Ghost Work in Amazon’s AI-Run Grocery Stores
By Nandini Shrotriya
American tech company, Amazon, along with many other multinational companies, heavily relies on the use of Artificial Intelligence to increase efficiency, reduce costs, and achieve a hyper-personalized user experience. Amazon Go, one of Amazon’s business ventures, piloted the “just walk out” technology. Instead of having cashiers or self-checkouts at grocery stores, sensor-based technologies were implemented. In 2023, there were 43 operating stores, not only in the United States but in England as well. The innovation was met with positive feedback until it was revealed that Amazon had over 1,000 remote human workers in India performing ghost work as they manually reviewed the transactions. This false perception of a fully automated AI system speaks to how human labour remains the crux of obscure human labour and corporate responsibility to maintain the perception of automation.
Ghost work is the hidden work performed by humans in supporting AI systems to make them appear more efficient. The erasure of human labour leads to their devaluation, making people be seen as replaceable. In turn, this form of precarious job has them underpaid and hired as contractors without the usual benefits seen in non-precarious work, as ghost work is classified as temporary labour. Additionally, framing human labour as AI allows companies to avoid scrutiny over wages and labour practices, as well as responsibility for working conditions.
The choice of having Indian workers is not incidental. It reflects a form of digital colonialism where labour from the Global South is stolen for profits while credit remains in the Global North. The exploitation of workers in India, along with Amazon’s well-documented history of unsafe and exploitative labour practices, reflects how it is not unreasonable to infer that the conditions faced by the workers or valuation of their work in India for Amazon Go were rendered invisible to maintain the corporate strategy of maximizing profit by minimizing labor visibility and responsibility.
This case highlights broader concerns about how AI can be used to deflect accountability for labor practices and can reinforce global inequalities.
