1) Reviewing *short* blocks of computer code for possible improvement

and (what is fundamentally the same action)

2) Suggesting corrections to grammar—sadly, the ALs are not as fond as am i of em dashes, random commas, neologisms, underused words and nonStandard capitalisation .

My first exposure to what could be termed "AI" was the computerized interpretation of EKG tracings (tho at the time was simply called just that "computer interpretation" ). This has steadily improved over the decades, at first it was quite lame, now is almost always correct.

What is fascinating though is when CI either misses something entirely, or (as more oft the case) makes a wrong call. I find it fascinating to try to understand why the CI made the error that it did. Generally, sometimes after more than a little effort, I am able to do so and thus learn more about EKG interpretation myself.

I can respect your fascination.
I translated more to the second conclusion right off, but could have walked away with the first if I wasn't so possessed.

I fear AI not for its power, but I worry about it's Black Holeness.
Going there has no escape.

I have little idea what the OP is about and no idea about all the alphabet soup, which severely limits its readability.

FWIW, I can cope with a lot of molecular biology, which is pretty much all alphabet soup, but I only post translations here.

AI doesn't terrify me except when it's used to track and restrict human interaction the way it has been used in China. It's just another thing which will be a good servant and a terrible master, like governments, corporations... and molecular biology.