Pinned
🚨Exclusive: OpenAI used outsourced Kenyan workers earning less than $2 per hour to make ChatGPT less toxic, my investigation found
(Thread)
Billy Perrigo
1,389 posts
Billy Perrigo
@billyperrigo
Tech correspondent @TIME
- Can't quite believe I'm writing this: Today, 150 African workers behind ChatGPT, TikTok and Facebook voted to unionize at a landmark meeting in Nairobi. These AI workers are invisible, underpaid, and the backbone of the tech in all our pockets:
- New: Facebook staff flagged a post by a BJP politician as hate speech internally, but left it online for more than a year until I got in touch to ask about it. My investigation:
- Exclusive: India’s ruling party is creating WhatsApp groups on an industrial scale ahead of elections this spring. My investigation found those group chats include widespread fake news and religious hatred (THREAD) time.com/5512032/whatsa…
- BREAKING: Today, Facebook (Meta) and Sama received legal letters accusing them of multiple violations of Kenyan law, and impending lawsuits, as a result of details revealed in my recent story (thread!) 1/
- Officially, India has the world's second-worst COVID-19 outbreak. Unofficially, it's almost certainly the worst.
- Replying to @billyperrigoTheir working conditions reveal a darker side to the AI boom: that AI often relies on hidden, low-paid human workers who remain on the margins even as their work contributes to a multibillion-dollar industry. (6/8)
- “We will never come back here,” Mohammed Efaz, a Muslim preparing to flee Delhi, told @TIME on Saturday. “The divide between Hindus and Muslims is unbridgeable now.” After the Delhi riots, what next for India's embattled Muslim minority?
- It was already dangerous to be Muslim in India. Then came the coronavirus. My report on the rise of the #CoronaJihad narrative:
- Replying to @billyperrigoThese workers were tasked with reading and labeling text that described, often in graphic detail, situations like murder, child abuse, rape, animal abuse, torture and self harm Content warning: this screenshot includes a description of sexual abuse (2/8)
- 🚨NEW: At Facebook’s external content moderation facility in Africa, employees describe traumatic working conditions, alleged union-busting, and pay as low as $1.50 per hour, my investigation found. 🧵Thread (1/)
- Replying to @billyperrigoThe purpose of their work? Well, without a filter over the top, ChatGPT would spew racism and sexism, just like its predecessor GPT-3. These Kenyan workers were helping OpenAI build that filter. (3/8)
- Replying to @billyperrigoThe lowest-paid among them earned a minimum of $1.32 per hour after tax. Some said they were mentally scarred by what they read, and described mental health support as both inadequate and difficult to obtain. (5/8)
- Replying to @billyperrigoTo get there, OpenAI first needed labeled examples of different types of toxicity: sexual abuse, hate speech, and violence. Every day, these workers told me, they were expected to read and label up to 250 snippets of these types of material. (4/8)
