The problem I always had, historically, with games on DVD is they'd kind of "beat up" the optical drive, seeking all over the disc constantly while the game was loading and then being played.

It wouldn't have been so bad if they simply copied everything onto the console's hard drive or SSD when first installed. But games like God of War on my PS2 are what I blame for the drive wearing out so it got constant I/O errors trying to run anything.

I had the same experience with a couple of PS3s that our kids used and "wore out". It was always the optical drive failing. If you played only downloaded games, they'd run fine.

I think I can explain this. I'm not sure how well, I'll do but I can try.

For "nice" objects, we all have an intuition about how many dimensions the object is, which is roughly how many variables it takes to specify a given location on the object. For example, on a line embedded in your standard Cartesian two dimensions, say y=3x+2, you can just tell someone the x coordinate and they know exactly where they are. This works in general, so typical 3 dimensional space needs 3 coordinates. Etc. Similar, a circle (even though it lives in 2 dimensions) just needs 1 variable to tell someone where you are on it because you can take an angle from the center.

However, lots of objects turn out to have a notion of dimension that doesn't quite match up this way. This is roughly what people mean when they call something a "fractal" (although some people use fractal more to mean a thing deifned by a recursive procedure. Defining it precisely is tough.) Now, to talk about objects that are fractals, like say the Koch snowflake https://en.wikipedia.org/wiki/Koch_snowflake we want a notion of dimension that makes sense for them. Ideally, for a given object we want a notion of "dimension" which gives the correct value of dimension when we apply it to an object that is a traditional object with a known dimension, like a line, or sphere or ball. The Minkowski dimension is one such approach. It is easiest to explain in 2 dimensions, but the same basic idea looks in any number of dimensions. Roughly speaking, you take the object you want, and you pick some tiny number epsilon, and then you make a grid on the plane of little boxes each which are epsilon by epsilon, and then you ask how many tiny boxes does it take to cover the object? Then, you take that number and divide it by log (epsilon), and then you let epsilon go to zero. This essentially measures how pointy thick the object is. If the object is thick this will just be two. If it is just a few isolated points (even infinitely many but not a lot) this will be zero. It isn't obvious that this is the right thing to count but turns out to work pretty well. You then generalize this to any number of dimensions for your space your object lives in by using boxes of the corresponding dimension number. So, for a fractal which lives on the real line, you use intervals of length epsilon, in 3 dimensions you use cubes of side length epsilon and so on.

Now, the Minkowski dimension is not the only notion of dimension we have that works this way. There's a related idea called the Hausdorff dimension which is defined in a more complicated fashion but turns out to often be easier to calculate. For example, it is not too hard to show that the Hausdorff dimension of the Koch snowflake is (ln 4)/(ln 3) which is about 1.26. (This reflects the intuition one might have that the Koch snowflake is much closer to being a thing made of lines than it is to being a thing with normal area). Now, it turns out that the Minkowski is always at least as large as the Hausdorff dimension, and they are equal in many situations. But there are some annoyingly simple situations where they are not. For example, if you take the interval of points between 0 and 1 and just take the rational points, then it has has Hausdorff dimension zero and Minkowski dimension one, which is about as far as you could want. Sometimes we'll calculate Haussdorff dimension for an object but really care about the Minkowski dimension, where finding that takes more work.

In the case of Hong Wang's work. She and Zahl proved that if you had what is called a Kakeya set in 3 dimensions (where you want a set that can fit a needle in every direction but you don't care about rotating it around, you just want a needle to fit at any orientation you choose), is that the Haussdorf dimension of the set, and the Minkowski dimension have to be equal and in fact equal to 3. But note that this isn't obvious in part because one doesn't have the rotation requirement here.

Does that help/make sense?

A bit more about the history of the Kakeya problem, which Hong Wang worked on. It comes from originally a more concrete question, namely how small an area do you need to rotate a 1 unit long needle 360 degrees if it is allowed to also move around as you rotate it? The naive thing is to use a circle, but you can also do a bit better by using a triangle, and then rotating it a little bit at the corner, moving the needle to the next corner, and then continuing the rotation. But you can do a bit better than this by cutting out small regions from the sides of the triangle. You instead use a deltoid https://en.wikipedia.org/wiki/Deltoid_curve. Kakeya asked if this was the best possible construction But Beiscovitch made a very clever construction to show that there is no minimal area. That is, for any epsilson>0, you can make a region of area less than epsilon where you can rotate the need. https://en.wikipedia.org/wiki/Kakeya_set#Besicovitch_needle_sets . Thinking about these ideas in abstract settings and higher dimensions lead to what was called the Kakeya conjecture, even though it post-dated Kakeya's own work. Hong Wang Joshua Zahl (not me, different mathematician with the same first name and last initial) proved that conjecture.

We don't know what exactly they were thinking, but if I had to guess, they may have not realized that the airplane could sink at all, and had read/heard about other disasters both real and fictional involving people getting hurt or killed in lifeboats.

Huh? I'm confused then by what you are trying to argue here. I gave a list of examples which were hard to reconcile with the claim that these systems were just " just really really fancy search engines that regurgitate what they have ingested." You then switched to asking about practicality and useful pay off. Another user, ceoyoyo said you were moving goalposts. You then made a comment that this was instead "progress," and I'm having trouble figuring out what you meant. And I'm still pretty puzzled, since despite your claims here, you haven't given any reason in this discussion to think reality is any way whatsoever. If you think my examples don't show that these are not just regurgitating the data in their training sets, then please explain why, because I'm not seeing any argument for it here. If you think that solving math problems that have been open for 40 to 70 years counts as just regurgitation, then it seems likely that you are using a notion of regurgitation which would include the AI doing almost anything.

I'm not sure why you think I have done this, or that I'm a bot. You can easily check that my Slashdot account has been around for well before there was any issue of "bots" on the internet. Heck, you can go through my profile and check that my writing style hasn't changed much in the last decade or so. (And you might even recall I've had conversations with you before here on a wide variety of topics.) I didn't downmod any of your comments, and I'm not sure why you think I would have. I will comment that given how many people seem to really dislike you on Slashdot (admittedly it may just be one or two very obsessed people and I'm not sure what you've done to arouse their ire that you have these ACs constantly following you around) I don't see why you wouldn't just assume one of them went and used some mod points. I don't have any alts, and if I did, there are a lot of people I'd be burning points on before I got to you, since you are normally marginally reasonable and civil, even if I've disagreed with you on a bunch of things in the past. But I'd really rather focus on the specific claim at hand that the AIs are merely regurgitating. Can you explain why you are convinced of that and how that plays out? And again please note that's distinct from any question about usefulness or practicality, or whether they are worth the tradeoffs.

That's true, but in this case, we also have Hugging Face who would be able to contradict Altman if he were lying. This really looks like this is exactly what happened. A model broke out of a sandbox with a goal of hacking a specific target and then partially succeeded at that. I'm not sure what you would find more convincing than that than that as the next step. To people need to be hurt in order for the risk to be recognized?

Does that mean that we're now in agreement that "systems aren't creative they're just really really fancy search engines that regurgitate what they have ingested" is not an accurate assessment then?

With the exception of the Jacobian conjecture, none of these have any practical applications. But it is mistake to think that these problems were not solved previously due there not being any "good reason" to solve them. All of these were well known problems where major mathematicians had worked on them. Erdos 1196 was a problem which saw a lot of work by Jared Lichtman who is one of the leading young number theorists now. The other problems are even more so. Both the unit distance problem and the double cycle cover conjecture were problems that had many mathematicians work on, and the Jacobian conjecture is a problem where many people have had partial results about it and where people have spent years solving special cases.

The rest of your post isn't on the topic in question, but is about broader issues with AI, which may be valid, and may be worth discussing. There's no question that there are some pretty serious negatives. To use one personally annoying example, my spouse works in a library and she has to deal with trying to explain to patrons that just because ChatGPT said a book exists doesn't mean the book is real. But the point of my comment remains focusing on the claim that these systems are " aren't creative they're just really really fancy search engines that regurgitate what they have ingested." That's not accurate, and the results in math show that. And if we are going to understand these systems, both their positives and negatives, we need to understand their capabilities.

Are we still saying this? These systems have solved a wide variety of open math problems some of which were borderline famous. Erdos 1196 https://www.erdosproblems.com/1196 was one of the first. Then the Unit Distance Conjecture https://arxiv.org/abs/2605.20695 . Then the Double Cycle Cover Conjecture https://en.wikipedia.org/wiki/Cycle_double_cover. And now most recently, one of these systems came up with a counterexample to the Jacobian conjecture https://mathoverflow.net/a/513385/127690, a conjecture which is old and famous enough that I've literally talked about it in class when I've taught linear algebra. At least in pure math, these capabilities are well beyond being mere regurgitation. It is likely that's the case in other fields as well, and this is just one where it is slightly easier (due to the completely self-contained nature of the problems and the nature of mathematics) and where it is easier to see that it really has done something notable.

Well, the next step seems to be claiming that mandatory nutrition labels on food is a violation of free speech.

OK, which particular aspect of the LLM learning do you think doesn't exist in the human brain? Backprop exists (and the other training methods the brain uses has similar effects), embeddings more or less exists and seems to be very similar to the LLM embeddings (shown by brain activation in various areas linearly matching with embeddings) - see e.g. High-level visual representations in the human brain are aligned with large language models (Adrien Doerig et al, 2025), and speech/text generation by humans has similar error patterns as for LLMs.

They obviously don't learn in the EXACT same way, but there's also clearly a lot of bits that are the same, and I'm fairly sure nobody knows how much is similar yet, particularly not weighted by importance.
 

I paid for a year of access to their service last year (recently let it expire/lapse) ... and I agree with another Slashdot poster who called it "impressive".

I guess you can fight legal battles endlessly over what you're allowed to do with content that was made freely available for you to access over the Internet.... But those "vast music libraries" they "stole" were the same ones the average user of services like YouTube are welcome to pull up and listen to any time they like.

We're really just arguing about if it's ok to write code so a computer can analyze the music and use it to create new music based on ideas it "learned" from the content ... vs human musicians doing the same thing.

To me, the impressive part of SUNO was the way I could supply my own original lyrics as text, complete with instructions on how I'd like to hear the words sung, and have it churn out a realistic-sounding result with a backing track fully assembled to go with it. If you listen to enough SUNO content, you start to get a sense that specific music genres it uses result in only a certain resulting sound/feel/vibe. I could tell it to regenerate something I told it to create in the style of an "Irish jig", for example -- and over dozens of attempts? I'd wind up with maybe 4 or 5 really different ways it constructed it, and the rest feeling like small changes to those basic constructs. But to me, that's ok. You shouldn't try to use an AI music creation tool to crank out complete, "ready to play/perform" pieces of music that got rid of human musicians. A SUNO creation should be identifiable as a SUNO creation when a discerning listener hears it.

I see SUNO handling relatively "low effort" music creation needs like advertising jingles or as a tool to inspire a musician to build from what it gave them as a staring point. For a lot of background music, such as what's needed in a video game? It makes sense too.

I feel like it's not only possible, but likely America saw relatively major power outages at close to the same "one per month" rate in the past too? The electrical grid is basically designed with an assumption it only stays up with the help of a crew of linemen who get tasked with locating points of failure and fixing them ASAP.

I remember some years back, I lived in a small city right on the edge of the Potomac River in western Maryland. They were originally set up with "feeder" power lines coming from two directions in to town. At some point, Potomac Edison power company decided to just discontinue one of those feeder lines and let the city get by from the other one. Every time a car hit the right power pole coming in to town, after that? Power was out for the whole community.

Seemed insane to me that they'd purposely remove redundancy they already had in place? But I'm sure it was all about the economics -- with bean-counters realizing the lower grid reliability was still "adequate" per the total population there, and they'd save all the money maintaining the additional lines and poles.

I also remember living in a suburb of St. Louis, Missouri where they had a power outage lasting over 2 weeks. A storm came through and knocked down a lot of trees. (The community took pride in having all the trees growing there, but I guess didn't consider how bad that was for above-ground power lines running right past all of them.) They had to get crews from other states out to replace blown transformers on poles and the whole bit, to get it back up and running.

If better uptime was a major issue, you'd think they'd bury those power lines. But again, it's about cost-savings instead.

(What the heck does a PalmPilot organizer have to do with any of this?) Well ... I remember back when they first got popular, my buddy was a software developer at the company I worked for. At lunch one day, he mentioned how he absolutely loved the Palm OS platform, simply because it had so many limitations. He said when writing for the Windows PCs, by contrast? You had so many system resources and so many options, you could pretty much code anything you could come up with. Sloppy code was a non-issue too, on modern systems. Only the coders reviewing the source would know any better. He liked the mental challenges involved in maximizing what you could get from a Palm device with a small monochrome screen and the whole bit.

I feel the same way with movies. All these mega-mergers may give a few big-name film-makers massive financial resources to create new movies. But most of it is unnecessary to make an amazing film. What you need is a great story, and good acting (which really isn't some monopoly held by the big Hollywood stars!). Less is very often more. (Consider how well the first Star Wars trilogy held up over time, using simple backgrounds like a mostly empty desert for Tatooine. I prefer that to the crazy "busy" AI generated scenes in the newer movies.)

Data center use of power is very high, but water is often being overestimated. See https://theconversation.com/ai-has-a-hidden-water-cost-heres-how-to-calculate-yours-263252 for a good basic discussion of the issues involved.