), Matlab/Simulink, Rust, R, Lisp and VHDL/Verilog there higher than these three.
  Is there really so much legacy apps still in active maintenance?

It's time for some company to create a local-cloud product family configurable by a common Joe. Something one can setup locally, like a wifi router, configure with vpn and public dns and have it accessible w/o _their_ cloud.

Business model....eh... that's harder - they all like subscriptions, right?

This is interesting. It feels like a trip in a Model T across the US in the early 20th century...
Though chargers with a diesel genset is abomination for recharging EVs...