发表机构
internet Research Lab（互联网研究实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究通过GigSaathi聊天机器人收集印度即时零售配送工人收入数据，揭示不透明算法导致收入波动大、无最低工资保障，且激励作为劳动控制工具，凸显该行业零工工作的不稳定性。
AI 中文摘要
印度的即时零售行业呈爆炸式增长，Blinkit、Zepto、Swiggy Instamart 等平台通过密集的社区暗店网络配送杂货和日常必需品。这一增长使该行业成为印度零工经济的主要催化剂，但其对庞大而灵活的配送劳动力的依赖，引发了对工资、工作条件和社会保障的紧迫担忧。这些平台使用不透明的算法来设定薪酬、分配任务和评估绩效，收入因天气和工人可用性等因素而异。由此产生的信息不对称使工人无法重建其薪酬的确定方式，也无法了解薪酬如何随时间和地点波动。我们提出了 GigSaathi，一项旨在解决这种不对称性的试点干预措施；该聊天机器人使配送工人能够系统地收集收入截图，并将其处理为个人工资的时间序列数据。我们还补充了从新德里 Blinkit 门店收集的激励数据。我们的分析揭示了基本工资、激励、工作时间和行驶距离的巨大变异性，且大部分收入取决于未公开的变量；Blinkit 作为该行业最大的公司，不保证每公里最低基本工资。我们发现收入与每单耗时之间几乎没有相关性，表明工人未因配送时间而获得相应报酬，并表明激励作为劳动控制工具发挥作用——每单收入较高的订单，其激励占比反而较低。激励平均约占每周收入的四分之一，但差异很大。这些发现进一步证明了印度即时零售零工工作的不稳定性，其工资由不透明的算法支配，迫使工人追逐未知的激励，而没有任何更高收入的保证。
英文摘要
India's quick commerce sector has grown explosively, with platforms such as Blinkit, Zepto, Swiggy Instamart, and others delivering groceries and daily essentials through dense networks of neighbourhood dark stores. This growth has made the sector a major catalyst for India's gig economy, but its reliance on a large, flexible delivery workforce raises pressing concerns about wages, working conditions, and social security. These platforms use opaque algorithms to set pay, allocate tasks, and evaluate performance, with earnings varying by factors such as weather and worker availability. The resulting information asymmetry leaves workers unable to reconstruct how their pay is determined or how it fluctuates over time and place. We present GigSaathi, a pilot intervention designed to address this asymmetry; a chatbot that enables delivery workers to systematically collect earnings screenshots, which are processed into time series data on individual wages. We supplement this with incentive data collected from Blinkit stores in New Delhi. Our analysis reveals substantial variability in base pay, incentives, working hours, and distance travelled, with a large share of earnings contingent on undisclosed variables; Blinkit, the sector's largest firm, guarantees no minimum base pay per kilometre. We find little to no correlation between earnings and time spent per order, indicating that workers are not compensated proportionately for delivery time, and show that incentives function as a tool of labour control orders with higher per order earnings carry a lower incentive share. Incentives average roughly a quarter of weekly earnings but vary widely. These findings add to mounting evidence that quick commerce gig work in India is precarious, with wages governed by opaque algorithms that push workers into chasing unknown incentives without any guarantee of higher pay.
