"ICE" Kid Prisons

> In summary, after comprehensively considering the generalization performance...

I might be AI-brained but whenever I read the word "comprehensively" in a scientific paper, I get the feeling that it was written by AI.

"Of course, that doesn't mean we should ignore and stop working on the affordability of housing. But it does mean that at some point, those won't be our most pressing problems and humans do have a habit of fighting the last war, not the next one."

VC's think they're better-equipped to judge "pressing problems" than most and I think that's dumb.

Isn't that their job though? To find market opportunities for providing value by solving problems people have (or think) they have?

So on some level, it's literally their job to find solutions for problems.

There is no level on which it is their job to find solutions for problems. That is not a job description nor what they do. Their job is to guess which investments will earn them money.

Problems solving is accidental and done by other people. Also, their have zero incentives to care about social problems or about problems of people who are not rich.

I dunno, even though the authors address its use, making the task Tower of Hanoi doesn't meet the excitement of the title.

Especially since it's a recursive problem so each step is naturally broken up into subtasks. And the algorithm of what subtasks to break it up in to is public. This makes it much easier for it to get down to a case that the LLM can reliable solve.

I guess that the subtask decomposition of many (sub)problems is known and in the training distribution. How many real-world problems are resistant to divide-and-conquer? Presumably most/all of the unsolved mathematics conjectures. What else?

And yet the reverse paper was posted ad nauseam, covered by every news slop site, and overblown with really negative takes.

> ...is a skill that is in practice purely developed in order to do homework.

I would argue that it helps kids learn how to organize and formulate coherent thoughts and communicate with others. I'm sure it helps them do homework, too.

Feels like it would be easier to count the ills that _don't_ go back to this sort of thing.

I've been using for over 4 months and I think it's great - both Python and R. Plenty of DS features that make VS Code feel more competent as data exploration tool.