Artificial intelligence is becoming remarkably human in its manner. It converses, remembers, explains and occasionally apologises. Yet as companies make their machines more human-like, some are preparing to treat their people more mechanically, as costs to be removed when a process becomes more efficient. The danger is humanising the technology while dehumanising the organisation.

Alex Garland’s Ex Machina shows why this inversion is so easy. Its machine displays intelligence, vulnerability and charm. A person responds as though convincing behaviour proves an inner life like his own. Fluent AI encourages the same instinct because it sounds patient, remembers preferences and appears to understand. But a human-style interface does not make the system human. AI can perform humanity without being human.

This distinction matters in business. AI can sound considerate, but it does not carry responsibility, lose its livelihood or experience displacement. The immediate question is not whether a machine should be treated like a person. It is whether organisations will continue treating their people like people.

That does not mean resisting efficiency. Capital is finite, productivity matters and some tasks and roles will disappear. Retraining will not solve every case, and no company can preserve every job regardless of economic reality. But job losses should not become the automatic proof that an AI programme is working. Efficiency and humanity are not opposing objectives.

The purpose of productivity is to release resources for more valuable work. When automation saves time, management can remove that time from the cost base or redirect it towards customers, better decisions, new services and neglected work. Cost reduction may be necessary. It is simply not the only possible return. The greater opportunity is to turn saved effort into new capability.

This is why training should be part of the investment case, not a gesture added after deployment. People need practice in directing systems, checking outputs, protecting confidential information and applying judgement where the machine is weak. Redeployment can also preserve knowledge that does not appear in a process map, such as why a customer makes an unusual request or which exception signals a larger problem. Implementation, coaching and job redesign are one investment.

Research by Daron Acemoglu, David Autor and Simon Johnson distinguishes automation from technologies that make human expertise more valuable. Related MIT Sloan research argues that AI often changes tasks rather than removing whole occupations. Research from Peking University’s National School of Development adds a useful perspective: the labour-market challenge is the gap between the speed of technological change and the speed at which human capital adjusts. This makes education and vocational training part of the productivity strategy, not simply a response to displacement. Earlier technological revolutions created lasting value for the same reason. Machines improved, but people also learned to build, operate and create around them. Stronger technology creates more value when people can develop with it.

There are limits. Some roles will shrink, and some transitions will not be possible. Responsible leadership cannot promise that every job will remain. It can examine redeployment, education and job redesign before treating dismissal as transformation. Where redundancy is unavoidable, it can act with clarity and dignity. Redundancy may be necessary, but it should not be the first measure of AI success.

Every significant AI programme should therefore face two questions: how much more efficient has the organisation become, and how much more capable have its people become? Efficiency appears in time, cost and throughput. Capability appears in better judgement, broader skills, stronger customer outcomes and work that was previously out of reach. If only the first improves, the company may be cutting cost without transforming itself. A smaller cost base is not the same as a greater capability.

AI will make many organisations more efficient. That is its promise, but it should not define the limit of management’s ambition. Leaders can use the gain to give people better tools, broader skills and more valuable work. As machines become more human-like, leadership must ensure that organisations become more human, not less.

References

• MIT Sloan, “How artificial intelligence impacts the US labor market” (2025).
• Acemoglu, Autor and Johnson, “Building Pro-Worker Artificial Intelligence” (2026).
• Peking University National School of Development, report on AI, human capital and labour-market adjustment (2026).