The paradox of reliability
As AI becomes more capable and dependable, people may need to intervene less often. That looks like progress—and often is. Yet fewer interventions can also mean fewer chances to practise the skills required when the technology fails or encounters a genuinely unusual situation.
How capability is lost
Expertise is not maintained solely by watching competent work being done. It depends on judgement, experience, feedback and deliberate practice. If AI takes over the routine and intermediate tasks through which newcomers learn, fewer people may develop the deeper understanding required for supervision and command. Experienced professionals can also become less practised if they rarely exercise their judgement.
A lesson from other fields
Aviation, medicine and other safety-conscious professions have long had to think about automation, supervision, handover and skill maintenance. The details vary, but the wider principle travels: effective automated performance does not by itself guarantee that a human can respond well when automation is no longer enough.
Human command must remain a capability
A nominal human-in-the-loop arrangement is inadequate if the human has neither the competence nor the time to recognise an emerging problem. Command requires understanding the task, appreciating uncertainty, questioning a recommendation and being able to intervene. Those abilities should be maintained deliberately rather than assumed.
Build practice into the system
Organisations can preserve capability by rotating staff through non-AI tasks, running supervised exercises, testing intervention skills, providing explanations of system decisions and retaining meaningful human involvement in selected cases. Training can include failures and unusual situations rather than only normal operations. Regular checks should focus on what staff can actually do, not merely whether they have completed a course.
The succession question
Who will become tomorrow’s expert if today’s apprenticeship tasks disappear? That question applies to junior professionals, technicians, managers and future commanders. The AI strategy and the human capability strategy must be developed together; otherwise the organisation may become increasingly dependent on a tool it is decreasingly able to challenge.
More articles and information about the books: ai-humancapability.com