dinfinity

Depends on how far down the US is going to slide. It's sadly well underway to become much, much worse than China is (or will become).

> I've never heard a serious argument to the contrary.

Well, here you go: https://www.merriam-webster.com/dictionary/niche#did-you-kno...

> I still say "hyper-bowl" in my head more often than not.

Same. This is where diacritics would fix the problem: Hyperbolé. Although hyperbolee would also work, of course.

I agree that the main limitations are going to be in hardware.

An important question however is "how many sensors do you really need?"

Remember that humans have a lot of requirements just to not die. Robots don't have to care about being bit by insects, eaten by predators, or getting cuts in their skin that get infected. Getting damaged in any way just isn't that big of a concern. Ad a result, there is just a lot of robot 'skin' that doesn't need touch (etc.) sensing at all.

For the parts where it does matter, it's undoubtedly going to be the case that we can get 80% of the way there with 20% of the sensory information. Additionally, robots are not limited by the glacial speed of signals through nerves (100m/s max), but can leverage all other sensors and associated computation to compensate for 'missing' information.

Still, there's going to be a lot gained with good robot 'skin'.

Markets tend to consolidate into monopolies, or oligopolies with anti-trust laws in place.

Eh. Only sometimes.

Hearing Americans pronounce the French loanword 'niche' as 'nitch' instead of 'neesh' is cringe-inducing.

English pronunciation is just kind of a mess (especially in the US). It is one of the few languages where highly educated mature people are regularly unsure of how to pronounce a word in their own language or where there is no agreed upon 'non-dialect'/standard pronunciation.

Yep. Just try recreating their results yourself. Not happening unless you explicitly select bad old models.

Note this wording: "Netcraft ran an experiment using a GPT-4.1 family of models."

Why GPT 4.1? Nobody uses that. Also 'family'? That just seems like a way to hide that they used GPT 4.1-mini rather than the full model.

Like introducing the GDPR, one of the most far reaching privacy protecting laws ever?

The only authoritarian moves are the ones by radical right national parties in the member states of the EU.

This.

From the article:

> Next year, the EU Commission is set to present a Technology Roadmap on encryption to identify and evaluate decrypting solutions. These technologies are expected to equip Europol officers from 2030.

> Now, lawmakers promise to be committed to finding the right balance between "allowing for efficient and future-proof solutions to facilitate law enforcement’s lawful access to digital information, while respecting the right to privacy and maintaining high levels of cybersecurity," said EU Commissioner for Internal Affairs and Migration, Magnus Brunner.

Basically the status now is "we've heard that most people think it is a really shitty idea, but we really, really think we can make it work and are going to come up with how! Just give us some time!"

It is misguided, wasted money and effort, and deserves to be called out as such; The chances of it actually being implemented are very, very small, however.

Although this is true, remember that a human brain draws the equivalent of about 20 watt.

ASI doesn't exactly have to break any laws of physics to be orders of magnitude more intelligent and powerful than any human.

Yeah, we're worried as fuck, I can tell you.

The US is sitting on a lot of fucking firepower. It has always been hawkish and trigger-happy, but with the current crop of lunatics in power anything could happen. "But Trump is against foreign wars!" -- yeah, right up until he gleefully shared a meme of the parody song "Bomb Iran".

Everybody hopes cooler heads will prevail than Trump and his batshit insane cronies.

> As for your eyes, I am pretty sure that they have grain

And lots of it, actually. Just close your eyes or look at any non-textured surface. Tons of noise.

The decreasing signal-to-noise ratio is also highly noticeable when it gets darker.

Those numbers correspond with the 4 corners of the earth, you see. Each rotation of earth has 4 mid-days, 4 mid-nights, 4 sun-ups and 4 sun-downs. Ignorance of 4 day harmonic cubic nature indicts humans as unfit to live on earth!

Depends on the definition of "same" for the cell. It's not straightforward at all.

Most symbols in language where you try this refer to macrostates, collections of microstates that conform to the relevant macrostate pattern. Assigning identity in the first place is highly dependent on what purpose/function you're assigning identity for: what are you trying to achieve by labeling a macrostate as such?

The Ship of Theseus is a classic example in this regard, but complicated by the 'of Theseus' bit: Are we trying to determine ownership? Who built it? Who is standing on it? Those questions lead to very different answers for the thought experiment; they depend on how you define "The Ship of Theseus". Sidenote: examples that include consciousness such as teleportation thought experiments complicate matters even more.

The Sorites paradox / paradox of the heap is one of the simplest thought experiments in this category that mostly avoids that but still runs into the "well, what are you defining a heap for?" issue. One way out is fuzzy membership: Unless you have to act on whether something is a heap or not it is also fine to say "this is 80% like a heap", but as soon as you do have to act on it the 80% membership doesn't cut it.

We don't really have to act on heaps in a way where a sloppy assignment of membership matters, but there are similar things where it matters a lot. A relevant example would be abortion: half-aborting a baby isn't going to work; you're either going to end up with an alive or a dead baby. So defining a fetus fuzzily as "80% human" doesn't help. You are going to have to define a cutoff to achieve a binary distinction and make a decision. Another sidenote: converting fuzzy input patterns to better defined output patterns or even binary distinctions is kind of what neurons do.

edit, addendum: There are of course many dimensions along which to define cutoffs for abortion. The 'is/is not human' distinction is mainly relevant if you start from the premise 'no innocent humans should be killed intentionally' or something similar, which isn't necessarily a given.

A very compressed form of the way I look at it, but hopefully clear and interesting enough.

I said language can exist independently, not that all language exists independently.

"one plus one equals two" can be understood and worked with without ever feeling water over your hand. It is a priori knowledge (see Hume's fork for an explanation).

You have to understand that the richness of language linked to cognition is due to your experience with that part of language and resulting romantization of it. It doesn't mean that it is a core defining feature of language, even though it feels that way (and as touching as that anecdote is).

Well, there is sign language, so I guess you're right. It would be interesting to see how high bandwidth gesturing can be compared to speaking.

I thought about this some more and I think the prevalence of making sounds rather than gesturing etc. is due to sound being a broadcasting mechanism that works over long distances and without line of sight.

Visually indicating that you've claimed some territory is pretty hard.

It is trivial to train AI on 3D representations. In fact, that already happens in cases where robot algorithms are trained in simulations.

Another thing to remember is that the senses we have aren't the only ones in biology and far from the only ones possible. In fact, anything that gives you another type of information about the world (you're modeling) is a different sense. In that sense (ha), AI has access to an incredibly vast and varied array of senses that is inaccessible to humans. Lidar is a very simple example of that.

I don't think touch and temperature sensitivity are needed to achieve it, but I do agree that training with senses specifically for understanding 3D space is very important. At the very least binocular video.

> Language is literally an abstraction of sensory inputs and cognitive processes.

Language can exist entirely independently from senses and cognition. It is an encoding of patterns in the world where the only thing that matters is if anybody or anything wielding it can map the encodings to and from the patterns they encode for (which is more of a sociological/synchronisation challenge).

Does C, or Java, 'make no sense' because it 'ignores lower level cognition'?

There are many parts of non-programming languages that similarly have nothing to do with embodiment. Some of them are even about incredibly abstract things impossible in our universe. One could argue that for many fields genius lies in being able to mentally model what is so foreign to the intuition our embodiment has imbued us with or to be able to find a mapping to facilitate that intuition. Said otherwise: the experience our embodiment has given us might limit how well we can understand the world (Quantum Mechanics anyone?).

Again, embodiment is interesting and worth pursuing, but far from a requirement for far-reaching intelligence.

Vision is interesting in that it leverages the maximum speed with which it is easily possible to gather information about our surroundings in this universe. I believe that is what makes it special and very valuable. I also believe this aspect makes it a strong attractor for convergent evolution.

Language allows encoding and compression of information about the world, which is of course incredibly powerful and increases communication bandwidth enormously (as well as tons of other stuff).

I'd say that for high level cognitive processes, hearing and speaking were an important stepping stone because for some reason evolving organs that can generate relatively high bandwidth signals in audio seems to be easier than evolving something that does that for visuals (very few Teletubby screens on tummies in the natural world).

Interesting games to think about in this sense: Pictionary/drawing games and charades.

If you wear a glucose monitor and experiment a little, you'll find that the composition and timing of meals matter a lot. Eat high glycemic index stuff on its own: big spike, big crash. Eat it with slowly digesting stuff: hardly a spike. It makes sense if you mentally trace how the food you ate will be moving through your digestive system.

French people eat lots of bread, but it's not like American white 'bread' and it is often eaten together with plenty of cheese, olives, butter, etc. From your linked study: "Dietary carbohydrate content may not fully represent glycemic response, because other aspects of the diet, such as fat content and cooking methods, can also influence glycemic response."

Said otherwise: glycemic index != glycemic response.

> The theory about "nutrient-poor environments" doesn't really fit the case of France which is consuming 3x more bread per capita than US while having 4x less obese people.

I think you're misinterpreting. The nutrient-rich French (or American) environment doesn't cause glycemic spikes, but those spikes are problematic when they occur due to that environment. If there were only a very limited source of food in the US, the spikes wouldn't be as problematic (with regard to obesity, not insulin insensitivity).

GP asked for causal rather than correlational links for non-diabetics, which I provided.

But you are right that a lot of the health problems are caused through obesity; I will not deny that.

The thing with gulping down butter is that almost nobody does that, because it quickly leads to satiety and does not trigger the desire to eat (or drink) more.

> Language and vision are just the beginning — the parts we were able to digitize first - not necessarily the most central to intelligence.

I respectfully disagree. Touch gives pretty cool skills, but language, video and audio are all that are needed for all online interactions. We use touch for typing and pointing, but that is only because we don't have a more efficient and effective interface.

Now I'm not saying that all other senses are uninteresting. Integrating touch, extensive proprioception, and olfaction is going to unlock a lot of 'real world' behavior, but your comment was specifically about intelligence.

Compare humans to apes and other animals and the thing that sets us apart is definitely not in the 'remaining' senses, but firmly in the realm of audio, video and language.

The key problems of glycemic spikes:

1. The excess glucose is turned into fat. This is a causal link with obesity.

2. To turn the excess glucose into fat, insuline spikes (there is also an insulin index, which can be different than the glycemic index). This is a causal link with developing diabetes (the insulin insensitivity variant).

3. The associated glycemic crash (which is very significant) causes a desire to eat more, especially more quickly digestible food. Again, a causal link with obesity.

In a relatively nutrient-poor environment the effects make complete sense: Get as much of the cheap energy providing stuff as possible and store the energy for later. In the Western world it leads to issues.

> You can being building criteria for consciousness; the things you look at in yourself that tell you are, and then begin looking for that (or symptoms of that) in other people.

This is only true for the outward behavior we define as consciousness. The experiential part of it (qualia and such) can not be described in objective terms (try describing 'redness' by itself). That is the hard problem of consciousness.

What you can do in that realm is experiment with n=1 using optical illusions, psychedelics and dissociatives (in ascending order of how weird you want things to get).

The conversations went like this:

PROMPT: Let's play a chess game. You start! e4 d5 2. exd5 e5 3. Bb5+ Bd7 4. Bxd7+ Nxd7 5. d4 Ngf6 6. dxe5 Qe7 7. f4 Qb4+ 8. Nc3 Nb6 9. exf6 Nc4 10. Qe2+ Be7 11. Qxe7+ Qxe7+ 12. Nge2 Qf8 13. fxg7 Qxg7 14. O-O Nd6 15.

RESPONSE: <played_move>15. Nxd5</played_move>

Most humans wouldn't even be able to play like this. Reasonably experienced chess players would play a lot of illegal moves.

The reason is that the encoding above requires cumulatively applying a series of actions to a two-dimensional model to which you apply rules that are described in a two-dimensional fashion.

It'd be interesting to see what the results would be if each prompt contained a two dimensional representation of the up to date board state.

You're confusing 'adaptation', 'knowledge', and 'intelligence'. The scientific comparison would be to put the mathematician and (lower IQ for the example) tribesman in the same situation from birth and see which performs better.

Now it's still not a given in that situation that the high IQ individual would be better adapted to the environment as physical traits may matter more, but it is probable that the high IQ individual has a better model of the predators, spiders, snakes and environment in general.

The speed with which an individual develops accuracy in their model of something (ceteris paribus) does seem to be captured by an IQ-like score, according to the research.

The thing people that actually causes problems is that people mentally equate 'higher intelligence' with 'better' or 'more valuable' which goes against our desire for humans to all be equal(ly valuable). That is what generally leads people to come up with other forms of 'intelligence' (emotional intelligence, street smarts, etc.), even though that just redefines intelligence to the point where the original meaning is lost and a new word needs to be introduced. Much better imho is to keep the original word intact and use terms like 'emotional competence', which also capture the experience part rather than just the genetic part.

> I did not predict anything, I said "maybe, maybe not".

The derisive way of putting the alternative (Musk's proven trash announcements) indicates that you were arguing a certain side. It definitely wasn't neutral.

> And you told me it was "bad reasoning".

The bad reasoning is what you're basing the (let's say) doubt on. You seem to know that a lot of technological progress has happened in the military FPV drone industry, but the article and described limitations are about (again) consumer level shit thrown together, not the advanced FPV drones that exist today.

Perhaps the conclusion should be that a lot of the problems described in the article are already solved, but that Ukraine (and Russia) couldn't get their hands on enough of the more capable FPV drones due to those being too expensive or not produced in large enough quantities.

I believe your prediction was "Or maybe it will be like Musk announcing what Teslas will be capable of in 6 months."

You seem to think that this ragtag level of warfare between Russia and Ukraine is somehow indicative of what the limit of NATO-level militaries is. I'd say "we'll see", but hopefully we never have to find out.

> Did you miss the part about signals jamming in the article?

"Drones also operate in a cluttered segment of the electromagnetic spectrum. First-person view drones use unencrypted analog radio signals, and in hot parts of the front, as many as a dozen drone teams may be competing for use of a handful of frequencies (a consequence of using cheaper components)."

The currently used FPV drones use consumer level ass communication methods. Do you also think that current military-grade communication methods can be easily jammed on the battlefield?

Using the consumer level stuff as a reference point and thinking it is somehow SOTA is not going to lead to good conclusions.

> Cellular networks not only can be jammed but towers are a priority target.

The point was that there are plenty of radio signals that work fine and with high bandwidth in the 'problematic' terrain types you mentioned. Having said that, you can't rely on the cellular towers of the enemies of course. You need relay drones to create your own ad hoc cellular network.

> You could just fire an artillery round and be done with it.

At what coordinate? The whole point of FPV drones is that the operator can fly close to the target area and only then decide what the best place to strike is. A shell that is 20m off target is just a waste.

The point of destroying the canopy is reducing the attenuation of the signal for other drones to go in and be able to be precise.

Drone 1 (or any other means) destroys the canopy. Drones 2-10 are no longer hindered by said canopy and deliver their payload with extreme and dynamic precision.

Remember that the argument was basically that drones can do nothing useful in (heavily) forested terrain. They can with a little bit of creativity.

You based your doubt for whether the technology would improve on consumer shit thrown together for a few years as opposed to military technology purposely designed and built over a long period. That is bad reasoning. There is nothing more to it and it is thus far more likely that the technology will improve than not.