Examine how emerging technologies like AI and IoT-based surveillance can address governance and accountability gaps in critical public infrastructure. Illustrate with the example of Indian Railways' level-crossing safety systems.

Critical public infrastructure usually fails less for want of rules than for want of independent verification — compliance is self-reported and unauditable. AI and IoT-based surveillance can convert such trust-based systems into evidence-based accountability, as Indian Railways' response to level-crossing (LC) failures shows, though technology supplements rather than substitutes structural safety investment.

The accountability gap in critical infrastructure

• At non-interlocked LC gates, the gateman manually closes the gate and confirms closure to the Station Master through a verbal "private number" exchange — a single point of failure with no means of independent verification [1].
• The Cuddalore (Tamil Nadu) accident of July 2025, in which a train struck a school van at a manned but non-interlocked gate killing three schoolchildren, exposed this deficit: closure was reportedly conveyed while the gate stood open [1].
• Accountability was therefore individual and post-facto (suspension, arrest) rather than systemic and preventive.

How AI and IoT close it

Gate sensor/CCTV → AI check on closure time
   → live feed: Station Master + engineering control room
   → automated alert if gate stays open  → intervention before train arrival

• The Railway Board has ordered AI-enabled, solar-powered CCTV at all non-interlocked LC gates, streaming live feeds to the Station Master and engineering control room and auto-alerting when a gate is not closed in time [1]; solar power enables off-grid rural sites.
• This creates real-time detection, a tamper-resistant audit trail and shared situational awareness, replacing self-certification.
• Parallel deployments — Kavach (indigenous Automatic Train Protection), electronic interlocking, and AI-based Distributed Acoustic Sensing for elephant intrusion — show technology-led safety layering; consequential accidents fell from 135 in 2014-15 to 31 in 2024-25 [2].

Limits

• Surveillance detects, it cannot physically stop; hence structural fixes remain primary — as with the elimination of all unmanned LCs on Broad Gauge by January 2019 via ROBs/RUBs and manning [3].
• Risks: alert fatigue, connectivity and maintenance gaps, and use of footage purely to punish frontline staff rather than redesign processes.

Technology thus works best as an accountability enabler embedded in institutional reform — sensors that inform, protocols that act and audits that learn. Extending such verification-by-design across all LC gates, backed by Rashtriya Rail Sanraksha Kosh funding and staff training, would align railway governance with the citizen's right to safe mobility.