Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Drowning in AI Generated Garbage: the silent war we are fighting

ploum.net

Post reply on HN
Print view View original

341–350 of 452 posts

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#341

Post by drdaeman »

tetraca wrote: post #331

Upload your fake text and images anywhere. Upload it everywhere. Upload it as yourself. Upload it on behest of your willing friends. Corrupt your face in dozens of ways, then tag each as yourself. When everything is garbage, shadow profiles will mean nothing, social credit scores will mean nothing, analytics will mean nothing. Centralization can be turned on its head - the watering hole will be revealed to be the dan…

While I honestly do want to support this approach because unlike the alternative it might (just "might", though) work, I regretfully must also admit that the huge problem with this is collateral damage. There are millions of humans out there that are either highly susceptible to or are already caught into believing all sorts of nonsense, getting scammed in all forms, costing them money, time, well-being and even lives. The issue with fake flood is that until the saturation/breaking point (when the understanding that existing trust system is fatally broken becomes truly mainstream) it's going to make this situation worse.

What's extremely frustrating is that the only alternative I've ever seen proposed are forms of a never-ending moderation arms race, some variant of a "war on fakes". Which is even scarier in the long run, because while it mitigates the damages here and now (good), it would never solve the underlying problem, and only further cement some parts of it (enforced trust invites potential abuse).

And as usual, humanity is currently stuck between two incredibly shitty options (traditionally going through a path of least resistance, no matter what could be at the end of it - because that's how large systems work), hopefully still searching for a magical third silver bullet.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#342

Post by visarga »

nradov wrote: post #190
neilv wrote: post #45

I've been known to start ranting when I'm searching on some topic that attracts SEO garbage (e.g., Python), and realize I'm in the nth different SEO garbage article search hit. I wonder whether a winning solution to the coming massively-parallel "AI" megaspew will emerge, how it will work, and whether Google will be the ones to do it. Or will the bulk of consumers overwhelmingly submit, and accept it as normal, much…

Search engines like Google are probably a dying technology. The next disruptive innovation in that space will be "answer engines" that use GPT like technology to dynamically generate specific answers to questions rather than linking users to third-party sites. This will allow for controlling the entire user experience and serving ads embedded in the generated content.

I don't think there will be ads. We'll be able to run a local chatBot just like we run SD.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#343

Post by wizzwizz4 »

bitL wrote: post #211
wizzwizz4 wrote: post #151

Earlier quoted context omitted.

It will be: GPT output is usually either garbage, repetitive, very wrong, subtly wrong, or inconsistent in writing style in a way humans aren't. (It sort of starts trying to say something different, in the "same style" as it started with, then switches style later .) When humans really can't tell the difference, well… https://xkcd.com/810/ seems appropriate.

Without revealing the origin, the latest models tested in private betas solve many of the points you mentioned.

This is why I'm not leaking all my discriminating heuristics. https://en.wikipedia.org/wiki/Goodhart%27s_law strikes again.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#344

Post by visarga »

2devnull wrote: post #66
marstall wrote: post #48

Earlier quoted context omitted.

> These new tools are awesome if you’re a creator. you're typing "generate me an article about x" into gpt-3. who's the creator here?

Why would anybody need to type anything? Gpt will type the gpt prompts. AI will create the AI that will create the AI. There will be people involved but increasingly less.

We need an AI that writes AI papers, and one that implements AI papers. They could be both chatGPT. Then just let them run in a huge datacenter for a year.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#345

Post by YurgenJurgensen »

pigsty wrote: post #18
poszlem wrote: post #9

While the current speed of the AI research makes me quite uneasy, I cannot really dismiss what we are seeing as: >What we are witnessing is thus not "artificial creativity" but a simple "statistical mean of everything uploaded by humans on the internet which fits certain criteria". at least not more than I would my own brain. I don however fully expect a strong luddite-like backlash happening when this is going to be…

It’s weird how readily so many people will dismiss AI art as just being a product of its input, when that’s really all humans are as well. All artists (myself included) are just the ongoing product of various inspirations (inputs). Dali is a world-renowned artist, but his art wasn’t born in a bubble and ripped from the void. He studied and referenced artists before him, then changed it. Artists who saw his work later…

Only boring humans are just the product of the content they consume; interesting people have concerns about relationships with other humans, organisations, resources, threats and opportunities that come from being an entity living in a world full of things they don't control and things they'd quite like to control. It's also fairly easy to identify artists who have done nothing but look at other art.

Hayao Miyazaki once said "Almost all Japanese animation is produced with hardly any basis taken from observing real people, you know. It's produced by humans who can't stand looking at other humans. And that's why the industry is full of otaku!" While I'm not actually a huge fan of his work, it's easy to see where he's coming from when you see the likes of "In Another World with My Smartphone" or the hundreds of other trash light novel adaptations. They're clearly fiction being produced by people whose lives are nothing but consuming other fiction, and all they can do is reference that. They have nothing to say.

For another example, there's a huge number of 20th century war films. They weren't made because their creators had read a lot of Wikipedia articles about how destructive wars were becoming. The likes of Platoon are a product of mortal fleshbags who actually saw combat and genuinely feared for their lives. Dr. Strangelove was created by people who had a genuine belief that they were at risk of being annihilated in a nuclear fireball.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#346

Post by shagie »

gibspaulding wrote: post #244
Mountain_Skies wrote: post #15

In just the past couple of weeks, I've noticed lots of machine generated cooking recipes. They follow the typical SEO pandering blog format, where there's an unnecessary story about the author's trip somewhere or their child's school play being the inspiration for making this particular recipe, followed by the actual recipe details. The machine generated stories are even more pointless and meandering than what humans…

Holy crap. I just got an account with OpenAI set up last night so I've been looking for interesting prompts to play with. Can't say I'm going to try the recipe, but this looks at least as good as most blog-spam recipe sites. Prompt: Write a recipe and blog post for onion soup. GPT3 Response: Onion soup is a classic comfort food that can be enjoyed at any time of year. It’s a simple and delicious dish that comes toget…

A bit ago, with the Watson craze, they tried to figure out what else it was good for.

This Cocktail Concocted By IBM's Watson Isn't Half Bad - https://gizmodo.com/this-cocktail-concocted-by-ibms-watson-i...

Would You Drink a Cocktail Invented By a Computer? How IBM’s Watson become the world’s most creative bartender

https://www.smithsonianmag.com/smart-news/ibms-watson-can-mi...

---

The reason that blogspam recipes are so common is that recipes have a special place in copyright law in that they aren't.

https://copyrightalliance.org/are-recipes-cookbooks-protecte...

https://www.copyrightlaws.com/copyright-protection-recipes/

https://www.copyright.gov/help/faq/faq-protect.html (and in particular https://www.copyright.gov/circs/circ33.pdf )

From circular 33:

> A recipe is a statement of the ingredients and procedure required for making a dish of food. A mere listing of ingredients or contents, or a simple set of directions, is uncopyrightable. As a result, the Office cannot register recipes consisting of a set of ingredients and a process for preparing a dish. In contrast, a recipe that creatively explains or depicts how or why to perform a particular activity may be copyrightable. A registration for a recipe may cover the written description or explanation of a process that appears in the work, as well as any photographs or illustrations that are owned by the applicant. However, the registration will not cover the list of ingredients that appear in each recipe, the underlying process for making the dish, or the resulting dish itself. The registration will also not cover the activities described in the work that are procedures, processes, or methods of operation, which are not subject to copyright protection.

And so, you can grab a recipe and be 100% sure that it isn't copyrighted. Then you can add some SEO text around it and some advertisements on the side and you've got yourself a page that you are sure that no legal department will come after you for.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#347

Post by bufferoverflow »

keiferski wrote: post #11

People really love to make predictions about technologies way too early. It’s been a year or two, tops, since this stuff has gone mainstream. I very much doubt that the consumer ecosystem will remain the same while content undergoes an AI-powered paradigm shift. We will most likely see a growth of curated sites that help weed out the nonsense, as has been typical for most new developments. And on that note, I think t…

The problem is, Google isn't very curated. All the AI-generated SEO spam is going to pollute its indexes.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#348

Post by shagie »

ghaff wrote: post #252
DoughnutHole wrote: post #227

Earlier quoted context omitted.

Well to me there's two distinct types of background: 1. Background directly related to the recipe/cooking in general. 2. The writer's personal anecdotes tangentially related to the recipe. Some cooking blogs have great background (eg. J. Kenji Lopez-Alt's work on Serious Eats) but I would say they're mostly type 1 - information that informs you about the recipe, how they developed it, technique etc. They have type 2…

Absolutely. There is content around recipes that is good and content that is, well, not. But a lot of people seem to object to anything other than a recipe.

It's because you can copyright everything except the recipe and often a recipe by itself fails SEO in today's world.

Re: Drowning in AI Generated Garbage: the silent war we are fighting

#349

Post by rrobukef »

sam0x17 wrote: post #338
tetraca wrote: post #331

Upload your fake text and images anywhere. Upload it everywhere. Upload it as yourself. Upload it on behest of your willing friends. Corrupt your face in dozens of ways, then tag each as yourself. When everything is garbage, shadow profiles will mean nothing, social credit scores will mean nothing, analytics will mean nothing. Centralization can be turned on its head - the watering hole will be revealed to be the dan…

I'd be careful. By doing this you can use the date these AIs became readily available as a tool for bisecting between "legitimate profiles" before the year X and "potentially fake" profiles from after the year X, which can then be used to very effectively train an AI to tell the difference...

Sure, it also mean limiting the dataset of our children (i.e. the bane of ML), it means bit-rot for existing datasets.
Post reply on HN