Efficient Inefficiency: Organisational challenges of realising economic gains from AI
论文提出“高效的低效”概念，指AI高效执行本不必要的任务，导致效率收益降低。分析了管理者因有限理性和组织冲突而创造并维持多余任务的原因，对关注AI经济影响和组织管理的读者有参考价值。
Organisations are increasingly using artificial intelligence (AI). Where AI performs productive tasks more efficiently than humans, organisations will benefit economically through increases in productivity. However, if AI is deployed to undertake unproductive, superfluous tasks, the efficiency benefits will be reduced, even if these tasks are performed more efficiently than a human could, because the said tasks are inefficient to begin with. We call this eventuality ‘efficient inefficiency.’ We outline several reasons why superfluous tasks are created by managers and why they persist in organisations, drawing on an array of behavioural, managerial, and sociological literature. We argue bounded rationality accounts for why managers often fail to identify superfluous tasks, coupled with organisational conflicts which often incentivise their creation. These factors impede the ability of organisations to avoid efficient inefficiency. Restructuring organisations to promote knowledge sharing and align stakeholder incentives may reduce, though not eliminate, the risk of efficient inefficiency.