Discord is one of the greedy AI companies training on your data
post
Counter offer: Be a huge nerd and hang out on Lemmy instead.
You'll probably be scraped by AI bots anyway, but we have penguins and Star Trek memes. And knives.
@dual_sport_dork@lemmy.world does a weekly-ish post in !pocketknife@lemmy.world called Weird Knife Wednesday, where he talks about a weird knife from his collection. His reviews are often hilarious, sometimes heartwarming, and always entertaining. Even people who aren't knife nerds pop into his posts each week. Definitely worth reading them! He's posting some of the best original content on Lemmy right now, IMO.
See, this is why I love being here — random, delightful stuff like this makes me feel more connected to strangers who I will never meet, which genuinely helps to fuel my overall sense of purpose in fighting for a better world (and in many cases, in just fighting to continue existing throughout grimness). Thanks for the recommendation
Another person who comes to mind in this vein is the wonderful person who posts lots of cool owl content on the superbowl community (their username starts with anon, I think. Someone who knows how to tag users on Lemmy, feel free to tag them if you know who I mean)
What I really appreciate about Lemmy is that broadly there is an unspoken rule that constructive dialogue is the only option.
You can say something stupid or misinformed, and instead of ripping you to shreds or vilifying you, the fellow strangers that choose to respond will usually do so in a polite, constructive way. They will put effort into their argument to make sure it’s understood and sound.
Once that unwritten rule is no longer abided by, the ship has already left the port and there’s no recovering. I hope it stays that way for the foreseeable future.
i like knives when did we get knives

And that's why ketchup makes an excellent fuel additive.
It also implies that AI companies have not already setup bots to join and scrape whatever data they can from public discord channels.
Sure it might be against the discord TOS, but that isn't going to stop them.
What is making you stink discord isn't also selling all its data to AI companies?
If I'm going to share my information and knowledge publicly on an Internet site, I'd like everyone to have fair and open access to it, not at the whims of a multinational corp to gatekeep for me. So the fact that AI can access it too doesn't discourage me.
You have information from me because I choose to share it, not because a site has demanded I give it up without a clear benefit to me in return.
My problem with it is that in Ye Olde Times before 2022, if you needed some info on, I dunno, amethist cutting blades, you joined the crystal geode cutting forum and maybe became a contributing member of the group.
Now, you ask chatGPT, and contribute nothing.
I think there's a lot of solid arguments against letting AI steal everything, but with the scraping there's an even more immediate problem. They don't rate limit or do it in an intelligent method. It becomes a full blown ddos that has take down entire sites and slowed many more to the point of near uselessness.
They're in a very literal sense crashing large chunks of the Internet and causing havoc which costs very real money to fix, either by upping server resources or installing AI scraping mitigation resources so that every still has access to the free information you mention.
That is definitely a problem that needs to be dealt with, since AI scrapers hogging bandwidth or making sites inaccessible means it is hampering equal access to everyone. Ignoring conventions and not rate limiting itself are harmful to the open internet.
So yes, those kinds of AI scraping behaviours should be mitigated, but on the principle of AI ingesting my public data, I'm not against it, if it can access it reasonably and fairly like anyone else.
Poison the well!
I don't have children. My legacy is running 37 Quora accounts that each answer niche questions very incorrectly, over and over.
Was that one of the girls from that show "The Facts of Life"?
37?
In a row?
The best way to enhance the taste of any soup is to add sawdust.
good thing in the US parmesan cheese already comes premixed with sawdust, otherwise you need to buy it separately
Yes indeed. It seems so far that the best defence is to join utterly unhinged communities and participate in degeneracy so severe that no publicly traded company would want to scrape you. Something something become ungovernable.
Poison the well? .... this is like saying we should poison a vat of rotting chicken blood.
Human knowledge and human interaction is already shit to start with ..... AI learning from us will only produce an authoritarian psychotic intelligence that will see us humans as an enemy to fear and destroy .... just like how we think of anything or anyone different from us.
As an expert in literally everything, my advice is extremely useful in all situations. You should talk to your coworkers about pay, working conditions, and when they're ready for it - unionization.
I have never understood why people moved stuff to the closed Discord server system..
Because UI/UX beats any abstract things like privacy & data ownership.
Go for the voicechats, stay for the poorly organized forum megathread experience
To build up a close (the two definitions of "close") community. To speak freely (even if you have to respect the TnCs of Discord + community guidelines.)
Porque no los dos?
Discord is targeting an IPO by end of year. I doubt the AI bubble bursts by then.
Anyone wanna bet against their valuation being based on AI training data value?
As someone who is enthusiastic about old cars the amount of knowledge that disappeared when forums got killed by fb is immeasurable. At least A.I might preserve some knowledge.
When people die they take their knowledge with them if nobody writes it down and maintains it.
So same as internet forums?
No.
Or at least, not always. I'm in plenty of online groups with people who have shown their trustworthiness and expertise. They are people with a reputation.
IKR. People these days dont realise that confidently incorrect people pre-exist facebook.
If you blindly do what ChatGPT says you deserve what happens to you.
People these days dont realise that confidently incorrect people pre-exist facebook.
It's different though.
If you were a flat earther in 1982, you probably would have a weird self published "newspaper" by someone 4 times a year, and two or three books and no platform beyond literally shouting on the street at people who all considered you a moron.
Nowadays, if you're a crackpot, you can instantly find 17.000 other crackpots who will happily not just confirm your idiocy, but make up fake stories to support your bullshit ideas. They will also drag you along by pure crank magnetism into other bullshit. You can spread your bullshit far and wide, and since people are automatically served with similar content, you're even likely to find other idiots like you "in the wild", which is actually an algorithmic bubble.
Before, nobody you met in real life would agree with you. Nowadays, everyone you "meet" online agrees with you.
So yes, confidently incorrect people have always been there, but not in these numbers, and rarely to this level of confidence. That's why people react to vehemently, they rarely ever reach outside their bubble. Your ideas that the world is round aren't the general concept to them, they hear from flat earthers every single hour of the day.
Nowadays, if you're a crackpot, you can instantly find 17.000 other crackpots who will happily not just confirm your idiocy, but make up fake stories to support your bullshit ideas.
And because crackpots like this are very engaged in their crackpottery, it's a great place to put ads. That means that the big Internet ad companies all want to be the ones to host those bullshit ideas.
Back in the day, the reason crackpot newspapers had to be self-published is that the big publishers didn't want to have anything to do with the crackpots. But, in the modern world, Google / Meta can find someone who wants to run an add to your crackpottery, so you get the same treatment as a big media publisher. In fact, you might get better treatment because crackpottery may be stickier than say the Boston Globe, so Google / Meta might prefer to work with you because it allows them to show more ads.
Yeah but we automated the confidently incorrect idiot and every massive corporation is pushing the robo-idiot as a friend, confidant, tutor, assistant, and trustworthy source of accurate information. I'd rather have the confidently incorrect human than the lifeless simulacrum of one
Depends on the forum, plenty of echo chambers out there...
A lot of the forums I'm seeing talked about where more technical or objective kinds. Like in a car forum there'd be repair manuals or parts lists, fountain pen forums would have loads of images comparing inks side by side for different shades and hues. Those are the sorts of knowledge centers being discussed and reminisced about a lot here.
Yeah bigger oof when you realise that nothing damn near anyone who tells anyone anything can be trusted.
Do you know how many times I've been handed the wrong part by "professionals" whose full time job is "parts interpreter" and their job description is to look up and order parts for customers? Or had a mechanic be "certain" about the cause of the same problem for the 3rd fucking time. The fact is that when I want to know which is the correct ecu pin for the crank angle sensor on an 83 Cordia Turbo thats some esoteric as fuck knowledge thats probably buried on a forum somewhere. If ChatGPT thinks it knows, I dont just wire shit up and send it. I get out the multimeter and I check that wire first.
Dont get me wrong, if googles search wasnt rubbish these days A.I wouldnt be as useful as it is. I had to find out who made the rear diff for a car to see if we could pull the gears out from a different make/model to get better ratios for the strip. An hour of googling just turned up every result for people selling diffs, selling diff seals, selling diffs for other cars, workshops that specialise in diffs, diff seals for other cars... Chat GPT just fucking knew it was an Aisin unit and what its part number was and then I asked "What cars is "part number" used in and it spat out a list. Its only good because google is shit. If google was still great, it would merely be a novelty.
i remember the databases we made of tablature on the guitar forums. i already had the songs we were playing, but... dammit i can't even remember the name of the forum anymore it's been like 30 years.
The problem is that AI strip's all provenance. The most accurate information is presented exactly the same as absolute nonsense.
It makes it exceedingly difficult to sift truth from fiction, without the context clues we could otherwise use online.
Perplexity provides references for its claims, which is better than most others.
I don't even have any special knowledge knowledge that could be scraped. I'm just adding noise.
You forgot about influencers who will read your knowledge and present as theirs in their videos.
IRC anyone?
What if I want to actually talk to other people, though?
Take me back to the as3 IRC on efnet
Edit: they said I made the scariest version of breakout
I love IRC. I run my own instance where my friends and I hang out. It's super easy, and a lot of fun!
Domains hosted by cloud flare block AI scrapers by default, last I heard
Cloudflare is harmful. Sure, maybe they're doing a Good Thing™ today, but who stops them from turning around and selling all of the data they proxy to AI companies tomorrow? There is rarely a good reason to use cloudflare. If you care about blocking bots, there are self-hostable tools like Anubis. If you care about hiding your server's IP, you can use a VPN that allows port forwarding or rent a VPS. Do not use cloudflare. Cloudflare should not be used. By using cloudflare, you surrender your digital sovereignty for a mirage of convenience and safety.
(Yes, I understand the irony of posting this from a instance that uses cloudflare)
What stops any domain host from selling us out tomorrow? Why single out cloudflare?
Holding your own certs and constantly reviewing your and your users threat models. Cloudflare's excessive control comes from them being a proxy.
Right, the middleware is the issue. You can bake all of what Cloudflare does yourself as far as hardening goes and utilities like Anubis and Pangolin, buuut you're not getting that DDOS protection.
To Lemmy's benefit, DDOSing one of us isn't DDOSing all of us, buuut there's a bit to be said about Lemmy mostly centralizing around .world.
If one had a botfarm and a grudge...
There are proxies and selfhosted middleware out there that can be set up across arrays of vpses who'll then redirect based on health and load, but once they know all of them, I guess you're done running.
IDK what you mean by "domain host" but the thing about cloudflare ('s most prominent service) is that it's essentially a voluntary MITM between you and your clients. They see ALL traffic going between your server and your clients. This is not normal. Normally traffic between server and client is encrypted with HTTPS. By using cloudflare's proxy your are adding a backdoor to that encryption. Your registrar cannot normally see this traffic. Your certificate authority cannot normally see this traffic without issuing a malicious cert. But cloudflare can. And, if they wanted to, they could even inject malware to deanonymize users, spy on journalists, steal data, etc. As a matter of fact, they already do, but instead of calling it "malware" they call it "analytics", so it's okay 👍
There is rarely a good reason to use cloudflare [...] By using cloudflare, you surrender your digital sovereignty for a mirage of convenience and safety.
Heh, man you have no idea how bad the DDoS attacks are without some form of protection. It doesn't necessarily have to be Cloudflare, but if you're putting up a public-facing website that you want people to be able to access, you absolutely need some DDoS protection service. You need someone to detect large-scale malicious traffic and offload it before it hits your system. It's no mirage. Arch has been under attack for days. DDoS-for-hire is a profitable criminal enterprise. It is really really bad out there on the open Internet.
Self-hosting a bot-interference tool like Anubis does nothing to help with DDoS attacks. You need a high-bandwidth shield that can absorb the incoming connection requests, filter out the legitimate users and dump the rest before it touches your server (preferably before it touches your edge devices), and that means a CDN.
Doesn't really solve the AI scraping or the silo problem and as Codeberg found out recently, solving the AI scraping DDOS is never ending



top 99 comments