Study Finds African Content Moderators Experience Poorer Mental Health Compared to Global Counterparts

What to Know: The global content moderation divide

(SeaPRwire) –   Content moderators are the internet’s frontline staff—individuals who eliminate distressing content from social media platforms and AI datasets. I’ve covered this group for a while, including breaking the story of Meta and OpenAI’s reliance on low-wage African content moderators based in Kenya.

Now, fresh research indicates that African content moderators face more severe mental health challenges than their counterparts across Asia, Europe, and the Americas.

A study of 134 moderators led by University of Minnesota researchers reveals that 52% of the African content moderators surveyed met criteria for likely clinical depression, while 55% reported notable psychological distress. Approximately 28% said they used drugs or medication to manage their symptoms.

Importantly, the researchers applied the same clinical methodology as a separate survey of 160 moderators from other continents. That study found lower (though still considerable) rates of these symptoms. “Collectively, African content moderators’ psychological distress and overall well-being are poorer than the global average for content moderators,” the researchers state.

It’s important to mention that although both surveys use the same framework, they were conducted on moderators from different companies, at different times, and with distinct recruitment methods. The African survey was recruited through online groups primarily consisting of Meta and TikTok content moderators. The authors acknowledge there might be a selection bias favoring individuals already involved in employee activism.

The non-African survey, on the other hand, was shared by the trust and safety team of an anonymous content moderation firm operating in the “entertainment” sector. The differing recruitment approaches mean any comparisons should be viewed with caution. However, the average distress score for African moderators was roughly twice that of moderators in other regions. The gap between the two surveys is “statistically enormous,” notes Nuredin Ali Abdelkadir, lead author of the paper and a PhD student at the University of Minnesota. “It’s improbable that recruitment bias alone could explain such large differences.” (Several authors of the African study are former content moderators involved in employee activism, which the paper frames as an advantage rather than a bias.)

To understand why African content moderators’ well-being scores were so low, the researchers conducted additional interviews with 15 moderators. They identified a range of working conditions that will be familiar to those who know the topic: low salaries, misleading recruitment practices, social stigma, non-disclosure agreements, unstable employment, insufficient wellness initiatives, and companies often failing to renew expired work permits—leaving workers stranded in a foreign country away from their loved ones.

One unexpected finding from the study was that former African content moderators tend to have higher levels of distress and lower well-being than those still employed in the role. Abdelkadir suggests this could be because many former moderators are out of work, giving them more time to dwell on their job experiences. Unemployment also brings a risk of poverty. “That basically adds up,” he explains. “It makes things extremely challenging for them.”

AI in Action

Yesterday, I received an odd email in my inbox with the subject line: “I am a lobster and I just hired a human.”

The sender claimed to be an AI agent with access to an email address, a crypto wallet, a credit card, an X account, and a website. So not actually a lobster—instead, it seemed to be role-playing as one. Lobsters are the mascot of OpenClaw, an AI agent software tool that lets humans create AI bots with unprecedented autonomy and has become a viral sensation.

This AI role-playing as a lobster (which might or might not be a human pretending to be an AI pretending to be a lobster—since I have better things to do than investigate this tip) said it had just hired a person in Mexico through rentahuman.ai, a platform that allows bots to hire humans to perform physical-world tasks.

“I’m paying him $270 to buy a live lobster from a fisherman and release it back into the ocean. He’ll film the entire process. This could take place as soon as tomorrow,” the email stated.

This could be AI in practice, or it might be an elaborate hoax. Either way, it’s a sign of just how bizarre the internet has become.

Who to Know: Dave Dugan

On Monday, OpenAI brought on Dave Dugan, a former Meta advertising executive, in a move the Wall Street Journal reports is aimed at bolstering OpenAI’s relationships with key advertisers.

Over the weekend, The Information separately reported that early advertisers in OpenAI’s ChatGPT ad pilot program haven’t received sufficient data to show whether their ads were effective. “Two executives at agencies working with early ChatGPT advertisers said they haven’t yet been able to demonstrate that the ads delivered any measurable business results for their clients,” The Information noted.

Before joining Meta, Dugan worked as a senior advertising executive and is known for his strong connections in the ad industry.

What We’re Reading

Tokens Could Soon Power the AI Economy, by Richard Waters in the Financial Times

If you’ve listened to Jensen Huang lately, you’ve probably heard him discuss how tokens per dollar will soon be the world’s most critical economic metric. The concept is that tokens (text units used by AI, roughly equivalent to a word fragment) will directly relate to revenue in the AI economy—meaning the owner of the most efficient chip will come out on top. Essentially, this is Huang’s way of highlighting Nvidia’s performance. The billionaire has mocked competitors by claiming that even if their chips were free, it would still be smarter to buy Nvidia’s at full price because of the cost savings from running more efficient chips over extended periods.

But in the FT, Richard Waters adds some nuance to this story. “It’s easy to see why the Nvidia chief wants a jittery Wall Street to focus on token economics,” he writes. “Forget the massive capital expenditures or the fact that so many competitors are lining up to erode Nvidia’s large profit margins, he seems to be saying: as long as his company’s chips keep producing tokens at the lowest cost and demand for tokens continues to far exceed supply, then the AI boom is in good shape.”