Recent reports question the reliance of Amazon’s ‘Just Walk Out’ technology on human labour, despite initial promises of AI automation. Discrepancies in the system prompt human intervention, raising concerns about transparency and labour dynamics in tech industry.
Amazon’s “Just Walk Out” technology, which debuted in January 2018 at the first Amazon Go store in Seattle, was heralded as a significant advancement in AI-powered shopping. This system allows customers to pick items and leave the store, with the items automatically charged to their accounts.
Recent reports have emerged suggesting that the technology relied more on human labor than initially disclosed. Sources indicated that a team of approximately 1,000 workers in India verified transactions via camera feeds. Amazon countered these claims, describing them as “misleading and inaccurate,” and clarified that human reviewers were used primarily to train the system for greater accuracy.
Despite the technology’s promise, discrepancies often required human review, with reports indicating that in 2022, there were 700 reviews for every 1,000 transactions. This review process entailed checking camera footage, often causing delays in receipt issuance.
Amid these revelations, Amazon has announced layoffs among U.S. staff involved in installing and supporting “Just Walk Out” systems. The technology will continue to be utilized in smaller third-party and 40 Amazon Fresh grocery stores.
The practice of employing cheaper overseas labor for tasks such as AI training is common in the tech industry. According to experts like Benjamin Shestakofsky from the University of Pennsylvania, this method can obscure the labor force’s role, potentially devaluing their work and reducing labor costs.
