The Human Toll of the AI Boom: Uncovering the Hidden Costs of Artificial Intelligence Development

If you frequently use apps from top-tier technology companies like OpenAI, Amazon, Microsoft, or Google, there's a good chance you've benefited from services facilitated by remote work, commonly referred to as cloudwork. Organizations, both large and small, increasingly depend on outsourced labor through platforms like Scale AI, Freelancer.com, Amazon Mechanical Turk, Fiverr, and Upwork.

Recently, these platforms have become vital for artificial intelligence (AI) companies, enabling them to train their AI systems effectively. OpenAI, for instance, relies on Scale AI and Remotasks for data labeling to enhance applications like ChatGPT and DALL-E. Social media networks also utilize these platforms for content moderation. Beyond tech, universities, businesses, and non-governmental organizations (NGOs) regularly hire for roles such as translation, graphic design, and IT expertise on these platforms.

Cloudwork platforms present a significant income opportunity for an increasing number of individuals. A recent study by University of Oxford scholars Otto Kässi, Vili Lehdonvirta, and Fabian Stephany revealed that over 163 million people have signed up on these sites. Notably, Freelancer.com boasts more than 67 million registered workers—equivalent to the entire population of the United Kingdom. Many of these freelancers reside in low- and middle-income countries, providing their services to clients worldwide.

From the client's perspective, the appeal is undeniable. Online labor platforms offer access to a cost-effective, often skilled workforce available around the clock. The intense competition for jobs on these platforms means that clients wield considerable power in negotiating terms and pay, explaining why AI companies are eager to outsource the extensive manual tasks involved in their application development.

However, the situation for workers can be challenging. Many freelancers spend significant amounts of time searching for and applying for jobs, often without remuneration. Most platforms provide no assurance that the work available will meet minimum wage standards. Additionally, if disputes arise with clients, the lack of a clear appeals process can leave workers vulnerable to unpaid wages.

For the third consecutive year, the Fairwork project at the University of Oxford conducted an evaluation of 15 cloudwork platforms, including Freelancer.com, Amazon Mechanical Turk, Fiverr, and Upwork. These platforms were rated on a scale of 0 to 10 based on five principles of fair work: pay, conditions, contracts, management, and representation. Criteria assessed included timely payments to workers, health and safety protections, and the presence of clear communication channels for resolving worker-related issues. The findings were derived from a survey of 752 workers across 94 countries, supplemented by insights from platform managers.

Shockingly, none of the platforms achieved more than half of the basic standards necessary for decent work. Popular freelance sites such as Fiverr and Freelancer.com scored just 2 and 1 points, respectively. Platforms focusing on small tasks like data labeling or content moderation received particularly low scores: Appen (3), Clickworker (1), Scale/Remotasks (1), Microworkers (0), and Amazon’s Mechanical Turk (0). On average, surveyed workers earned a meager $2.15 USD for their time spent on these platforms.

These results highlight the significant shortcomings of many cloudwork platforms regarding workers' rights and protections. Due to the international nature of the industry, these platforms often sidestep national labor laws, creating a digital labor landscape reminiscent of the Wild West, where workers face substantial risks. Disturbingly, this model is rapidly proliferating, fueled by the demand from AI companies and the broader trend towards remote work that has emerged post-pandemic.

Notably, some platforms like Terawork and Comeup have pledged to implement improvements, such as establishing a minimum wage floor. However, the overarching challenges in the sector will only be effectively addressed through regulations that ensure these companies adhere to the same labor standards as other industries.

At the international level, the International Labour Organization (ILO) is deliberating the creation of standards aimed at safeguarding workers’ rights within the digital economy. This presents a crucial opportunity to tackle the issues faced by workers on cloudwork platforms. However, effective change necessitates the involvement of national governments around the globe.

Regrettably, so far, regulatory discussions surrounding the platform economy have largely concentrated on sectors like food delivery and ride-hailing. Urgent attention is needed from policymakers and regulators to consider the unique challenges confronting online remote workers and to channel comparable energy into devising solutions for this often-overlooked, yet significant, workforce.