On November 11, Oxford University Professor Mark Graham delivered a lecture titled The Human Labor Behind Artificial Intelligence. He shared insights from his Fairwork research project, which aims to improve working conditions in platform and gig economies through research, advocacy, and the development of standards.
Professor Graham began by explaining the research methodology. He noted that digital platforms have created a globalized labor market, intensifying competition among workers. On these platforms, an oversupply of labor drives down wages and worsens working conditions. Two main factors contribute to this phenomenon: increasing internet access in impoverished areas, leading more people to seek work online, and the invisibility of platforms, which makes them difficult to hold accountable or regulate. The research team interviewed thousands of workers across continents, uncovering numerous risks and harms faced by most of them.
These platforms often do not sign formal employment contracts with workers, meaning they are not legally obligated to pay minimum wages. Workers bear additional costs related to their jobs, resulting in low pay, high expenses, heavy economic burdens, and long working hours. Moreover, workers are exposed to various risks. For example, food delivery platforms encourage risky driving, and in some countries, workers face dangers like kidnapping and assault.
Additionally, platforms’ rating systems are influenced by social biases and discrimination, putting workers in structurally disadvantaged positions where they must comply with platforms’ decisions or risk being eliminated. A deeper issue lies in how workers are treated as machines rather than individuals with unique skills, making them entirely replaceable and hindering their ability to build collective power. Even when workers organize to resist, platforms often evade responsibility.
Fortunately, governments worldwide are increasingly strengthening regulations on platforms, but the reality of workers facing risks and harms persists globally.
The Fairwork project was established against this backdrop. It focuses on four key stakeholders—companies, policymakers, consumers, and workers/unions—with the goal of creating a rating system to evaluate companies and platforms. Through extensive discussions and research, the project developed five core principles: fair pay, fair conditions, fair contracts, fair management, and fair representation. Each principle includes two specific criteria, and platforms must meet these standards to achieve a corresponding rating.
The rating system is based on a rigorous evaluation process to ensure fairness and consistency. These ratings serve as positive examples for responsible corporate management while warning companies with low scores. By engaging with low-scoring companies, the Fairwork team raises awareness of these issues, reducing risks for workers and improving their conditions.
The Fairwork team also collaborates with governments and local teams in various countries, delivering public lectures and disseminating their findings through podcasts, comic books, videos, press releases, and billboards. In addition, they work with local labor lawyers, scholars, and unions to raise workers' awareness of fairness. Their efforts have led to real changes in 304 companies, impacting millions of workers' job conditions. The team continues to monitor these companies’ progress to ensure sustained improvement.
In the concluding section, Professor Mark Graham discussed what ethical and fair artificial intelligence entails. He outlined future plans for the project and introduced his forthcoming book, Feeding the Machine: The Human Labor Behind Artificial Intelligence, soon to be available in Chinese. Professor Graham emphasized that the project aims to make people aware that real human labor underpins algorithms and that this labor should be conducted fairly and ethically.
The lecture concluded with a lively discussion between faculty, students, and Professor Graham about the practical implementation of the rating mechanism. He described it as a narrative system that, through continuous dialogue with companies and workers, helps assess risks, provide oversight, and ensure companies progress in the right direction.