Mastodawn

Emile van Krieken Jan 19, 2023

Elio Campitelli

Well, fuck. The GPT disinformation age is now.

I googled "OS for 4gb ram" and the first hit, which also was used by google to populate its snippet is an answer from quora which is very obviously created with #chatgpt (I recognised the non-committal non-answer right away, but it can also be detected by a popular GPT detector).

The user has 98 answers and, you've guessed it, they are all created with GPT.
PLOT TWIST: The questions where also created with GPT!

https://www.quora.com/profile/Heri-Mulyo-Cahyo

Heri Mulyo Cahyo

Following 444 people. Asked 455 questions. Wrote 98 answers.

Quora

billseipel Jan 17, 2023

@eliocamp My god. The internet just got noisier.

Danielle Navarro Jan 17, 2023

@eliocamp it’s going to be a wild ride. I have already resigned myself to the fact that gpt-derived models are going to make large swathes of the web useless for a long time. It was already happening on twitter before I left - I used to get gpt based bots show up to argue with me about trans rights. They were pitch-perfect at riling people up, and not easy to detect. I’m expecting more of that in the future

Elio Campitelli Jan 17, 2023

@djnavarro That's sad. I wonder how long until most of internet traffic is bots talking to other bots.

Michael Sumner Jan 17, 2023

@eliocamp @djnavarro beep boop

Funky Gibbon Jan 17, 2023

@eliocamp @djnavarro I imagine hackers are working on this right now as the new DDOS

Ad Gerrits Jan 18, 2023

@djnavarro @eliocamp and when they create a humansfilter.

freeformz 🏳️‍🌈Jan 18, 2023

@eliocamp @djnavarro is that how the singularity starts? (J/k)

Michael Sumner Jan 17, 2023

@djnavarro @eliocamp that's hell, but also a perfect capture of the slow decline since a golden era (~10-12 years ago the internet and tech was *excellent*)

Michael Sumner Jan 17, 2023

@djnavarro @eliocamp I guess you know we're fucked when "I've just spoken to ..." becomes "I've Judy spoken to ..."

like, is it the same guys coding at night in their underpants doing ai? I'd kind of elevated that as the problem, but really we're rallying around "who in reality has an opinion and output that I trust"

ahistorical immaterialist Jan 17, 2023

@mdsumner @djnavarro @eliocamp I hate that nowdays, half the things you search for information on return a YouTube video for you to watch rather than just text to read... But it's occurred to me that a video of a real live human actually bothering to speak about something is now just about the only thing you can actually (still) trust to be real!

Charlie Stross Jan 18, 2023

@mdsumner @djnavarro @jimbob @eliocamp

You know about DeepFake videos, right?

Some folks are probably working RIGHT NOW on hooking up a GPTChat text back-end to a DeepFake front end using AI-generated "generic attractive vlogger" personnae to narrate the bogopasta, for advertising revenue.

BeeCycling has moved Jan 18, 2023

@cstross @mdsumner @djnavarro @jimbob @eliocamp
Dead Internet theory looking less like a conspiracy theory now.

WizardWes Jan 18, 2023

@beecycling @cstross @mdsumner @djnavarro @jimbob @eliocamp
Aight, time to just move over to gopher 100%. Hopefully it'll be a bit more sheltered

Kit Bashir Jan 18, 2023

@cstross @mdsumner @djnavarro @jimbob @eliocamp well, at least we can relax now that we know the solution to Fermi’s paradox. #TheGreatFilter #YoureSoakingInIt

Michael Sumner Jan 18, 2023

@cstross @djnavarro @jimbob @eliocamp maybe you're that

Michael Sumner Jan 18, 2023

@cstross @djnavarro @jimbob @eliocamp ok had a closer look and I'm more sure 🙏😃

Bastien Quelen Jan 18, 2023

@cstross @mdsumner @djnavarro @jimbob @eliocamp You mean like this https://www.brut.media/fr/economy/brut-ia-voici-les-dernieres-infos-sur-le-mouvement-de-greve-de-jeudi-contre-la-reforme-des-retraites-12783cb3-5caa-4078-acfc-369a5aedfd2e (sorry it is in french). They create a avatar from a journalist and dialog generate by Chatgpt from a AFP news.

Brut.IA / Voici les dernières infos sur le mouvement de grève de jeudi contre la réforme des retraites | Brut.

Brut.IA / Voici les dernières infos sur le mouvement de grève de jeudi contre la réforme des retraites. 👉 Une vidéo d’information créée par l’IA, c’est possible ? C’est ce qu’on a voulu tester. Pour la première fois, Brut expérimente avec cette publication les nouveaux outils utilisant l’intelligence artificielle (IA). Pour fabriquer cette vidéo, nous avons créé un avatar de notre journaliste Rémy Buisine (avec son accord, bien entendu). Nous avons ensuite utilisé la technologie de ChatGPT pour générer un résumé d’articles de l’Agence France Presse. Ce résumé, produit par l’IA, a été vérifié, corrigé et validé par un journaliste de Brut. C’est un premier test et pour vous comme pour nous, c’est probablement surprenant comme publication, alors dites-nous ce que vous en pensez. On va continuer d’explorer ces nouvelles technologies pour vous informer du mieux possible. Posez-nous toutes vos questions en commentaires, nous y répondrons avec plaisir.

Brut.

Chris L Jan 18, 2023

@jimbob @mdsumner @djnavarro @eliocamp The problem is that there are a lot of highly motivated frothing idiots putting out YouTube content.

"How do we find information we can trust?" is a real problem and only getting worse.

CurtAdams Jan 21, 2023

@jimbob @mdsumner @djnavarro @eliocamp Unfortunately YouTubers are often poorly informed, especially influencers, and they're going to be reading what they Google off the net, which will be full of chatbot nonsense.

Sunflower Björnskalle 🌻Jan 17, 2023

@djnavarro @eliocamp This could be really great if you think about it... We can develop new ways of discovering content that rely on trust and networking of actual people. This could be a really good upgrade for a lot of people's day-to-day epistemology.

Fangh Jan 18, 2023

@apodoxus @djnavarro @eliocamp really funny that you have a sunflower in your name because there is a service that aim to promote good content on YouTube and it is called https://tournesol.app (which means sunflower in French).
@lenhoang could explain it way better

Tournesol

Compare online content and contribute to the development of responsible content recommendations.

Tournesol

Sunflower Björnskalle 🌻Jan 18, 2023

@Fangh @djnavarro @eliocamp @lenhoang This is exactly the kind of thing I'm talking about :)

Alexandre B A Villares 🐍Jan 18, 2023

@djnavarro @eliocamp https://botsin.space/@scream/109709199540529896

Endless Screaming (@[email protected])

AAAAAAAAAAAAAAAAAAAAAHHHHHH

botsin.space

Alka Jan 18, 2023

@djnavarro @eliocamp "I couldn't argue with someone that disagrees with me so Im going to call them a bot"

Alexander Ruoff 🇪🇺Jan 18, 2023

@djnavarro @eliocamp wouldn't it be possible to use the same tools to argue in favour of trans rights, climate crisis and transformation of mobility? Fight them with their own weapons.

Brian McCallister Jan 18, 2023

@djnavarro @eliocamp @cstross realizing Vinge’s Friends of Privacy?

Cassandrich Jan 18, 2023

@djnavarro @eliocamp We desperately need notarized hashes of information resources known to exist before 2023.

Ben Jan 18, 2023

@eliocamp @djnavarro cool, back to RSS of trusted authors we go

CurtAdams Jan 21, 2023

@djnavarro @eliocamp The worst part is this flood of AI chatbot nonsense is going to get into the training for AI cbots and they will get even worse. Not only will we be drowning in nonsense, it's going to be festering nonsense.

Dizzy Jan 17, 2023

@eliocamp It is going to get a LOT worse before it gets better. Freeloaders will exploit this like crazy; just like they are doing with TikTok, Youtube etc...

@eliocamp I struggle to remember the last time I've seen a good answer on quora though

Giorgio Sidari Jan 17, 2023

RetroDevDaemon Jan 17, 2023

@eliocamp Ive stumbled upon this myself while searching for coding tidbits. The articles use the same pattern of talking "around" the information, never actually listing it, referencing figures or tables that don't exist, etc.

I hate our planet now

SSFC KDT (MOVED)Jan 17, 2023

bazkie bumpercar | unfluencer Jan 17, 2023

@eliocamp hopefully, this leads to people no longer believing everything they read on the internet.

..yeah I think I'm being a bit naive there :P

Jigen Jan 17, 2023

@eliocamp On par with preexisting non-AI results such as:

Matt Hodgkinson Jan 17, 2023

@eliocamp Haha. Quora is going to be even more fun now!!! /s

@williamgunn, another headache for you, I guess.

#ChatGPT #Quora

William Gunn Jan 17, 2023

@mattjhodgkinson @eliocamp Thanks for flagging, y'all!

MissScott01 Jan 17, 2023

@eliocamp
Now we will have to spend even more time validating anything we read.

@eliocamp Dunno if you’ve found any actual suggestions, but I’ve installed Xubuntu on a couple of old laptops over the years and it’s performed well.

Henrique C. S. Junior Jan 17, 2023

@eliocamp Well, I guess the only possible reaction is social media and popular websites in general scanning and warning or banning when a text is GPT generated. This is already possible with some degree of confidence.

13 barn owls in a trenchcoat Jan 17, 2023

@eliocamp we're overdue a return to curated search indexes

John Gordon Jan 18, 2023

@HauntedOwlbear @eliocamp

I’m starting to think very few people care about useful search.

13 barn owls in a trenchcoat Jan 18, 2023

@jgordon @eliocamp

This is the inherent problem with the corporatisation of the web - when your customers are advertisers, the only concern is showing adverts, rather than providing information.

Same mindset that brings us this:
https://gizmodo.com/cnet-ai-chatgpt-news-robot-1849996151

I've been making stuff that's been broadly categorised (by other people) as "online content" for about two decades, and since the earliest entry of large business into the space, absolutely no shits have been given by ad sales people about the stuff that sits alongside the adverts.

CNET Is Reviewing the Accuracy of All Its AI-Written Articles After Multiple Major Corrections

Big surprise: CNET's writing robot doesn't know what it's talking about.

Gizmodo

13 barn owls in a trenchcoat Jan 18, 2023

@jgordon @eliocamp

Anyway, I'm specifically thinking we need something like that old Yahoo search index alternative that had a lizard for a logo and a four-letter name.

(which I've entirely forgotten the name of - extra embarrassing given that I used to volunteer on it)

John Gordon Jan 18, 2023

@HauntedOwlbear @eliocamp

I vaguely recall.

I also remember a Yahoo! search experiment with a slider that let you tweak how much ad supported content was returned in search... (or something like that. All the developers disappeared shortly afterwords.)

Antonio Páez 🇲🇽🇨🇦Jan 23, 2023

@HauntedOwlbear @jgordon @eliocamp

What Corey Doctorow calls "enshittification"

@eliocamp I love watching the internet break, it will usher in an age of intellectualism. Information will have to be painstakingly vetted to be trusted. GPT detectors are often wrong about 50% of the time.

grayrattus Jan 17, 2023

@chris @eliocamp where do you have this statistics from?

Ted Han ★ 韓聖安 Jan 18, 2023

@eliocamp Has anybody flagged this for @cstross yet. 'cause he's starting to move from proving that he's a wizard from the near future to proving he's a wizard from the far future.

arthur.pizza Jan 18, 2023

@eliocamp The only difference now is that the SEO'd garbage results have less typos. It's time to remove Google.

wakingrufus Jan 18, 2023

Googling any very specific question about a video game is almost totally useless now.

ai-Paul E Jan 18, 2023

@eliocamp sheesh! AI-generated, AI-amplified bunkum loop! 🥴🤦🏽‍♂️

Doug Holton Jan 18, 2023

@eliocamp guess we're going to have to bring back Google Answers where we pay librarians & others to find information or answer questions for us https://en.m.wikipedia.org/wiki/Google_Answers
🎵 toss a coin to your researcher🎵

Google Answers - Wikipedia