VoterFrog @ VoterFrog @lemmy.world

Posts

0

Comments

146

Joined

2 yr. ago

The examples they provided were for very widely distributed stories (i.e. present in the data set many times over). The prompts they used were not provided. How many times they had to prompt was not provided. Their results are very difficult to reproduce, if not impossible, especially on newer models.

I mean, sure, it happens. But it's not a generalizable problem. You're not going to get it to regurgitate your Lemmy comment, even if they've trained on it. You can't just go and ask it to write Harry Potter and the goblet of fire for you. It's not the intended purpose of this technology. I expect it'll largely be a solved problem in 5-10 years, if not sooner.

Soooo… Interstellar was wrong with all the shaking of the camera?

All for the cinematography :) I will say that there's a small caveat in really extreme situations like close to a black hole. Spacetime gets so warped there that your head and your feet take very divergent paths through spacetime, enough to stretch you out and even break you apart at the atomic level. You'd definitely notice that...

In case of accelerating ship, I wonder what would happen in local frame once you hit/get really close to c. You’d get decelerated out of nowhere? Just as if you hit something?

Oh boy, special relativity is another fun one. So here's the thing: there's no "universal speed" that you're moving so you're never any closer to c no longer how long you accelerate for. To accelerate is to change your reference frame and there are no special reference frames.

Which is to say that any physical test you could run inside your ship will give you the same result, always. Accelerate for 13 billion years at any rate and check the the how fast light moves within your ship, the answer is always c.

This is where the name relativity comes in. You have to think in terms of relative speed. Your speed relative to earth will indeed advance closer and closer to c but never reach it. There's a bunch of really wild and crazy implications behind this.

Like that acceleration doesn't change the relative speeds of things uniformly. Keep accelerating at 1 meter per second per second and every second Earth's relative speed changes by less than 1m/s. And look up relativity of simultaneity, another consequence of special relativity. It's fascinating stuff.

One mind-altering fact that I love is that there's no "acceleration due to gravity," once you're in free fall, until you hit the ground. Hop in a space ship with no windows and fly off straight in some direction. Turn off the engines and watch an accelerometer. It'll never read anything until you run into something.

You could fly past a planet, a massive star, even a black hole. Your path through space could be full of curves and loops but you'll never feel it. It's popular to think of those things as like crazy high G turns but they're not. You're just flying in a straight line through space time.

On the flip side, say someone knocks you out and puts you on that ship. You wake up and instead of being weightless, you can walk around the ship like normal on earth. Are you on earth or is the ship in space accelerating at a constant rate? Again, there's no way to tell. They are, physically, the same.

I mean you do. All the time. We all do. You're allowed to use them, you're just not allowed to copy them. It's in the name you know, copy right.

I know. But we're both talking about the same thing. Everyone gets irrelevant and ostensibly novel ads all the time. Cat litter, beauty products, diapers, whatever. They just so happen to have focused their attention on cat litter when they just as easily could have focused on dozens of other products and noticed the same result. And, in truth, it's unlikely that they are actually novel, just unnoticed before.

I'd be incredibly skeptical of the claim that they've never been served a cat litter ad. Everybody gets served ads that are misses. They're obviously easy to ignore which makes it difficult to recall what they were about. But I have no doubt that they would've been served cat-related ads plenty of times before. Cats are, after all, one of the most common pets.

So they don't give a shit that he's doing it.

Well half the country either doesn't give a shit that he's doing it or are straight up happy about it. It's literally like watching Hitler rise to power.

I won't hold my breath on enough of them changing their votes to change the poll numbers.

Huh? League is bigger than dota 2. By like a lot. Because League has much more mass appeal than dota. And that was their point, it has a dedicated but niche fanbase.

Hard disagree. There's plenty of games that are little more than dressed up choose your own adventure stories. Plenty that are meant for chill and relaxing gameplay. Plenty that do little more than guide you through horror scenes. And so on.

And even beyond that, most people don't even play a game long enough to have any real "skill development over time." I read from the Civ7 director recently that if you've ever finished a game of Civ you're literally in a minority of the player base. And that tracks with what I've heard about other games as well.

Most players of any given game never finish it. Most of those quit at the first sign of frustration and most are on the easiest game difficulties. This would indicate to me that the majority's conception of "fun" has little to no relation to skill development in the game. They're there for the moment to moment experiences. Rubber band mechanics are there to evoke those fun experiences more often in the majority of the player base.

The thing is I don't think it has anything to offer to bring in people from outside the genre. Some people really enjoy it but you kinda have to already be into that kind of thing (DOTA).

I think you're overstating the importance of games as a platform for skill development as opposed to a platform for, you know, having fun. The fact is that the vast majority of players play any game on one of its lowest difficulty settings.

Rubber banding is made for the core of the game's audience and challenge-seekers are just not large enough to be that core. Some of those rubber banding mechanics can and are disabled at higher difficulty settings. Others are needed at higher difficulty because the AI can't compete and the investment in dev time to improve the AI just isn't worth it because, again, very few people actually play the game at those difficulties.

Yes. Not just "imagine you were given a tool box." Imagine you've been told your entire life that all you're good for is being a basic handyman and you work incredibly hard against people biased into believing that about you to become an astronaut. You finally make it to space. You've done it. You've proven you're good for more than handiwork.

And you get onto the space station and your peers spend their time joking that you actually ought to be the station's handyman instead.

Sad part is that voters trust him more on the economy than the Democrats.

There are multiple facets here that all kinda get mashed together when people discuss this topic and the publicly available/public domain difference kinda gets at that.

• An AI company downloading a publicly available work isn't a violation of copyright law. Copyright gives the owner exclusive right to distribute their work. Publishing it for anybody to download is them exercising that right.
• Of course, if the work isn't publicly available and the AI company got it, someone probably did violate copyright laws, likely the people who distributed the data set to the company because they're not supposed to be passing around the work without the owner's permission.
• All that is to say, downloading something isn't making a copy. Sending the work is making a copy, as far as copyright is concerned. Whether the person downloading it is going to use it for something profitable doesn't really change anything there. Only if they were to become the sender at some later point does it matter. In other words, there's no violation of copyright law by the company that can really occur during the whole "training" phase of AI development.
• Beyond that, AI isn't in the business of serving copies of works. They might come close in some specific instances, but that's largely a technical problem that developers want to fix than a fundamental purpose of these models.
• The only real case that might work against them is whether or not the works they produce are derivative... But derivative/transformative has a pretty strict legal definition. It's not enough to show that the work was used in the creation of a new work. You can, for example, create a word cloud of your favorite book, analyze the tone of news article to help you trade stocks, or produce an image containing the most prominent color in every frame of a movie. None of these could exist without deriving from a copyrighted work but none of them count as a legally derivative work.
• I chose those examples because they are basic statistical analyses not far from what AI training involves. There's a lot of aspects of a work that are not covered by copyright. Style, structure, factual information. The kinds of things that AI is mostly interested in replicating.
• So I don't think we're going to see a lot of success in taking down AI companies with copyright. We might see some small scale success when an AI crosses a line here or there. But unless a judge radically alters the bounds of copyright law, at everyone's detriment, their opponents are going to have an uphill battle to fight here.

Because that's not what copyright is for. It exists to give the creator exclusive rights over distribution. That's it. So unless the company is planning to distribute the work and they obtained a copy willingly and legally distributed to them, then copyright is the wrong law to lean on.

I mean, it's in the name. The right to make copies. Not to be glib, but it really is

A copyright is a type of intellectual property that gives its owner the exclusive legal right to copy, distribute, adapt, display, and perform a creative work, usually for a limited time.

You may notice a conspicuous absence of control over how a copied work is used, short of distributing it. You can reencode it, compress it, decompress it, make a word cloud, statistically analyze its tone, anything you want as long as you're not redistributing the work or an adaptation (which has a pretty limited meaning as well). "Personal use" and "fair use" are stipulations that weaken a copyright owner's control over the work, not giving them new rights above and beyond copyright. And that's a great thing. You get to do whatever you want with the things you own.

You don't have a right to other people's work. That's what copyright enables. But that's beside the point. The owner doesn't get to say what you use a work for that they've distributed to you.

Training literally is consuming. A copyright license doesn't get to dictate what computer programs the work is allowed to be used with. There's a ton a entertainment mega corps that would love for that to be the case, though.

You're saying that you're not allowed to do a statistical analysis on a copyrighted work. It's nonsense. It's well-established that copyright does not prevent that kind of use.

Copyright licensing allows the owner to control how a work is distributed, not how it's consumed. "Personal use" just means that you can't turn around and redistribute a work that you've obtained. Not that you're not allowed to consume it in a corporate setting.