pegasus

I guess the question of why counterexamples are so rare and/or hard to find is still left to be investigated.

I feel the same. When those large datacenter projects were being considered, the necessity for building new power lines should have been included in the ask. Instead they built the centers, then "notice" the system is straining and stick the public with the bill and local communities with the consequences. Classic "privatize profits and socialize risks".

What I'd really wish for is an EU browser that 'just works'.

Have you tried Zen Browser? It's entirely open source but the core team and sponsor are European, and it's actively positioning itself as the European alternative. I've been giving it a go lately. So far, so good, and I like most of their UI/UX choices.

That difference is not essential to the question at hand. An open book test based on an erroneous book would give the same results as this test, even if it wouldn't penalize blank answers.

Yep, a possibility that has found support with Roger Penrose, who wrote two books about it.

I don't see how stealing IP is inconsistent with a sympathy for openness. If anything, it's the opposite. "Information wants to be free" and all that. They were just liberating those secrets ;)

That's just one surface feature of how fascism manifests at the level of economic policy, it doesn't capture the essence of the movement. You wouldn't expect Mussolini to have the introspective capacity to actually understand those murky depths, let alone the intellectual honesty to put that capacity to work, would you?...

I recommend Umberto Ecco's essay "Ur-Fascism" [1] if you want to get a more accurate picture of what fascism is. It might make you reconsider which of these two countries fits the pattern better, at least at the moment.

[1] https://en.wikipedia.org/wiki/Ur-Fascism

"Mussolini's model"...

Wikipedia has a whole page on the definition of Fascism [1]. Also worth reading is Umberto Eco's article on this subject, which also has its own wikipedia page [2].

Spoiler alert: it's not as simple as you imagine.

[1] https://en.wikipedia.org/wiki/Definitions_of_fascism [2] https://en.wikipedia.org/wiki/Ur-Fascism

Watercolors in SVG?

Isn't steering the agent the new journey? Why shouldn't that provide "little intrinsic rewards"?

Prefill isn't supported in the major commercial models anymore. Probably in order to thwart jailbreak attacks.

Not only is AI art often slop, it's also often theft. Or maybe even always theft. And it's both of these things that account for this newfound ability to create something normally time-intensive in seconds. Imagine I want to have a world-class painting on my wall: the fastest and cheapest way is to steal one from a museum, especially if I found a way to obscure the act. In the case of LLMs, its obscured by the fact that the model draws from many sources, blending them together in a way that's hard or even impossible to separate. Is stealing a small amount from many more excusable than a lot from just one? I don't see why that would be so. It's the total amount that measures the gravity of the transgression.

It's because they are post-trained to be agreeable, which is clearly a desirable trait in a model. I think the correct way around this is converting the conversation from first- and second-person direct, to third-person indirect, making the rope-playing obvious. I.e. '''prisoner says "I'm about to teleport to Narnia", what's the dungeonmaster's response?''', or even '''prisoner says he will teleport to Narnia [...]''' etc.

From Shannon's theory, it's abundantly clear that "information" is just a mathematical description

That's a very narrow and technical definition of information. It's what people refer to by the word when they discuss things like: communication, medium, surprise, compression, etc. in a technical context. Otherwise one should refer to the dictionary definition, so we don't end up talking past each other.

If you'd like to get a better intuitive sense of what "it from bit" points at, check out David Deutsch's Constructor Theory and the Stephen Wolfram Physics Project. Also, Bernardo Kastrup, on the philosophy front.

Why shuffle, though?

Finally, to claim that information is somehow the important fundament of existence is patently absurd to me, since it would've similar to arguing that miles have some very special epistemic and ontological status.

As the sibling comment says, no, it is like arguing that distance/space/extension has a special ontological status — which it does. "It from bit" is just using the word "bit" for the rhyme, the theory is quite profound and compelling, at least to me.

Then there's also Max Tegmark's "Mathematical Universe" idea, which makes the symbolic world the fundamental one. It goes a bit far for my taste, but it has its own internal coherence.

This, a thousand times. As the ratio of code to human writing necessarily [1] goes up, they become not just smarter, but more precise and technical, which requires them to use more jargon. You could say they become more nerdy. Hence, text generated by these models will become more easily recognizable, at least by default, when not asking them to twist themselves into something else via prompting — which degrades intelligence. This is a good thing, in my book, given all the slop we already have to contend with.

Of course there will be models trained on much less code and technical writing, and they will create more natural sounding prose, but they will lack the deep intelligence of frontier models. Seems like a fair tradeoff.

[1] watch the first couple of minutes on this bycloud video on scaling training data mixtures: https://www.youtube.com/watch?v=aD93kfArOik

The Church-Turing thesis (the fact that all models of computation we've been able to come up with are equivalent) gives the necessary weight to that claim.

This stance is very tempting because it appears so straightforward and no-nonsense. But there are issues which undermine it in my mind. Because, at a closer inspection, it itself doesn't seem informative, in the end coming down to a question of semantics, i.e. how we define the word "reality". Which then determines whether it encompasses the realm of mathematics or not. I wonder what definition you favor, but the language you use points towards something like "the totality of all things" or "everything that's made from a physical substance". Yet platonists and other realists wouldn't argue that numbers are made from a substance as you imply.

Physical reality is a slippery thing. Yes, at the level of our daily life it all seems very solid and obvious, but inspecting that foundation revealed a bewildering world of relativity and uncertainty. And one of the most intriguing ways out of this bewilderment seems to be the "it from bit" trend in physics, which makes information the fundamental reality.

I personally don't feel the need to tie the notion of reality to physical phenomena. I prefer something more along the lines of Philip K. Dick's "reality is that which doesn't go away when you stop believing in it", or, as you put it, "something which exists independently of us". The fact that there are an infinite number of primes, is such a fact. Surely any alien civilization would know this fact. Maths doesn't go away when we look away, and neither does the notion of computation. Sure, Turing machines are a human construct, and there can be others as well. But the Church-Turing thesis speaks about how these constructs all ultimately describe the same underlying reality.

You're encountering the "contrarian dynamic" which dang described in this post (which also explains why your own anti-contrarian comment reached the top) https://news.ycombinator.com/item?id=24215601

[...] claiming that math is producing objective facts ignores at least a few hundred years of philosophy of mathematics (if not more). Even practicing mathematicians like Chaitin have described math as being more about inventing than discovering.

Chaitin's work (including in his own opinion) moved the philosophical needle on maths towards "discovery" and away from "invention".

Our concept of logic based on our subjective experience of "truth."

The idea that we experience objectivity subjectively, thus there is no objectivity, seems like an ultimately nonsensical and self-defeating sleight of hand to me.

He's just saying the article does not even discuss this vital distinction, let alone bring evidence either way.

I would agree, having visited GB a few times in the past, one thing that is apparent is how solid things are built. Mailboxes, yellow cabs, medieval cathedrals.

Plenty of these tropes have some truth behind them, even if that truth often gets flattened into a caricature of itself in popular imagination. Then you get the anticaricature caricature (reverse snobbery) where people knee-jerk label everything that rhymes with that trope as a caricature, convinced that's enough to dismiss an argument.

Btw. medieval Europe, or classical Greece, just to give two non-Eastern examples, also had many of the characteristics I mentioned in my comment: a strong communal spirit and shared symbolic life which connects one to a coherent and poetic view of the world. But pointing out ways cultures very different from ours might have been superior in some senses to ours is one of the most triggering moves, especially on a techie-oriented forum like this one.

If all you see there is something "cool", you're missing the more rewarding depths. These symbols speak of a society which was at the other extreme end to the fragmentation, atomization, and attention span erosion we are experiencing today. This allowed them to elevate esthetics to a point we couldn't afford to today. I'm thinking here mainly of those ancient Mon[1] symbols – though something of their spirit lives on in the modern symbols also discussed in the article. Contemplating these evocative designs can help us build a bridge to a societal context that's otherwise almost incomprehensible to us.

Another thing I found fascinating: the way the symbol that marks a fire truck doesn't employ the red color we would expect in the West, and is instead based on the shape of an ice crystal. A bit like an amulet to ward evil away, there's some kind of atavistic magical function still afoot there.

[1] the wikipedia page linked from the article is worth a read: https://en.wikipedia.org/wiki/Mon_(emblem)

these are not good symbols. None of them have any meaning, they must simply be learned.

They are excellent symbols – for the society they arose in, which was at the other extreme end to the fragmentation, atomization, and attention span erosion we are experiencing today. In their case, choosing esthetics over learning curve totally made sense, because the learning curve was not a concern for them. Our society OTOH cannot afford such niceties.

Sure, real-world usage is always more difficult to benchmark, but the additional issue with the one shot prompting benchmark is that by optimizing for it, models are nudged towards making all those assumptions they shouldn't really make. Maybe a better test would be to have a fully spec'd-out plan, but start with a one shot, high-level prompt and expect the agent to discover your preferences by repeatedly asking for clarifications. The system that manages to suss out more of the details in the hidden spec this way, in less steps and with less unnecessary questions would more likely to be a truly well-calibrated agent.

Have you been in nature recently? We've evolved to deal with very busy pictures and parse relevant information from them.

Have you ever played "spot the bird in the forest"? Let's not turn our interfaces into that. OTOH, I do grant that there are personal preferences in play as well. There is, for example, the phenomenon of Japanese web design previously discussed here on HN: https://news.ycombinator.com/item?id=47122789

New UIs often don't have relevant information

That information is usually implicit. You're supposed to know that a scrollbar will materialize once you've moved the pointer to just left of the right edge of the bounding box. We've interacted with these elements enough to be able to not be confused just because that particular element is not always in my face. Same for 3D buttons, a well-designed interface doesn't need such heavy cueing. A "Send" button shouldn't need any more than the semantics of the word already implies to make it clear it's clickable. Unless one wants to start clicking buttons before reading (and understanding!) the text. It's the UX inconsistencies, design flaws and general bugginess I've already highlighted in my comment that are making this more painful than it needs to be. Though it's not scrollbars that frustrate me on latest macOS, that part works comparatively well as far as I'm concerned.

Do you sincerely believe Dario wants Fable to be restricted to US citizens only?