Automation and the future of work: tasks, skills and reskilling

Automation and the future of work is less a story about robots replacing people than about tasks moving between people and machines. Most jobs are bundles of tasks, and technology tends to take over some of them while making others more valuable. The workers and employers who do best are those who see which tasks are shifting early and invest in the skills that grow alongside the machines.

Tasks, not jobs

The most useful idea in this debate is that automation acts on tasks, the individual activities that make up a job, rather than on whole occupations. A bank teller’s job includes counting cash, answering questions, spotting fraud and selling products. ATMs automated the first; the others remained and, in many branches, grew in importance.

Economist David Autor made this case in a 2015 article in the Journal of Economic Perspectives titled “Why Are There Still So Many Jobs?” His argument: automation substitutes for some tasks but complements others, raising the value of the problem-solving, judgment and interpersonal work that machines handle poorly. Those complementary effects are easy to overlook in headlines about job losses.

This framing also explains why estimates of automation risk vary so widely. Early studies that judged whole occupations produced alarming numbers; a widely cited 2013 study by Carl Benedikt Frey and Michael Osborne put about 47% of US employment at high risk. When the OECD looked instead at the tasks people actually perform, it estimated that about 14% of jobs across its member countries are at high risk of automation, with a further 32% likely to change significantly. That is still a large share of the workforce facing disruption, but the main story is change rather than disappearance. The OECD also stresses that a job being technically automatable does not mean it will be automated: cost, regulation and customer preferences all matter.

Three kinds of automation

“Automation” covers very different technologies, each suited to different work.

Technology	What it automates	Typical examples	Main limits
Industrial and service robotics	Physical, repetitive tasks in controlled settings	Welding, palletizing, warehouse picking, delivery and greeter robots	High upfront cost; struggles with messy, varied environments
Robotic process automation (RPA)	Rule-based digital tasks	Copying data between systems, reconciling invoices, generating standard reports	Brittle when screens, formats or rules change
Artificial intelligence	Pattern recognition and judgment-heavy cognitive tasks	Document review, forecasting, customer support drafts, code suggestions	Errors and “hallucinations”; needs oversight and good data

Robotics

Industrial robots have been common in car and electronics plants for decades, and the International Federation of Robotics tracks their spread in its annual World Robotics reports. Service robots, like the lobby robot pictured above, are newer and narrower in what they can do. Physical automation tends to advance slowly because hardware is expensive and real-world settings are unpredictable.

RPA

Robotic process automation has nothing to do with physical robots. It is software that mimics the clicks and keystrokes a person makes in business applications. RPA is cheap to deploy on stable back-office processes, which is why it spread quickly through finance, insurance and shared-service centers.

AI and generative AI

AI extends automation into tasks that used to require human judgment, such as reading documents, classifying requests or drafting text. Generative AI pushes this into writing, coding and analysis. Early evidence suggests it often augments workers rather than replacing them: a National Bureau of Economic Research study of 5,179 customer support agents found an AI assistant raised productivity by 14% on average and by 34% for novice and low-skilled workers. We explore the business side in our analysis of generative AI for business.

What the World Economic Forum expects

The World Economic Forum’s Future of Jobs Report 2025 is based on a survey of more than 1,000 employers representing over 14 million workers across 55 economies. It is a survey of employer expectations, not a forecast model, so its numbers are best read as direction rather than destiny. Its headline findings for 2025 to 2030:

• Churn is large. Employers expect new jobs equivalent to 14% of today’s employment, about 170 million, to be created, and the equivalent of 8%, about 92 million, to be displaced, for net growth of about 78 million jobs.
• Technology is a top driver. Among technology trends, 86% of employers expect advances in AI and information processing to transform their business, and 58% say the same of robotics and automation.
• Skills keep shifting. On average, employers expect 39% of workers’ existing skill sets to be transformed or become outdated over the period, down from 44% in the 2023 edition.
• Training needs are big. If the global workforce were 100 people, 59 would need training by 2030, and employers expect 11 of them to be unlikely to receive it.

Reading the forecasts carefully

Projections like these are useful but easy to misread. The WEF figures are global totals that hide very different stories by country, industry and occupation: a net gain worldwide can coexist with steep losses in particular roles or regions. The OECD estimates describe technical potential, not a timetable. And history suggests that new technologies also create jobs that are hard to imagine in advance, much as the web created roles in digital marketing and app development that did not exist a generation earlier. A follow-up OECD paper examined what actually happened to occupations previously classed as high risk, a reminder that forecasts should be checked against outcomes. The sensible takeaway is not a precise number but a direction: substantial churn, with the pain concentrated where workers lack routes into growing roles.

Which skills gain value

The same WEF report ranks analytical thinking as the core skill employers most want, followed by resilience, flexibility and agility, and leadership and social influence. The fastest-growing skills are technological: AI and big data, networks and cybersecurity, and technology literacy. But creative thinking, curiosity and lifelong learning also rise in importance. Manual dexterity, endurance and precision show the clearest net decline.

In plain terms, skills gain value when they do one of three things:

1. Direct the machines: knowing how to use, configure, check and question automated tools.
2. Do what machines do poorly: complex judgment, persuasion, care, negotiation and handling the unexpected.
3. Connect the two: combining deep knowledge of a domain, such as nursing, logistics or accounting, with fluency in the tools that are reshaping it.

Reskilling that works

Reskilling means training workers for a different role; upskilling means deepening skills for their current one. Both are easier to announce than to deliver. The programs that work tend to share several features:

1. Start from tasks. Map which tasks in each role are likely to be automated and which new ones will appear.
2. Tie training to real jobs. People engage when a course leads to a defined role, pay band or project.
3. Make time for it. Training squeezed into evenings has low completion rates; protected work time signals that it matters.
4. Learn by doing. Short courses paired with on-the-job practice beat long classroom programs.
5. Measure outcomes. Track redeployment, retention and performance, not just course completions.

Say a company plans to automate 40% of the tasks in its 200-person accounts payable team over three years. Rather than cutting jobs at the end, it might identify roles in supplier analytics, exception handling and process improvement, then train volunteers for those roles as automation rolls out. The cost of training is often lower than the cost of severance plus hiring new people, and institutional knowledge stays in the building.

For individuals, the practical move is to audit your own job: list your weekly tasks, mark which are routine and rule-based, and invest in the ones that require judgment, relationships or technical fluency. If your employer is in the middle of the broader changes described in our guide to digital transformation, ask where the new roles are likely to appear and volunteer early.

The bottom line

Automation reshapes work task by task. Robotics, RPA and AI each take over different kinds of activity, and credible estimates from the OECD and the World Economic Forum point to substantial disruption alongside net job creation, not a jobless future. The winners will be organizations that redesign roles and reskill people deliberately, and workers who build analytical, technological and human skills that complement machines. For more on technology and the workplace, explore our Tech & Innovation hub.