js8

AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape

I don't know what this guy thinks AI is, but this strikes me as delusional.

In my view, AI (LLM) is two things mixed together:

1. A reasoning engine on top of relatively rich fuzzy modal logic, implemented through variety of rules, which implement very common concepts.

2. A huge dictionary of words defined (with lot of detail) in the said logic, together with many known facts about them. Maybe bigger than Wikipedia.

Now, how on Earth do you want to gatekeep either of this? You can't gatekeep the 1st, logic of common sense, that's almost as difficult as gatekeeping a Turing machine (a concept of a computer). And gatekeeping the 2nd is ridiculous too, as it was built mostly from already published sources like a giant Wikipedia.

If anything, the opposite, to gatekeep AI is actually dystopian. It would mean end not only to right to compute, but also end of right to scientific knowledge.

(And I think, honestly, Chinese understand this. Trying to control-export AI makes as much sense as trying to control-export an English dictionary.)

I found it funny that Fable reinterpreted the Starry Night as "an oil-well in a desert", while Grok went for a "Stalin statue in a prison camp" angle.

People tried to distill a lot of things.. even oil.

Yes, I think this idea, that it should be "magical", is what makes it feel scummy. (Apparently I am not alone https://news.ycombinator.com/item?id=48988475). It makes AI providers sound like snake oil salesmen, and rightfully so.

Meanwhile, technological and engineering (STEM) progress have always been made by emphasizing externalization of the deductions (as opposed to reference to an opaque expert judgement) and reproducibility of experimental results.

I would even call the frontier AI labs anti-scientific. We need to understand how inference is done to avoid mistakes, not rely on intuition, even if the intuition is enclosed in a reproducible machine. The idea that AI should be this closed is a return to pre-scientific days.

Even if they published the conversation, Anthropic (and likely other closed model publisher) no longer provide logs of the actual thinking process.

I more and more see LLMs as a kind of scam; not useless, but really just a big database of fuzzy facts with some Prolog on top as rediscovered by the learning algorithm. Most likely could be made much cheaper to run, were humans allowed to actually inspect the algorithm.

I really like the cabin from Sol. It could be an art piece.

I think you're correct, I'll just point out that compilers are also chaotic today and it is often accepted. A small change to source program can lead to big change in performance. It's especially apparent in high-level and declarative languages - SQL or anything with GC.

I still think it is (in the source-code generating mode) a compiler, just like LiquidHaskell is. (I haven't used LH but it essentially can automatically supply functions based on the type conditions you specify.)

In my view, reasoning LLMs have learned a large number of close to true sentences of "logic" in natural language. These rules are not consistent (unlike typing rules of LH that allow creation of provably correct programs), but they work in practice better than LH. Essentially, they represent an unsound formal specification of natural language, together with many useful (almost)tautologies that helps solving constraint problems (just like LH does, but again unlike LLM, provably correctly).

LLMs proved that human language can be formalized reasonably close to soundness. I believe we can have a sound formalization of human language, and that kind of formal language will be once superior to LLMs.

It will also not be as complicated as LLMs and require only fraction of compute to use.

I would go further, it's tied to protestant work ethic. And there always was an alternative - gradual reduction of work hours rather than unemployment.

No, you missed what makes AI bad in our culture. It doesn't just take your work, by your choice. It also takes your self-worth (because our culture is based on working being worthy) and for many people also a job (which is a material representation of that cultural value).

The problem is the idea of the job. The idea that you're worthy of existence only if some other person certifies you're worthy of existence, based on how much they have been certified themselves.

Of course this converges to a system, where this certification is the most sought thing. And if you look who can certify in the real world and why, it's not a fair picture. These eigenpeople SEOed the capitalist form of pagerank.

We need unconditional basic income. It used to be that some leftist thinkers from the turn of the 19th century considered a wage job being not much better than slavery.

Ah you're right, thanks.

Should have used quotes.. I didn't mean it in any formal sense. What I am saying the nature of unit in complex plane makes it difficult to intuitively imagine invertibility and determinants.

I find it interesting that the counterexample uses C as a field. C is twisted and weird. Maybe the Jacobian Conjecture still holds for reals?

I have enjoyed Craft (1979). Artistically might be dubious, it's a sort of genre/style parody but I think writing is better than the actual Minecraft movie.

Turns out aliens love talking funny and safe balloons too.

Yes, although I wish he would call it properly triage calculus (see https://treecalcul.us/specification/), because although the languages are almost identical in power, the original tree calculus and triage calculus rules lead to different programs.

Personally, I grew beyond this. I tried Common Lisp, Forth and Haskell. I enjoyed books On Lisp and Let Over Lambda.

Now I think the best programming language is Barry Jay's Triage Calculus, which is close to combinatory logic or untyped lambda calculus.

But unlike lambda calculus (which is easily expressed in TC), triage calculus has a built-in quoting and introspection (similar to Lisp's CAR and CDR), which lets you easily add typechecking. So in TC, you can have any syntax you want, just by building the correct abstraction. It's truly an ultimate programming language.

TC shows that there is a false dichotomy between programming language features and syntax on one hand, and the function/API definitions in the standard library or user code on the other hand. Every time you write a piece of code intended to be reusable, you're adding to the language. It can all be expressed as some term in TC.

All the syntactic squibbles are a matter of habit. Although one syntactic feature you really want in TC is the let over lambda abstraction, because it's somewhat annoying to order combinators by hand.

Hate to break it to you, but every time you create a function, you're creating your own language, specific to the domain.

I suspect what you're actually objecting to is a lack of referential transparency, which is a problem in many languages (including Lisp, to be fair).

I think there should be "An Infinite Literature Contest", where the contestants would submit a formal grammar and the texts in the language would be judged for literary and other qualities.

So you think U.S. would be better at airplane manufacturing had Boeing gone bankrupt at some point in the past?

It's a pretty strong claim and I would like to see some theoretical justification, other than belief in the magical free market.

In my country, IIRC, we let Aero Vodochody to crash and be bought by Boeing. We are not better at making airplanes, honestly probably the worst than at any other point in history.

They also say "Wider adoption of AI has made it more difficult to measure task-level productivity"

I think there is a simple reason for that. If you automate something, you make the measureable/predictable thing faster. So the hard to measure/predict part of the job will take more share of the time, and overall difficulty to measure/predict goes up.

I think this is what happened with Agile Scrum - as developers became more productive (for unrelated reasons, two main sources of SW developer productivity before AI were compilers and open source), the bureacracy (amount of meetings) increased, because the ratio of hard to measure vs easy to measure went up. Bureacracy is hard to measure, so it went up (as a share of work). I expect this only getting worse with more automation, such as AI. So I predict an increase in share of bureacracy compared to pre-AI world.

Either way, IMHO main point is automation has the opposite effect on human job predictability, it lowers it. Tasks we can easily automate are those that are easy to predict.

The problem is you can't really hoard products. Most products depreciate - it's a force of nature called entropy.

I think the argument from symmetry still holds, but it leads into a different conclusion. Since products (goods, physical assets) depreciate in value over time, money must too decrease in value. Hence you get inflation.

I believe that "natural rate of inflation" is driven by natural depreciation of goods and the free market mechanism that exchanges money and products as you describe.

I think GP found it useful for this particular use case, as did I and my coworker who is even more enthusiastic about it (he calls it compressed human culture).

And no, Google search (or Wikipedia) won't come close to capacity of LLMs to find similar things.

"Fair" doesn't always mean according to everyone's preferences. I might want to have a full cake but getting a slice is fair.

Well there can be no consequences at T=0, but thanks to transparency, consequences can happen, by a collective decision, at T=1. Therefore having transparency is important on its own, it facilitates change towards fairness.

And that's what I am saying - we should still ask for transparency even in the environment of no consequences.

It's also possible that people are not sure about the lack of consequences, and again, transparency then prevents them doing bad thing even if actually there are no consequences.

But of course tautology is tautological by definition. (I am almost 50 and kinda tired of these eristic games on the Internet.)

You're wrong, it still matters. It's the first step, and it's an important step in maintaining fairness.

I don't know what your argument is. Maybe some US companies won't be financially viable - so what? I mean if it affects you, you have stake in those companies, fair enough. But it's not a geopolitical problem - most US citizens even (not to mention the rest of the world) will be better off if they have access to public models.

Not only that, but it also relies on assumption that everything that LLM does cannot be factored into small reasoning core (something like Lean or Prolog, but in more modal/fuzzy logic) plus knowledge base (like Wikipedia).

It then raises a question, which of these two is really the precious thing. (Although there could be a 3rd thing, a memoization of some shortcuts, like tautologies in logic.)

That's not true, and anyway it's quite simplistic view of history.