Making Sense of Machines: Ritualized Care, Agency Attribution, and Failure Interpretation in Human-Automation Encounters in India

<p>Machines in India are cared for, spoken to, honored, and blamed, well beyond ordinary use. New vehicles are blessed before their first drive, workshop tools are garlanded and rested during Vishwakarma Puja, and small tilak marks or deity images sit on machine casings without a second thought. Human-computer and human-robot interaction research has rarely asked what these everyday practices do to the way people read a machine that hesitates, fails, or behaves unexpectedly. This paper reports a formative, mixed-method study of how adults in India make sense of ambiguous automated behavior, and whether caring for and personifying machines shapes how they read that behavior. A survey of 156 adults gives a baseline for how common these practices are, how people explain failure, how much agency they grant a machine, and what they want from a system when it fails. Machine-care practices turned out to be almost universal: 96% of respondents reported at least one. In a scenario where a vehicle fails before an important journey and then starts again on its own, 62% chose a relational or meaningful explanation alongside the technical option they were also given, and 60% expected a machine they had looked after to perform well in return. The interviews complicate that last figure: most of what looks like relational trust there is closer to an ordinary belief that upkeep improves performance. The orientation toward the recovering vehicle did not carry over evenly to newer technology, either. About a third of respondents said they trusted AI more than mechanical machines and expected it to “know better,” a view far more common among younger respondents. I read the survey alongside twenty-one semi-structured interviews, conducted in Hindi and English with photographs and short scenarios and analyzed thematically. Across both urban and rural lives, people mostly treated machine care as ordinary maintenance, not as a belief that machines are conscious. That relational orientation showed up again and again, but it also ran into real limits: in platform work, in closed or inaccessible design, and among people who reject it on rational or minority-faith grounds. There is no single Indian view of machines here. Region, generation, religion, and class produce real differences, and I treat those differences as central to the findings, not as noise to explain away. The paper offers three things: a documented instrument set for studying culturally situated automation sensemaking, along with an honest account of where its measures blur together and how they could be sharpened; a preliminary vocabulary connecting ritualized care, agency attribution, and failure interpretation; and an early design agenda for settings where machines carry meaning beyond what they do.</p>