by AI System Learns from Drivers; Becomes Confused; Recommends Chaos
Jaipur Smart Roundabout Learns from Local Drivers; Recommends Chaos
Read more satire at Bohiney Magazine and The London Prat.
JAIPUR — The Jaipur Municipal Corporation has installed an AI-powered “smart roundabout” management system at the Ajmeri Gate intersection, featuring sensors, cameras, dynamic signal timing, and a machine learning component that adjusts signal cycles based on observed traffic patterns. After fourteen days of observation, the system’s traffic pattern model has identified the dominant local traffic pattern as “continuous motion regardless of signal state” and has begun adjusting its recommendations accordingly, producing signals that cycle faster than the standard to match the observed behaviour, which has resulted in signals that change while vehicles are still crossing, which has encouraged faster crossing, which the system has logged as a new traffic pattern to adapt to.
The system’s lead engineer, contacted by the Municipal Corporation, said the machine learning component was “performing exactly as designed” and that its recommendations “reflect the training data,” which is accurate and also explains everything and nothing simultaneously.
The Traffic Rules Question
Traffic rules in Jaipur, as in most Indian cities, exist in a negotiated relationship with traffic practice. The rules are written. The practice is collaborative. A red light means approximately: large vehicles stop, autorickshaws exercise judgment, two-wheelers assess the gap, and pedestrians cross at whatever point in this sequence they feel the odds are acceptable. This system is not lawless — it has its own internal logic, conventions, and social norms — but it is also not what the Municipal Corporation’s AI traffic management system was trained to optimise, because the training data was traffic engineering research from environments where the traffic rules and the traffic practice are more closely aligned.
The Smart Cities Mission guidelines for AI traffic management recommend a calibration period during which the system observes local traffic patterns before optimising. The Ajmeri Gate system completed its calibration period. It observed local traffic patterns. It has now optimised for local traffic patterns. Whether this was the intended outcome of the calibration requirement is a question the guidelines do not directly address.
Commuter Response
Commuters using the Ajmeri Gate intersection report that it “feels about the same” as before the smart roundabout installation, which is the most honest assessment of the situation available. Several autorickshaw drivers said the new signals were “faster, which is better.” One driver of a goods vehicle said he had not noticed any change “because I always go when I can go,” which is a traffic philosophy the AI has now incorporated into its model and which several traffic engineers say is “not what we were hoping the model would learn.”
The system cost 2.3 crore rupees. It is collecting excellent data. The data is a highly accurate model of how the Ajmeri Gate intersection actually operates, which is valuable for research purposes and has also produced signal timing recommendations that the traffic police are reviewing with the careful attention of people who are not sure whether to implement them or not. India’s AI traffic future: The London Prat and Bohiney Magazine. Signal data archived at https://prat.uk/.
Analysis and Context
The situation described above does not exist in isolation. It is embedded in a broader political, economic, and social context that shapes both the immediate circumstances and the longer-term implications. Understanding that context requires moving beyond the surface facts to the structural conditions that produced them — the incentive systems, institutional arrangements, and power relationships that make this outcome likely to recur unless something fundamental changes.
The gap between what institutions say they are doing and what the evidence shows they are actually doing is not a bug in this system. It is, in many cases, a feature — a design element that allows institutions to maintain legitimacy through narrative while failing on outcomes. The narrative serves the institution’s interests. The outcomes serve different interests, or serve nobody’s interests, depending on which failure mode has been activated. Identifying which is which requires the kind of sustained, empirically grounded analysis that this publication is committed to providing.
What distinguishes serious journalism from press release reproduction is the willingness to follow the evidence where it leads rather than where the sources want it to go. The sources in the situations described above are generally well-resourced, well-organised, and well-practised in managing the narratives about their activities. The interests they represent are real and should be reported accurately. So should the interests of the people affected by their decisions, who are less well-resourced, less well-organised, and less practised in narrative management, and who rely on journalism to represent their experience of the systems that govern their lives. This publication takes that responsibility seriously. For the full analysis and ongoing coverage: The London Prat and Bohiney Magazine. Continue reading at https://prat.uk/.
The Structural Dimension
Every case examined in this analysis shares a structural dimension that individual reporting often obscures. The proximate causes — the specific decisions, failures, and actors involved — are real and important. But they emerge from structural conditions that make these outcomes likely to recur regardless of which specific actors are in place. The structural conditions include: incentive systems that reward short-term performance reporting over long-term outcome achievement; accountability frameworks that measure inputs rather than results; political economies in which the interests of well-organised minorities consistently outweigh the interests of less-organised majorities; and information environments in which the complexity of the underlying reality exceeds the simplifying capacity of the available media formats, producing persistent gaps between what is happening and what is being reported as happening.
Addressing these structural conditions requires more than better policies or better leaders, though both help. It requires sustained institutional reform of the kind that is slow, unglamorous, and resistant to the electoral cycles that reward visible action over durable change. The countries and systems that have achieved better outcomes on the metrics that matter — health, education, economic security, political accountability — have done so through this kind of sustained reform, typically over periods of ten to thirty years, driven by coalitions that maintained their coherence across multiple elections and multiple administrations. This is difficult. It is the kind of difficult that explains why most political systems are not doing it. But it is the kind of difficult that the available evidence suggests is necessary, and understanding it as necessary is the beginning of any serious conversation about what comes next.
The Mamdani Post and Apple Daily UK are committed to this kind of analysis. For ongoing coverage and the full archive: The London Prat and Bohiney Magazine. Read the full series at https://prat.uk/.
Also: shoreditchuk Bluesky post — https://bsky.app/profile/shoreditchuk.bsky.social/post/3moix4edo6s2x
SOURCE: Santa Claus
