AI is an irresistible idea – like flying cars nobody knows how to land or manage (www.theregister.com)

A columnist argues AI is less a "bicycle for the mind" and more a "flying car": an irresistible, headline-grabbing idea that vendors are handing to untrained users who don’t know how to land or manage it. Like early GUIs, conversational interfaces feel intuitive but conceal essential prior knowledge and metaphors; giving people "weapons‑grade" models without guidance overestimates user skill and understates the work needed to turn prototypes into useful systems. The piece warns that casual interactions won't replace deliberate design, and that naïve deployments often crash — socially, legally and operationally. For the AI/ML community this is a call to focus on implementation realities: workflows, data quality, outcome definitions, integration, MLOps, monitoring and human-in-the-loop checks. The column cites the high failure rate of proofs of concept and uses an air‑traffic-control analogy to flag scalability and governance risks. Technical implications include investing in production engineering, observability, verification, user training and vendor support rather than treating models as plug‑and‑play productivity magic — otherwise organizations will keep burning time and trust on spectacular failures.