Honeywell and Ecolab Executives Warn Against Full AI Autonomy
Honeywell and Ecolab executives argue that high costs and reliability gaps make semi-autonomous AI models more practical than full autonomy for industrial environments.
Executives from Honeywell International, Inc. and Ecolab detailed the practical limitations of deploying artificial intelligence in industrial settings during Fortune’s AIQ Summit at the New York Stock Exchange.
Honeywell Chief Technology Officer Suresh Venkatarayalu stated that current frontier models cannot meet the near-perfect reliability required by industrial customers. He argued that "autonomy is also not about removing people," advocating for semi-autonomous or human-led models to mitigate safety risks.
Ecolab Chief AI Officer AJ Wijesinghe identified high token costs as a significant barrier, noting that some high-volume AI applications are "more expensive than having humans." Despite these costs, Ecolab is integrating horizontal AI and sensors for predictive maintenance, targeting $325 million in annual run-rate savings by 2027.
To address data sovereignty and accuracy, Honeywell is partnering with Nvidia to fine-tune open-source models. Both leaders concluded that the current state of agent technology is too immature to support a fully autonomous industrial future.
