In the spirit you propose—Did you verify this?

I notice that your comment history is all rapid-fire three-paragraph LLM responses. You do appear knowledgeable and respond quickly, but I've just dumped 10 minutes of my life into your attention in order to verify, parse, and filter through your responses.

I can't tell whether you're a person who thought about something. Therefore, I can't tell whether, for example, https://news.ycombinator.com/item?id=47393311 is an analysis I should take seriously (as I might, if it were spoken from experience) or just Markov-chain, Reddit-trained hypothetical fluff.

How can we increase the friction to presumptively exclude you, but provide accommodation if, for example, you're more comfortable in your native language and using the LLM mainly to bring your English writing to a level consistent with your personal expertise?

Not OP, and I’ve been looking at more meta-analyses than RCTs, but:

* Lei et al 2024 metaanalysis, generally positive associations but no causal evidence for psychosocial effects, especially when the social media use is family-directed - https://www.cambridge.org/core/journals/international-psycho...

* Balki et al 2022, metaanalysis, same thing: good for reinforcing existing real-human social connections, overcoming barriers to/increasing regularity and frequency of contact, acquiring access to resources; isolating to use outside that context; regular and frequent contact much more effective than occasional or episodic: https://aging.jmir.org/2022/4/e40125/

My overall impression seems to be what common sense tells us: to the extent it lets you overcome aging-related obstacles to interacting with real people, great; to the extent it’s brain rot, it’s isolating.

If anybody wants to do something with instagran.org, I know somebody who would be willing to part with it for a good cause…

The last time I chose to watch a movie in a theater instead of the comfort of my home, I went for the raucous audience aspect of the experience.

I love, love that. And if even one of my weird little side projects—including the ones I build with AI-powered tools—connects with a young person like that, I’ll be satisfied.

To me it’s not the “how” so much as the “what,” though.

I can only speak to my own experience with that sort of thing, but how much of what moved you was the invisible authorial hand behind the tutorials—deciding what’s fun to them to write about, and how to talk about it in a way that clicked with a young you?

I guess, what’s the difference between that website, the official docs for the language in question, the formal spec for the language, the .h files themselves that mechanically define the engine that compiles the language, a big pile of examples of working code in the language…

For that matter what’s the difference between what’s fun to do in the language and what’s boring?

I would grant that LLM tech would probably shine at the grunt work part of “please translate these docs into a grade-school-pitched, engaging, example-driven tutorial website; make it dinosaur themed.” But equally it could pitch it for a billion other audiences, and most will not bear fruit without guidance and refinement. And LLM frontends are already same-y, what will distinguish it to your young eyes? Knowing (or finding out) what’s worth doing is the tricky part—and that’s hard to separate from the humans on the receiving end.

I think about when tiktok made it “easy” to recut songs and memes, and to do basic compositing effects. The “how” required specialized software and serious skill for a long time, then suddenly it didn’t. But when it comes to the “what,” there are still people who are good at using the tools and ones who are bad—ones who make good content and ones who make bad ones… and the difference seems to cleave along the normal human lines: innate talent + practice + persistence.

As the old saw goes, contemporary art: “But I could do that!” “Yeah, but you didn’t.”

Alternatively, as the fox said, “C’est le temps que tu as perdu pour ta rose qui fait ta rose si importante”…