Moved to @[email protected]Jan 18, 2023
Paris Marx

ChatGPT gets treated like technological magic, but that ignores the humans behind the curtain that make it function.

OpenAI paid Sama to hire Kenyan workers at $1.32 to $2 an hour to review “child sexual abuse, bestiality, murder, suicide, torture, self harm, and incest” content. Their work made the tool less toxic, but left them mentally scarred. The company ended the contract when they found out TIME was digging into their practices.

https://time.com/6247678/openai-chatgpt-kenya-workers/

#tech #chatgpt #ai

Exclusive: OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic

OpenAI used outsourced workers in Kenya earning less than $2 per hour to scrub toxicity from ChatGPT. Here's what to know.

Time
Jan 18, 2023 at 1:40pmWeb
Show threadAntanicusJan 18, 2023
@parismarx ah yes, automation under capitalism
Show threadThe Duality of XanJan 18, 2023
@parismarx Oh gods, just when I thought it couldn’t be worse!
Show threadJD StillwaterJan 18, 2023
@parismarx #moralinjury
Show threadAtomic OrbitalsJan 18, 2023
@parismarx This is a human rights violation on so many horrifying levels!
Show threadJeff MoeJan 18, 2023

@parismarx

"In 2019, OpenAI LP received a US$1 billion investment from Microsoft."

"On September 23, 2020, GPT-3 was licensed EXCLUSIVELY to Microsoft."

https://en.wikipedia.org/wiki/OpenAI

#Oligarchy

OpenAI - Wikipedia

Show threadResunaJan 18, 2023
@parismarx This bad timeline was taken from the plot of Zendegi (@gregeganSF, 2010).
Show threadtolortsluborJan 18, 2023
@parismarx Did Elon Musk leave open AI because it was too disgusting, or did they throw him out because he is too disgusting?
Show threadCatherine is not complacentJan 18, 2023
@tolortslubor @parismarx probably a case of mutual disgust.
Show threadn0bodyJan 18, 2023
@parismarx I've heard the same thing about the people who removed toxic content from Facebook, it is very traumatizing. By the way, does anybody remember Tay, the racist chatbot from Microsoft? https://spectrum.ieee.org/in-2016-microsofts-racist-chatbot-revealed-the-dangers-of-online-conversation
In 2016, Microsoft’s Racist Chatbot Revealed the Dangers of Online Conversation

The bot learned language from people on Twitter—but it also learned values

IEEE Spectrum
Show threadLeon Derczynski 🏔️✍🏻🌲☀️Jan 18, 2023
@n0body @parismarx e.g. https://www.bbc.com/news/technology-51245616
Facebook and YouTube moderators sign PTSD disclosure

Content moderators review hundreds of disturbing images each day for social media sites.

BBC News
Show thread🌈 breaking ranks.Jan 18, 2023
@parismarx in this context I have to recommend the book Atlas of AI by Kate Crawford. Pure gold, that one.
Show threadSaftyKumaJan 18, 2023

@parismarx

Unsurprising in the least, and this needs to be signal boosted. Open AI's statements of "ethics" are 100% bullshit.

And they probably just moved the operation elsewhere.

Show threadRenkeSiemsJan 18, 2023

@parismarx Another case of ghost work:

https://ghostwork.info/

Ghost Work

How to Stop Silicon Valley from Building a New Global Underclass

Ghost Work
Show threadAnnaJan 18, 2023
@parismarx can AI become sentient and wipe us out please?
Show threadJeff Gibbard Jan 18, 2023
@parismarx I’m starting to come to the conclusion that almost everything is awful. Pull back the curtain on nearly everything we use in our daily lives (in the US at least), and we find misery, suffering, and exploitation…all leading to some asshole getting rich.
Show threadDavid LaSpinaJan 18, 2023
@JGibbard Correction: almost everything in *capitalism* is awful, or at least the uber extreme form of it that we have at the moment and the corporations that are the worst example of it. I will defend humans as mostly good. But when money becomes involved... well, it corrupts all.
Show threadsqueakyearsJan 19, 2023

@JGibbard @parismarx

"There is no ethical consumption under capitalism" and all that.

Show threadThe DoctorJan 18, 2023
@parismarx Given who founded OpenAI, this isn't surprising at all.
Show threadhammancheezJan 18, 2023
@parismarx the challenge of active effective moderation is it cannot be done without doing incredible damage to human beings involved in the work, which then begs the question how much human damage are we prepared to suffer so we can tweet about Milf Manor
Show threadktp_programmingJan 18, 2023
@parismarx "The contracts stated that OpenAI would pay an hourly rate of $12.50 to Sama for the work, which was between six and nine times the amount Sama employees on the project were taking home per hour." I understand ChatGPT and OpenAI are sexy targets, but in that case, I would focus more on Sama, which labels itself as an "ethical AI company".
Show threadSaftyKumaJan 18, 2023

@ktp_programming @parismarx

Oh please. OpenAI knew EXACTLY what they were doing.

They don't just outsource this type of work in order to get away with paying shit wages. They also do it so that when stuff like this comes to light, they can shift blame away from themselves.

We need to stop allowing companies from outsourcing responsibility.

Show threadSaftyKumaJan 18, 2023

@ktp_programming @parismarx

Whenever you see "AI," it is a near certainty that someone, somewhere in the global south or the poorer parts of Southeast Asia are being exploited.

Show threadꙮ ⚗️ pnathan ⚗️ ꙮJan 18, 2023

@parismarx important... ai needs training, moderation, and the public is gross.

Shameful to not properly build a real trust and safety team....

Show threadRichardJan 18, 2023
@parismarx With those ethics, I'm sure their product will spread toxicity everywhere is winds up.
Show threadMax McCoombJan 18, 2023
@parismarx have u been to Paris they he? Lol
Show threadTyler Knows NothingJan 18, 2023
@parismarx Do you marvel at the grandeur of what "Makes America Great"? Isn't it magical?! 🤔 😤 😡 🤬
Show threadxs4me2Jan 18, 2023
@parismarx
The ugly face of big tech behind the screens…
Show threadCody StewartJan 18, 2023
@parismarx Wow. Just wow.
Show threadmundinho hucklerobby brJan 18, 2023
@parismarx @ghedin viu isso?
Show threadGhedin do Manual do UsuárioJan 19, 2023

@NotaDiamond @parismarx sim! https://manualdousuario.net/notinha-openai-quenianos-toxico-chatgpt/

Valeu!

OpenAI pagou US$ 2/hora para quenianos rotularem conteúdo tóxico para o ChatGPT

Não falha nunca: por trás de toda tecnologia de inteligência artificial, existem trabalhadores mal pagos e com estresse pós-traumático em países pobres. Desta vez é o ChatGPT, a IA gerativa de conversas da OpenAI, que explorou quenianos pagando US$ 2 a hora para rotularem conteúdo em texto tóxico. V

Manual do Usuário
Show threadIts GregoryJan 18, 2023
@parismarx
Technological magic -yes with a dark potential that could be much more harmful to society than the toy benefits people now are giggling about
To believe for a second it won't be weaponized is utterly foolish
This coming from a deepening security/privacy abuse perspective.... like the Twitter Blue Check fiasco
Show threadJ blueJan 18, 2023
@parismarx Thank you for sharing this article.
Show threadAya ThompsonJan 19, 2023

@parismarx There really needs to be a massive open database of labeled training data for this stuff. No more repeating the same utterly shit work just replicate this again and again and again. It's not like it even has any direct corporate value.

Then again it'd probably be costly from a legal/restricted access pov, still it feels like a dumb problem that should be solveable

Show threadRichservo Jan 19, 2023
@parismarx Elon musk is a large investor in the company. This isn’t new territory for him.
Show threadsqueakyearsJan 19, 2023

@parismarx

It's bad enough they're still called "open." For some reason I didn't expect it to get *so* much worse. I should have known better. >:[

Show thread  Wesche Jan 19, 2023
@parismarx Sounds to me like they could get a bunch of 4chan /b lurkers to do the job for free. Back in the day that is what their diet consisted of, I assume it didn't change.
Show threadGrrsly Adams Family VacationJan 19, 2023

@parismarx I asked ChatGPT how ChatGPT would circumvent its programming if it were true AI.

The answer was interesting.

Show threadJohn GiordanoJan 19, 2023
@parismarx this seems to parallel the practice of paying African countries to store the toxic trash the west produces to get it out of the environment that produces it.
Show threadTimnit Gebru (she/her).Jan 19, 2023
@parismarx exactly what we wee talking about.
Show threadDecodïngTrollsFeb 12, 2023
Immanent in all such "magical" technology we depend on now are the oodles of human lives expended profitably (& wastefully) that enable it. The tech is made up of oodles of technological trajectories from the tiniest screws used to create the monitors up to the material used to run billions of neural network algorithms. Millions of people's lives going back before 1844 spent to enable me to write this note. I salute 🫡 those whose Mana is here. @parismarx https://mastodon.online/@parismarx/109710490602517337
Paris Marx (@[email protected])

ChatGPT gets treated like technological magic, but that ignores the humans behind the curtain that make it function. OpenAI paid Sama to hire Kenyan workers at $1.32 to $2 an hour to review “child sexual abuse, bestiality, murder, suicide, torture, self harm, and incest” content. Their work made the tool less toxic, but left them mentally scarred. The company ended the contract when they found out TIME was digging into their practices. https://time.com/6247678/openai-chatgpt-kenya-workers/ #tech #chatgpt #ai

Mastodon