Yeah most of the advance assumes you have the data ready at hand and just need to phrase the cards right, get the number of words right. Whereas for conceptual domains the biggest problem is: how do I encode this as question-answer pairs at all? What I want to read more of is people sitting down and writing in the first-person perspective how they go about it, like Michael Nielsen does here: https://cognitivemedium.com/srs-mathematics
> All that said, once again, that stand on a very large set of "if"s.
Yeah, probably I should have listed some more premises, i.e. that corporations maximize profits, the state maximizes power and security (I don't entirely buy the Realist framework, but if you want to predict how things work out "in the limit", rather than tomorrow, it seems alright?).
And naturally every "therefore" becomes weaker the further out into the future you try to predict.
Sometimes you can be in a situation where every actor taking locally-rational actions leads to globally catastrophic outcomes. It would be easy to argue I think that the July Crisis was like this: if you look at the incentives of each player, they had many reasons to do what they did, and nobody can perfectly what all other players will do, or what the future holds.
Doesn't this prove my point? In feudal Japanese society, wealthy merchants were lower status than poor samurai, i.e., they rich could not buy political power. "The wealthy" and "the ruling class" are not always the same group of people.
For every person using AI as a personal tutor there's a hundred people using it to produce AI slop articles, slop scientific papers, slop short stories, to checck out of living and let the AI do all their thinking and writing and creation. Voluntary disempowerment is already here!
A corporation that is fully staffed by AIs and only retains humans for legal reasons (as directors, for liability etc.) still needs money to coordinate. You need to pay for inputs, you need to pay to run the AIs, which consumes resources. Why would costs go to zero? The market is still a valuable tool for allocating resources even if no market actors are human.
Tangentially, this is why (despite being an extremist for privacy) I've somewhat soured on Internet anonymity. Very few people are using their anon status to further discourse, i.e. like Publius. Mostly it's just people who want to sever the reputational thread between their output and their person.
I think it's fine to be pseudonymous, like Gwern, because while your government identity is protected, the pseudonym is a persistent store of reputation across time.
Seems kinda silly to name such a huge project after a meme that will be stale in a few years, but then, this was likely all generated by AI, so it doesn't matter.
EDIT: Expanding on this a bit, because I want this comment to be more productive and less old-man-yells-at-cloud.
When I saw the link I got curious. It reminds me of the old skdb[0] project and of Open Source Ecology[1]. The idea is cool: a DAG of civilization from David Gingery[2] basic tools to jet planes and turbofans and rocket engines! Imagine that!
If it was real, it would be world-changing. The creator would be a personal hero of mine. But it's not real. It's a vague suggestion of the real thing.
Who is the creator? No-one knows. These vibeslop websites never have an About page, or contact info. No-one is putting their reputation on the line, regarding the quality or accuracy of the content.
If the About page said who made it, i.e. if someone was putting their reputation on the line, I might be more receptive. But the website has enough LLM design tics to make me suspicious.
It's sad. I come to Hacker News to see cool stuff and when I click on a link and see something obviously put together by an LLM I feel like I've been tricked :(
But the problem is people are not just delegating formulaic procedural prose to AI. They're using AI to write entire scientific papers, so now reviewers have to use Pangram[0] to screen submissions. Literary magazines have the same problem[1]. Maybe those people should know that their behaviour is bad.
> This is a very good essay when you get past the arrogant tone evident especially at the beginning (is that a form of engagement bait?…)
We have reached the point where you have college professors who defend[0] using AI to write scientific papers (in a seemingly AI-written tweet). Everywhere I go online, I see spam written with the exact same voice. Scientific journals and literary magazines are inundated with AI-written submissions. Software projects have shut contributions because maintainers are tired of reading AI-written slop pull requests.
What's the right tone to take here? "Please stop defecting"? "I wish you would kindly stop ruining the commons"? I don't know. Maybe, if we raise the reputational cost of slop, we get less of it.
But writing, calculators, search engines etc. are specific tools, you drop a skill to gain effectiveness in another. While AI is a general tool whose builders intend it to be able to perform all cognitive work. Past a certain point, what is left of the human?
Website: https://borretti.me/
Email: [email protected]
GitHub: https://github.com/eudoxia0
Twitter: https://twitter.com/zetalyrae