Training AI takes heavy toll on Kenyans working for $2 an hour | 60 Minutes

Summary

This video script explores the hidden human labor behind AI development, focusing on workers in Kenya who label data to train algorithms for tech giants like Meta, OpenAI, and Microsoft. Despite their crucial role, these workers endure harsh conditions, low wages (around $2/hour), and psychological trauma from disturbing content. The script highlights the exploitation of labor in regions with high unemployment rates, where workers are often misled by promises of good opportunities. It calls attention to the unethical practices in outsourcing AI tasks, with companies reaping massive profits while workers face exploitation and mental health challenges.

Takeaways

• 😀 AI is not making humans obsolete; millions of workers, known as 'humans in the loop,' are still needed to help train and improve AI systems.
• 😀 Many AI training jobs are outsourced to countries like Kenya, India, and the Philippines, where workers are paid very low wages for complex tasks.
• 😀 In Kenya, the unemployment rate among young people is as high as 67%, leading many to take AI jobs despite poor conditions and low pay.
• 😀 Workers like Naali Walo in Kenya label images and videos to teach AI, but these jobs are mentally taxing and pay only about $2 per hour.
• 😀 Despite being tasked with difficult and emotional work, like identifying disturbing content for AI moderation, workers receive minimal support and counseling.
• 😀 Some workers report severe mental health issues, such as PTSD, after long hours spent viewing violent and explicit material.
• 😀 The labor conditions in these AI training jobs are compared to modern-day slavery due to the exploitation of vulnerable workers.
• 😀 Workers in Kenya are often hired on short-term contracts with little job security, and companies frequently terminate projects before the workers are paid.
• 😀 The companies outsourcing these AI tasks, such as Sama, Meta, and OpenAI, pay low wages and do not always provide adequate protections or benefits for workers.
• 😀 Local governments, like Kenya's, face a dilemma of promoting tech investment while failing to protect workers from exploitation due to outdated labor laws.
• 😀 The workers in these jobs feel trapped, as they are desperate for employment, and any pushback could lead companies to move to other regions with cheaper labor.

Q & A

What is the main idea of the script regarding artificial intelligence and human labor?

The script challenges the common narrative that AI will replace human workers. Instead, it highlights the growing need for humans in the loop, particularly in countries like Kenya, to perform labor-intensive tasks that help train AI systems. These jobs, while crucial for AI development, are often low-paying and exploitative.

What types of tasks are workers in Kenya performing for AI companies?

Workers in Kenya are tasked with labeling images and videos, drawing boxes around objects, and categorizing them for AI systems. This includes identifying items in photos, tagging faces, and labeling vehicles and pedestrians to help train autonomous vehicles.

How are these AI-related jobs affecting the mental health of workers?

The workers are exposed to graphic, disturbing content, such as violence, pornography, and abuse, as part of their AI training tasks. This has led to significant psychological trauma, including flashbacks, anxiety, and social isolation. Despite claims of providing mental health support, workers report that the counseling is insufficient.

Why are the workers in Kenya paid so little for their work?

The workers are paid poorly because these jobs are outsourced to regions with large, underemployed populations and weak labor protections. The outsourcing companies, such as Sama, pay local workers a fraction of what they charge the tech giants, like Meta and OpenAI, who pay higher rates for the tasks.

What role does outsourcing play in the exploitation of AI workers?

Outsourcing allows big tech companies to avoid direct responsibility for the poor working conditions and low wages. By engaging outsourcing firms, they can disassociate from the exploitation of workers, while still benefiting from the labor provided by these low-wage countries.

How do tech companies justify the low pay for workers in countries like Kenya?

Tech companies, through outsourcing firms, argue that the pay is reasonable for the region. Sama, for instance, defends the low wages by claiming they are fair for the local economy. However, workers and local activists disagree, calling the compensation exploitative.

What kind of work environment do workers face in these AI training jobs?

Workers often face unrealistic deadlines and high-pressure environments. They are expected to complete complex labeling tasks in very short time frames, sometimes with just seconds to finish each task. The workspaces are described as AI 'sweatshops,' where workers are treated poorly and can be fired for even minor complaints.

What is the significance of Kenya's position in the global tech landscape?

Kenya has become a significant hub for digital labor, often referred to as the 'Silicon Savannah' due to its tech-savvy population and growing digital infrastructure. The government has actively courted major tech companies to set up operations in the country, offering financial incentives and promoting the country as a leading tech destination in Africa.

What are the challenges faced by workers who want to complain about their working conditions?

Workers face the challenge of being vulnerable in an economy where jobs are scarce, and companies can easily shut down operations or relocate to neighboring countries. This leaves workers afraid to speak out, as they fear losing their only source of income or seeing the companies they work for leave the region.

What legal protections exist for AI workers in countries like Kenya?

Kenya's labor laws are outdated and do not adequately cover digital or AI-related work. There are no specific protections for workers involved in AI training, and many workers have no recourse when they are mistreated, underpaid, or exploited by outsourcing firms.