There are about 3.6 billion people surfing the web and experiencing PNGs. That use case, consuming PNGs, seems to dwarf the perhaps 100 million (somewhat wild guess) graphic designers, web developers, and photo editing professionals who manipulate images for publishing (in any medium) or archiving.
If, on the other hand, you're considering the use cases envisioned by PNG's creators, or the use cases that interest the people processing or publishing images, yes, these people are focused on format itself and its capabilities.
I suspect this particular use of "use case" isn't terribly clear. Also these two considerations are not incompatible.
I could save my photos as BMPs like early digital cameras did but that doesn't make it practical or reasonable. My camera takes pictures as RAW or HEIF files. Why would I save my photos to a primarily lossy codec that's optimized and designed for distribution rather than preserving fidelity?
We used to do this with JPEG, in fact. And that's why many pictures on Facebook from pre-2018 or so all have a distinctive grainy look. It's artifacts on top of artifacts. Storage on phones isn't tight anymore, we don't need to store photos in a format meant to minimize bytes at the expense of quality.
There's more on Instagram than photos. Lotta meme pages, lot of people just uploading random screenshots and photos they downloaded that have been turned over a million times. Heck, all it takes is someone downloading their own photo from SocialMediaX to reupload on SocialMediaY, or just uploading a the WebP that they exported for their website.
Instagram hasn't even been primarily or even secondarily about photos for a long time. Indeed trying to "just" upload a photo is made super inconvenient these days.
Tangentially related but Instagram is really the worst plattform for photos. I don't understand why they crop and downsize (!) pictures. Not even Twitter does this, it's unironically a better photo plattform.
Unless you're uploading memes you've downloaded from elsewhere, this strictly isn't true. I'd consider myself an Instagram power user and the only thing that I and all the people I interact with is photos and videos. None of those are webp, or would have been worthwhile to save as webp as an intermediate format.
It is easy to dismiss, but the burden of proof would be on the plaintiff to prove that training a model is substantially different than the human mind. Good luck with that.
That makes no sense as a default assumption. It's like saying FSD is like a human driver. If it's a person, why doesn't it represent itself in court? What wages is it being paid? What are the labor rights of AI? How is it that the AI is only human-like when it's legally convenient?
What makes far more sense is saying that someone, a human being, took copyrighted data and fed it into a program that produces variations of the data it was fed. This is no different from a photoshop filter, and nobody would ever need to argue in court that a photoshop filter is not a human being.
I think the problem with "AGI" is that people don't want "AGI," they want Einstein as their butler. A merely generally intelligent AI might be only as intelligent as the average human.
One problem with the paper is that it defines AGI in such a way that if it fails to solve a problem that is inherently unsolvable, AGI can be written off as impossible. It tries to synthesise a definition from different sources whose own definitions don't have any particular reason to overlap in any meaningful way.
I'm just not sure "AGI" is a useful term at this point. It's either something trivially reachable from what we can see today or something totally impossible, depending entirely on the preference of the speaker.
AdSense had that rule when you manually placed the ads on your website. Ever since they started doing automatic placements with AI or whatever, they simply spam the page with ads.
Pretty much all article-based sites, recipes, news, blog posts, anything built with wordpress to blogspot. Their algorithm seems to ensure that there is always 1 ad visible on screen at all times. With font sizes as big as they are these days this means 1 ad every two paragraphs.
And the auto placement is enabled by default on new accounts, and all these new "features," get automatically enabled from time to time. I'm sure there is a mountain of webmasters that didn't even notice that their websites have gotten filled with ads.
The worst one is that interstital that appears whenever you click a link. I'm pretty sure Google had a rule against that type of popup, and then they literally made the popup themselves.
On the other hand, all of this can be disabled.
The question is how much money does a website need to make to stay online. If it could survive with fewer ads, I'm sure there would be fewer.
First it was "I hate how much ad companies track me and build profiles on me."
Now it is "I hate how ads are irrelevant."
People need to understand that ads will never be 100% perfect, otherwise you would buy something every time you saw an ad. 99.99% of the ads will miss the target, and that is normal. It would be insane if it worked any other way.
> First it was "I hate how much ad companies track me and build profiles on me."
> Now it is "I hate how ads are irrelevant."
This is an HN echo-chamber complaint, made by people who work for advertisers trying to come up with a way to make their ads seem less awful.
The fact is, relevant ads aren't better. They're still ads, and ads are still inherently bad.
If I'm looking for a used car, I do not want to hear ads from Bob's Lemon Shop about why they're the best place to buy cars. If Bob's Lemon Shop is the best place to buy cars, I'll find that out from independent reviewers who have shopped their before. An ad from Bob's Lemon Shop is relevant to my interest, but that makes it worse because now I'm susceptible to manipulation by the company that paid the most for ads instead of making a more rational decision based on true information from unbiased sources. Having more relevant ads is not good for me, it's good for advertisers.
Agreed. I've probably got a few thousand hours on youtube, more than just about everything else. It's immensely valuable, yet I refuse to pay for Youtube. Not quite sure why.
I'm perfectly happy paying for two $5 coffees a month that I hardly remember consuming, just because I was perched and a bit tired while on a walk in the city. I pay $25 for a more comfy seat for a 3 hour flight. I pay $15 for a single movie ticket, and another $15 for $3 worth of snacks. I pay $30 for a 30 minute cutting of my hair. I pay $20 for a 3 minute slingshot at the fair. I pay $30 for a 20min taxi.
Yet I refuse to pay $14 for Youtube that I use 30 hours a month, because with adblockers I don't have to. And if Youtube makes adblocking awful enough, I simply will pay. As annoying as youtube ads are, I'd never think to complain about them because it has an easy solution.
> There is a simple way to stop watching ads: pay for premium. It's 100% effective and works right now.
For now. With the ever increasing number of "premium" services that promised no ads, but slowly start introducing them, it is just a matter of time before YouTube does the same.
Yeah the problem for YouTube is that they bundle a bunch of other services with the premium package. They occasionally conduct surveys to gauge user awareness of these features. I myself don’t use any of them, just the ad-free experience.
Thus trying to reintroduce ads to the premium users will remove the only reason I’m paying for it in the first place.
Not the poster, but the point I think that Google is engaging in a clear-cut bait and switch. First, "free email, good UX", "free video hosting, minimal ads". Then, once the dependency sets in, use a standard playbook of degrading the lower tiers and charging for removing the inconveniences.
I am not claiming that Google is the only company doing that; it is not. But there is a reason that bait and switch is illegal in most places. My 2c.
This is not bait and switch, going by the definition on the Wikipedia page. It's closer to 'dumping' where goods or services are supplied below cost to drive out competitors.
> There is a simple way to stop watching ads: pay for premium. It's 100% effective and works right now.
It may be effective at not showing you ads on YouTube specifically, but then you’re helping Google build a more accurate profile on you (from your watch habits) to exploit further. Personally, I’m not comfortable with that because Google has proven time and again that it cannot be trusted.
I would pay for Nebula.tv if it had a few other specific creators.
Using YouTube enough for ads to bother someone does not imply it’s “enough” to pay for. There’s nothing paradoxical about it.
Humans don’t value things as a binary decision between it either being worth it as free or equal to the cost it’s being sold at. Everyone has a price point for a service they think is fair, for which they’ll start seeking alternatives when exceeded. This is how markets work.
Time spent does not correlate with cost-independent value. This is doubly true with social media platforms.
Requires getting out a credit card. Even simpler is an ad blocker.
As for the ethics of ad blocking, I'll consider unblocking ads when Google stops with the unethical (think Tai Lopez) and downright malicious ads (deepfakes of Elon suggesting to invest in crap like "Quantum AI"), and only then will I reconsider removing the blocker and maybe even paying.
Put simply, ad blockers provide a safer browsing experience.
Of course it's nicer to get stuff for free. Leeching is leeching, though, no matter how you try to justify it. Maybe you can find some alternative way to support the creators of the content you seem to be enjoying.
Unrelated to the paper, which is about asking LLM's to figure out which parts of a document were removed, but my assumption has been that to an LLM there is nothing "missing" in the sense that any input leads to valid computation and output.
For example, I asked ChatGPT to explain something I typed randomly
>It looks like you've entered “dosfi8q3anfdfiqr”, which appears to be a random string or perhaps a typo—it's not a recognized acronym, code, or term in any common context I’m aware of. Could you share a bit more about where you found this?
Although the answer is correct, my point is that anything you give to the LLM is going to be put under some bucket. The LLM can't say "I don't know what that is." Instead it says "that is a random string." As far as the LLM is concerned, it knows every possible input and concept that anyone could ever type into it, it's just that its "understanding" of what that means (after the tokens have gone through the neural network) doesn't necessarily match what any human being thinks it means.
This might be due to the system prompt and the training that it is supposed to be "a helpful agent". If you tell it not to ask clarifying questions, you get something more like "I do not understand your input". Tell it to be rude and never ask clarifying questions and I get "What an absolute mess. Fix it yourself"
Funny enough when testing this I also had to tell it to use English. It sees "dos" I suppose and tends to reply with exactly what you saw, but in Spanish.
“It's not a recognized acronym, code, or term in any common context I’m aware of” is pretty similar to “I don't know what that is”. I would assume that a model could be trained to output the latter.
This is news to me. I'm pretty sure the main use case for PNG is lossless transparent graphics.