TECH & AI
challenge when developing highly accurate machine learning models”. Gaps and inconsistencies in this history make it harder to build models that engineers can trust.
Anindya says he prefers to treat this as a path rather than a wall. ABB works alongside customers to validate, contextualise and enrich their data before any model development begins, turning historic weaknesses into a stronger foundation for future work.
His advice to manufacturers tempted to wait for cleaner data before adopting
AI is blunt: do not.“ Waiting for perfect data before embracing AI risks turning data quality into an excuse for delaying technologies that can deliver meaningful operational benefits today,” he explains.
Racing against retirement A refinery running a large fleet of ageing critical assets, among them compressors, pumps, furnaces and heat exchangers, faced two pressures: reduce the risk of downtime while capturing decades of operational knowledge before experienced engineers retired.
76 August 2026