I won’t pretend I understand all the math and the notation they use, but the abstract/conclusions seem clear enough.

I’d argue what they’re presenting here isn’t the LLM actually “reasoning”. I don’t think the paper really claims that the AI does either.

The CoT process they describe here I think is somewhat analogous to a very advanced version of prompting an LLM something like “Answer like a subject matter expert” and finding it improves the quality of the answer.

They basically help break the problem into smaller steps and get the LLM to answer smaller questions based on those smaller steps. This likely also helps the AI because it was trained on these explained steps, or on smaller problems that it might string together.

I think it mostly helps to transform the prompt into something that is easier for an LLM to respond accurately to. And because each substep is less complex, the LLM has an easier time as well. But the mechanism to break down a problem is quite rigid and not something trainable.

It’s super cool tech, don’t get me wrong. But I wouldn’t say the AI is really “reasoning” here. It’s being prompted in a really clever way to increase the answer quality.

China/russia/middle east not allowing it, is not the same as not being available. Did you even check the coverage map before replying.

So can you use it or is it not available then? And yes, I checked that map, where else do you think I got the list from??

Astronomers complain about light bleed from ground cities as well. No one was telling them to shut down the cities.

People claim we should turn down city lights all the time! Under what rock have you been living? But for city light bleed, astronomers have an alternative solution, simply place the telescope somewhere not near the cities. And yes, whenever a city tends to grow near one of those telescopes astronomers do kick up a fuss about it.

If you fill LEO with thousands of sattelites, there’s nothing astronomers can do about that.

Lol no just no… I dont know where you live but the majority of people in rural areas are not served, otherwise starlink would have never taken off and been sustainable.

I don’t know where you live, Mars perhaps?

https://nl.m.wikipedia.org/wiki/Bestand:InternetPenetrationWorldMap.svg

Clearly shows most of the Earth has internet access. Or do you think the US has no rural areas? They’re still above 90% somehow. Oh wait, I know, they must be using those mythical internet-via-sattelite services that existed well before Starlink did! I wonder where you’d find a mythical creature like the Viasat-1 for example.

Starlink took off because they promise higher speeds than some ISPs and most other sattelite companies do at lower cost, not because they’re your only option. Starlink has 3 million customers, which makes them the size of a small ISP.

Again this myth you keep spouting that the majority of the world has access is bullshit

Except for the fact that the data backs me up.

planes exist but you need to walk because you live to far from the airport is some classist bullshit.

Continuing your analogy, you propose demolishing the local university because people are entitled to fly to Ibiza, or their local supermarket. Or something, it’s not like it made much sense anyway.

You still completely failed to address the main point, that universal high-speed internet access is not critical for most of the world, certainly not for areas that have always managed perfectly fine without, and that filling up LEO is a disaster for astronomists that they don’t have a workaround for. If you’re not going to actually argue that point I think we’re done here.

Starlink doesn’t cover the globe, it’s available in the Americas, Europe and Oceania. It’s not available in most of Africa, the Middle East, India, China, Russia, Indochina. E.g. the majority of the world cannot access Starlink.

I don’t give a shit that Starlink is owned by Musk. Starlink as a company seems fine (it’s not X or anything), but I strongly dislike that their product messes with astronomy in such a major way that astronomists complain about it every chance they get.

You know that some of us are 10miles from town and considered rural? And the big Telecoms refuse to run broadband for us?

Sounds like your fight is with “big telecom” and with your local government for not putting up a good enough quote to run fiber. This isn’t an issue for large portions of the world, including rural areas, where they’ve figured out how to get them to lay fiber.

Internet access for a long time has been pushed as a priority and should be treated as a utility and that everyone should have access to it.

Access is not the same as high-speed access. Almost all of the world has some level of access, even in rural areas, through sattelites that are not in LEO. Enough to (slowly) browse, not enough to stream in HD. I don’t believe sacrificing considerable astronomical discoveries and progress is remotely worth it when feasible alternatives are available and have been used in large areas of the world already.

Yes, that was the joke. Clearly people didn’t get it.

Photo is clearly mirrored

You realise that this has held true for literally everywhere, and that it’s only a matter of time until they’re connected too? Between 2017 and 2023 an additional 20% of the world received internet access, a trend that doesn’t appear to be slowing down just yet. By 2030 approximately 80% of the world will have internet access, and somewhere between 2040-2050 we’ll consider the entire world to be connected.

I still see absolutely no reason to screw LEO and fill it with sattelites, just so that someone in bumfuck nowhere can Netflix or something. Internet access may be important for a western lifestyle, but the 90s barely anyone had internet and they lived perfectly fine without it. Even before Starlink sattelite internet existed (and still does), it’s just slower.

Most developing countries have pretty decent internet access already. Maybe not in the more remote areas, but again, access to the internet is not a requirement to live. Internet has barely existed for 30 years, I don’t think screwing up LEO in an attempt to bring faster internet to people who didn’t have it anyway is remotely reasonable.

https://isfdb.org/cgi-bin/pl.cgi?719892

Not living in an expensive city doesn’t equate to living in extremely remote areas. If you choose to live in an area with very few services, then don’t expect the rest of the world to bend over backwards to provide those for you at their expense. The sheer entitlement is hilarious.

Besides, there’s still internet, just not fast broadband.

Ehrm, no to both questions? You live in rural fucking Canada. Connectivity will be shit, that’s a given. If you choose a job that relies on that, you should move to where you can actually work.

Fast internet is a privilege, not something people “deserve”. Fucking up LEO so people can stream or Netflix or whatever is absolutely not worth it, and imo the practice should be banned. Starlink has been disastrous for astronomy already. Put fiber in if it’s so important, expensive but hey, people “deserve” it right?

He did what he was supposed to and delivered the evidence is an objective, neutral manner. It was congress, particularly the GOP, that (intentionally) fumbled the ball after.

For a month, we’d talk about aliens. Then, we’d talk about Trump or Musks latest shit takes about aliens.

It’s not a direct response.

First off, the video is pure speculation, the author doesn’t really know how it works either (or at least doesn’t seem to claim to know). They have a reasonable grasp of how it works, but what they believe it implies may not be correct.

Second, the way O1 seems to work is that it generates a ton of less-than-ideal answers and picks the best one. It might then rerun that step until it reaches a sufficient answer (as the video says).

The problem with this is that you still have an LLM evaluating each answer based on essentially word prediction, and the entire “reasoning” process is happening outside any LLM; it’s thinking process is not learned, but “hardcoded”.

We know that chaining LLMs like this can give better answers. But I’d argue this isn’t reasoning. Reasoning requires a direct understanding of the domain, which ChatGPT simply doesn’t have. This is explicitly evident by asking it questions using terminology that may appear in multiple domains; it has a tendency of mixing them up, which you wouldn’t do if you truly understood what the words mean. It is possible to get a semblance of understanding of a domain in an LLM, but not in a generalised way.

It’s also evident from the fact that these AIs are apparently unable to come up with “new knowledge”. It’s not able to infer new patterns or theories, it can only “use” what is already given to it. An AI like this would never be able to come up with E=mc2 if it hasn’t been fed information about that formula before. It’s LLM evaluator would dismiss any of the “ideas” that might come close to it because it’s never seen this before; ergo it is unlikely to be true/correct.

Don’t get me wrong, an AI like this may still be quite useful w.r.t. information it has been fed. I see the utility in this, and the tech is cool. But it’s still a very, very far cry from AGI.

This is true, but it’s specifically not what LLMs are doing here. It may come to some very limited, very specific reasoning about some words, but there’s no “general reasoning” going on.

Covid was going down by then I think.

Nuclear is so ridiculously energy dense that there’s no near future where we’re in danger of running out

Fissile material is not everywhere. A lot of it comes from Kazakhstan. I’ll let you check the map to see why that might be an issue. I also think you might be overestimating how much fissile material is economically feasible to extract.

I’m not saying we shouldn’t be also building up renewables but every powersource has its limitations and making a mix of all of these to cover each of the downsides is our best bet.

Nuclear is a bad addition to that mix. It’s too expensive, relies on rare materiald but most of all, it’s deeply, fundamentally inflexible. The variability of the cheapest energy sources (solar/wind) makes nuclear a bad idea because you can’t quickly turn a reactor on or off in an economically feasible way. These reactors are already so subsidized, they’d need to be running almost 24/7 to make it profitable. And with cheaper energy coming from solar and wind, we’re just not going to need that. And once a nuclear reactor needs to be turned off repeatedly, it’s profitability disappears into thin air.

Even a gas-powered power plant is a better idea at that point. It has some emissions unfortunately, but it is very quick to power and scale up when needed.

Probably not if the deafness is caused by brain damage.

I actually think this form factor makes more sense than a two-fold. With two-folds you get this weird not-quite-square aspect ratio, but this gives you something much closer to a very typical 16:9. Perfect for watching movies or streaming games.

I do want a proper android device though, which Huawei can’t provide anymore unfortunately. But I hope other manufacturers try their hand at this form factor. Still, massive props to Huawei on the design and engineering feat, it’s genuinely the first phone I’ve seen in years that made me reconsider what I want in a phone.

And no, obviously a tri-fold isn’t necessary, but neither are smartphones in general. Took me years to decide to purchase one and I was fine without one (started with a year-old Samsung S7, for reference).

Does that make any sense? No.

Glad he’s figured that out.

It’s not always true that appeals must go to some “higher court”. In some countries appeals may end up before the same judge, or another judge from the same court.