Enhanced Automation

In short: Machines have always taken over tasks, but today they can also do mental work: reading, writing, coding, diagnosing. The IMF estimates that about 40% of jobs worldwide, and 60% in rich countries, are exposed to AI. Office and service work is affected first; physical work follows once robots are ready, possibly within five to ten years for a given task. New jobs appear, but with a delay and with different skills. Automation also makes things cheaper, so less income may be needed, though housing and health care resist that. How work ends up shared will depend on three forces: concentration in the market, people choosing to work less, and policy stepping in.

The Question

Automation is not new: humans have built tools to extend their abilities for millennia. What is new is the speed, the range and the intelligence of today’s machines. This article asks: what will automation do to work, and how can we adapt?

What Gets Automated

Automation has usually replaced tasks that are routine, predictable and well defined first, whether physical (assembly lines) or mental (data entry, standard analysis). Harder to automate are:

• real creativity and solving new problems;
• work that needs a lot of emotional intelligence, such as care, counselling or complex negotiation;
• judgment in unclear, high-stakes situations;
• physical work in messy, unpredictable places.

That line is moving fast. Knowledge work built on pulling information together, such as consulting, finance or diagnostics, is now exposed earlier than the old pattern suggests, because AI is becoming better at parts of it, not just cheaper or faster.

The Impact on Work

• Exposure: the IMF estimates that about 40% of jobs worldwide, and about 60% in advanced economies, are exposed to AI. Exposure does not mean disappearance: in about half of those jobs, AI is more likely to help workers than replace them.
• Order of change: office and service roles are usually affected first, which can mean less demand for some roles and pressure on wages. Physical work follows more slowly, limited by how fast robots spread. But once robotics is ready for a given task, jobs in transport, manufacturing or skilled trades could shift within roughly five to ten years, much less time to adjust than past changes allowed.
• New jobs: historically, automation has created about as many jobs as it removed, but with a delay, and the new jobs need different skills. The World Economic Forum expects about 170 million jobs to be created and 92 million displaced worldwide between 2025 and 2030. The cost of the gap tends to fall on workers with the fewest alternatives.
• Changed jobs: even jobs that stay will change. Machines take the routine; people focus on judgment, relationships, creativity and oversight.

Falling Costs

When machines produce goods and services more cheaply, prices tend to fall: food, manufactured goods, transport, some services. So the income needed for a given standard of living could drop, and less paid work could go further.

How much this helps depends on which costs fall. Housing and health care weigh much more on living costs than electronics, and have so far resisted automation’s price drops. Cheaper goods alone won’t make up for less paid work.

How Work Could End Up Being Shared

Three forces could shape this, and they can run side by side:

• Concentration: left alone, employers tend to prefer fewer workers who are fluent with automation, rather than spreading work across more people. The default is a smaller group working full time, often in oversight roles, while others compete for fewer jobs at lower wages.
• Choosing to work less: if living costs fall enough, some people may choose to work fewer hours because they need less income. Not everyone, but those with some cushion.
• Policy stepping in: if unemployment becomes more than society will accept, history suggests governments react, usually late rather than early: shorter working weeks, more redistribution, or rewards for sharing jobs.

A plausible future mixes all three. The balance will depend on how fast the change comes, how evenly prices fall, and how quickly institutions respond.

How to Adapt

As an individual:

• build skills that are deeply human: communication, empathy, creativity, complex reasoning;
• work with automation, using AI tools as a multiplier rather than fighting them;
• keep learning: specific skills go out of date faster, so learning how to learn matters most;
• build financial resilience, such as an emergency fund, for the period between old and new work.

As an organisation:

• invest in retraining alongside automation;
• redesign jobs around what people do well;
• be open with staff about automation plans;
• consider the wider effects of large-scale job loss.

As a society:

• education focused on adaptability, not fixed content;
• a safety net strong enough for people in transition;
• tax systems that may need to change if more value goes to machines and less comes from taxing work.

If these aren’t enough, the response may combine redistribution, such as a Universal Allowance, with deliberate choices about which roles stay human, not because machines can’t do them, but because the human part is what gives them value.

Examples

• Bank tellers and ATMs: when cash machines spread in the United States, the number of bank tellers did not fall for decades. Each branch needed fewer tellers, but banks opened more branches, and tellers moved towards advice and sales.
• Freelance writing after ChatGPT: a study of an online freelance platform found that after ChatGPT’s release in late 2022, freelancers in writing-related jobs got fewer jobs and earned less, a fast and direct effect.

The Human Side

Losing work, or fearing it, hits more than income. It touches identity, routine and the feeling of being useful. Change is easier to face with some financial cushion, people around you, and a sense that you can still learn and choose. See Mastering finance basics and Managing fear in the Personal Guide.

Open Questions

• Who should benefit from the gains of automation: owners, workers, or everyone?
• Should working time be shortened as productivity rises, as it was in the twentieth century?
• Which jobs should society deliberately keep human?

Where Next

• Previous: Population Change and Ageing
• Next: Impact of Artificial Intelligence
• Related: Life After Automation, Universal Allowance

Sources

• Mauro Cazzaniga and others, Gen-AI: Artificial Intelligence and the Future of Work, IMF Staff Discussion Note (2024)
• World Economic Forum, The Future of Jobs Report 2025
• OECD, Employment Outlook 2019: The Future of Work
• Daron Acemoglu and Pascual Restrepo, “Robots and Jobs: Evidence from US Labor Markets”, Journal of Political Economy (2020)
• James Bessen, Learning by Doing (2015), on ATMs and bank tellers
• Xiang Hui, Oren Reshef and Luofeng Zhou, “The Short-Term Effects of Generative Artificial Intelligence on Employment: Evidence from an Online Labor Market”, Organization Science (2024)
• David Autor, “Why Are There Still So Many Jobs? The History and Future of Workplace Automation”, Journal of Economic Perspectives (2015)