AI Academy · Book
Executives & Directors · Module 01 · Chapter 011

AI and the Future of Work

"How many jobs will AI replace?" is the question everyone asks and the one leaders can do least with. The early evidence shows AI changing tasks, hiring and skills long before it removes occupations. The useful unit is the task, and the leader who owns the work owns its redesign.

≈ 14 min read

After this chapter you can

  • Treat the task, not the job title, as the unit of workforce change, and sort a role's tasks into three states.
  • Explain why the choice between automating and augmenting a task moves jobs, starting with entry-level hiring.
  • Read the early labor-market evidence and the main jobs forecast without overstating either.
  • Recognize that skills shift and that AI literacy is role-specific.
  • Communicate workforce change honestly, including where telling workers is a legal duty.

In August 2026, economists at Stanford’s Digital Economy Lab updated one of the most closely watched studies of AI and employment. Using payroll records for millions of American workers through June 2026, they reported two findings that sit uneasily together. First, there was no sign of widespread job loss across the economy. Second, employment of workers aged 22 to 25 in the occupations most exposed to AI stood about 19 percent below where it would have been had it kept pace with young workers in less exposed occupations. The gap had widened from 15 percent a year earlier, and it came mainly from fewer hires rather than more layoffs1.

Employment of 22- to 25-year-olds in the most AI-exposed US occupations was about 19 percent below trend by June 2026, up from 15 percent a year earlier, with no comparable gap for older workers.19%Employment gap,ages 22-25In the mostAI-exposed occupations15%The same gap ayear earlierJuly 2025 dataNoneComparable gap forolder workersSame occupationsSource: Stanford Digital Economy Lab, ADP payroll data · June 2026
Figure 1.11.1 The first signal is in hiring, not layoffs: fewer young people are being brought into AI-exposed work.

A month later, the Budget Lab at Yale, which tracks the whole US labor market, found no clear evidence of AI-related disruption in data through August 2026. The mix of occupations was changing, but not in ways that lined up with where AI is used2. So the honest answer to “how many jobs has AI replaced?” is: few that anyone can measure. And that answer misses what is actually happening. The change is real, but it is not arriving as a number of jobs. It is arriving as tasks that are done differently, hires that are not made and skills that matter more or less than they did.

The unit of change is the task

A job is not one activity. It is a bundle of tasks held together by a title. A project coordinator schedules deliveries and meetings, writes the daily status report, chases suppliers for missing parts, settles clashes between teams and keeps the sponsor’s trust. A financial analyst gathers data, builds models, writes commentary and argues for a recommendation. AI affects each of those tasks differently, so a question about the job as a whole has no good answer.

Economists have analyzed technology this way for more than twenty years. David Autor, Frank Levy and Richard Murnane showed that computers substituted for routine tasks that follow explicit rules and complemented the non-routine tasks of problem-solving and communication. Jobs did not simply vanish or survive; their task content shifted3. As The Four Industrial Revolutions showed, most workers now have at least some tasks exposed to AI, and exposure is not the same as replacement4.

The practical consequence is a change of question. “How many jobs will AI replace?” leaves leaders arguing with forecasts. “Which tasks in this role will change, and what does the role become?” is a question a leader can work on, one role at a time.

One role, three states

Inside any role, each task sits in one of three states. It can stay human-led: the person does it, perhaps with some help. It can become AI-assisted: AI drafts, searches or checks, and the person stays accountable for the result. Or it can become AI-automated: AI does the task, and the person handles exceptions and watches the quality.

A project coordinator's role holds tasks in three states - team clashes, the sponsor's trust and safety calls stay human-led; chasing parts and drafting notices become AI-assisted; status reports and bookings may become AI-automated.AI-automatedStatus reports, routine bookingsPerson handlesexceptionsAI-assistedChasing parts, draftingschedule-change noticesPerson checks andowns itHuman-ledTeam clashes, the sponsor's trust, safety callsPerson does itAIDOESMORE
Figure 1.11.2 One role, three states. Illustrative tasks for a project coordinator.

In this illustration, which is not drawn from a study, settling a clash between two teams, holding the sponsor’s trust and making a safety call stay human-led. Chasing suppliers for missing parts and drafting schedule-change notices become AI-assisted. The daily status report and routine bookings may be largely automated. The title is unchanged; the week is not. These three states are the task-level version of the three levels of change in AI Is Changing Everything.

Automate or augment: the choice that moves jobs

Which state a task moves into is not fixed by the technology. It is partly a choice, and it matters for jobs. The Stanford team found that the employment declines for young workers were concentrated in occupations where AI is mostly used to substitute for human tasks. Where AI is mostly used to complement workers, employment was flat or rising1.

Where AI substitutes for tasks, entry-level hiring falls and new work emerges more slowly; where it complements workers, employment holds or rises and new work appears.Where AI substitutesFewer entry-level hiresSlower creation of new workJuniors lose practice tasksWhere AI complementsEmployment flat or risingNew kinds of work appearPeople handle harder cases
Figure 1.11.3 The same technology moves jobs in opposite directions depending on whether it replaces tasks or strengthens the people doing them.

Eight decades of history point the same way. Autor and colleagues traced new job titles in the United States from 1940 to 2018 and found that the majority of today’s employment is in job specialties that did not exist in 1940. New work grew out of innovations that complemented what occupations produced. Innovations that automated tasks slowed it down, and their demand-eroding effect has grown stronger over the last forty years while the boost from augmentation has not5.

The field evidence on AI assistants fits this picture. As AI Is Changing Everything described, an assistant in a customer-support center lifted the least experienced agents the most: augmentation that made juniors better, rather than automation that made them unnecessary6. The lesson for leaders is not to avoid automation. It is to know which one they are choosing, task by task, and what it does to the people who used to learn the job by doing those tasks. Operational and Workforce Risk takes up what happens to expertise when juniors lose that practice.

Roles change shape before occupations disappear

Headlines count occupations. Organizations live in roles, and roles can change shape for years before an occupation shrinks in the statistics. The OECD’s 2023 review of the evidence concluded that, so far, AI was changing jobs and the skills they need more than it was replacing them7.

Headlines count occupations and job titles; below the surface leaders manage where the hours go, who gets hired, what juniors learn, what managers inspect, handoffs and what good work means.WHAT HEADLINES COUNTOccupations · Job titlesWHAT LEADERS MANAGEWhere the hours goWho gets hiredWhat juniors learnWhat managers inspectHandoffsWhat good work means
Figure 1.11.4 Much of the change happens below the line, long before a job title disappears.

Below the line, the change is already visible to anyone who looks. Hours move between tasks. The seniority mix of a team shifts, often through hires that are quietly not made, as the Stanford data suggest. What a junior learns in the first year changes, and so does what a manager inspects. Many leaders will redesign work long before they abolish a job family. That is not a softer version of the story. It is the management work itself, and it is where the outcome for people is decided.

The forecasts: large churn, uncertain balance

Forecasts exist, and leaders will be asked about them. A widely cited one is the World Economic Forum’s Future of Jobs Report 2025, built on a survey of more than 1,000 employers representing over 14 million workers in 55 economies. Extrapolating from their answers, the report expects structural change between 2025 and 2030 equal to 22 percent of today’s jobs: about 170 million jobs created and 92 million displaced, a net gain of 78 million. The largest declines in absolute numbers are expected in clerical roles such as cashiers and administrative assistants, and the fastest-declining roles include postal clerks, bank tellers and data-entry clerks8.

The World Economic Forum expects about 170 million jobs created and 92 million displaced worldwide from 2025 to 2030, a net gain of 78 million.170 MJobs created−92 MJobs displaced78 MNet changeSource: World Economic Forum, Future of Jobs Report 2025 · 2025-2030 forecast
Figure 1.11.5 A net gain hides the real story: 262 million jobs created or displaced, and different people in each column.

Read the number with care. It is a forecast built on employers’ expectations, covers all forces of change, not only AI, and says nothing about whether the people displaced are the people hired. As The Four Industrial Revolutions showed with Engels’ pause, a technology can raise output for decades while the workers who live through the transition gain little9. A positive global balance is no comfort to a team whose tasks are disappearing. The leader’s job is the transition inside the organization, not the global sum.

Skills shift, and literacy depends on the role

The same survey asked about skills. Employers expect about two-fifths of workers’ existing skills, 39 percent, to be transformed or become outdated between 2025 and 2030. AI and big data top the list of fastest-growing skills, followed by networks and cybersecurity and technological literacy. Analytical thinking remains the core skill employers want most, and seven in ten consider it essential.8.

Employers expect 39 percent of workers' skills to change by 2030, with AI and big data the fastest-growing skills.39%Of workers' skillsExpected to change or become outdated by 2030No. 1AI and big dataThe fastest-growing skillsSource: World Economic Forum, Future of Jobs Report 2025 · 2025-2030 forecast
Figure 1.11.6 Skills shift with the tasks, and the fastest growth is in AI and data skills.

Two misreadings are common. The first is that AI literacy means learning to code. It does not. Literacy is role-specific: enough understanding to judge AI output and redesign the work in one’s own role. A finance professional needs to know when a generated variance explanation is plausible and when it is wrong. An operations leader needs to know where automation is safe and where assistance is the ceiling. The second misreading is that training alone solves the problem. Training matters, but people drift back to the old way of working unless role definitions, what managers inspect and career paths change with it. Which people and capabilities an organization needs is a separate question, and it is the subject of the next chapter.

Tell people the truth about tasks

Sooner or later, people ask whether their job is safe. Many are already worried. In a Pew Research Center survey of more than 5,000 US workers, 52 percent said they were worried about how AI will be used in the workplace in the future, and 36 percent said they were hopeful. Almost a third expected AI to mean fewer job opportunities for them; only 6 percent expected more10.

Asked whether a job is safe, promising nothing will change breaks trust later and fear makes people hide experiments; telling the truth about tasks gets people to help redesign the work.Is my job safe?Promise nothing will changeTrust breaks laterUse fear as motivationPeople hide experimentsTell the truth about tasksPeople help redesign the work
Figure 1.11.7 Two easy answers fail. The one that works is specific about tasks and honest about uncertainty.

Two easy answers fail. Promising that AI will eliminate no jobs fails because the promise may not hold, and when it breaks, the leader’s credibility breaks with it. Using “AI will replace you” as motivation fails because fear does not improve work; it makes people hide what they are trying. The answer that works is specific: which tasks are changing, where people stay accountable, how they will be involved in the redesign and what support exists for those whose tasks shrink. Asking a team which part of its week it would gladly give up, and which part it would refuse to give up, is often the best way to start.

Story: the call center that became a design studio

Ingka Group, the largest IKEA retailer, offers a well-documented example of an organization that started with tasks rather than headcount.

Before. Until 2021, Ingka’s customer call centers handled every kind of enquiry, routine and complex alike, by person. That year the company introduced Billie, an AI chatbot named after the Billy bookcase. Over the next two years, by the company’s own figures, Billie resolved about 47 percent of the enquiries it received, 3.2 million interactions, and saved nearly 13 million euros12.

A headcount-first reading of those numbers was easy to make: nearly half the enquiries reaching the chatbot no longer needed a person, so the call centers could shrink. Ingka took a different route. It looked at what people could do that the chatbot could not, and found work it had never offered at scale: remote interior design advice. After. Ingka reskilled 8,500 call-center co-workers in remote interior design, digital sales, building customer relationships and solving complex problems12. In its 2022 financial year the remote channel brought in 1.3 billion euros, 3.3 percent of sales, and the company set a target of 10 percent by 2028. Asked by Reuters whether AI would mean fewer staff, Ingka’s people and culture manager said, “That’s not what we’re seeing right now”13.

Ingka launched the Billie chatbot in 2021, which resolved 47 percent of enquiries while 8,500 co-workers were reskilled; remote design sales reached 1.3 billion euros in FY2022; in March 2026 it announced 800 office cuts, citing complexity rather than AI.2021Billie launchedChatbot takesroutine enquiries2021-2347% resolved8,500 co-workers reskilledFY20221.3 billion eurosRemote design sales, 3.3%of totalMarch 2026800 officecuts announcedCited complexity, not AI
Figure 1.11.8 Ingka moved the routine tasks to AI and moved its people into work it had never offered at scale.

The story has an honest ending, and it matters. In March 2026 Ingka announced about 800 job cuts in its group functions, the office-based part of the business. Its new chief executive said the organization had “grown too complex”; the company did not attribute the cuts to AI14. Redesigning work is not a promise that no job will ever be cut. What it did was give the people whose routine tasks moved to a chatbot somewhere to go, because leaders created new work for them before deciding anything about headcount. The figures are the company’s own, and no independent evaluation has been published, so treat the case as an example of a sequence, not proof of a return.

What this means for leaders

Four lessons follow. First, answer job questions with task analysis. When a board asks how many jobs AI will remove, break the largest roles into tasks and show which are likely to stay human-led, become assisted or become automated. Second, decide automate or augment on purpose. The same exposure can shrink entry-level hiring or make juniors better, and the choice is made task by task, usually without anyone noticing it is a choice. Third, put the redesign with the leader who owns the work. HR is essential for skills, career paths and fair treatment, but it cannot redesign the project schedule, the month-end close or the design consultation without the person who runs that work. Fourth, decide where freed time goes before deciding headcount. Time saved is not value until someone directs it, as Productivity vs Realized Capacity explains.

Check yourself

  1. The clearest early sign of AI in US jobs data is mass layoffs.
  2. The same AI exposure can raise or lower employment depending on how it is used.
  3. Most of today’s US employment is in job specialties that existed in 1940.
  4. A positive global forecast for jobs means few people in your organization will be affected.
  5. AI literacy means everyone learns to code.
  6. Ingka’s redesign guaranteed that no jobs would ever be cut.

Reflection: decompose one role

What comes next

Once tasks change, the work asks for different things from people. That raises a question this chapter has deliberately left open: which people and which capabilities does the redesigned work actually need? The next chapter, The AI Talent Gap, takes it up.

Laws referenced

Not legal advice. Laws change; verify before relying on this, and consult counsel for decisions.

EU AI Act · EU

Regulation (EU) 2024/1689, as amended by Regulation (EU) 2026/1744

Risk-based rules. Prohibited practices include social scoring, untargeted scraping of facial images, and emotion recognition in workplaces and schools (with narrow exceptions). High-risk systems (Annex III: biometrics, safety components of critical infrastructure such as energy, water and traffic, employment and worker management, credit, education, essential services, law enforcement, migration, justice) need risk management, data governance, documentation, logging, human oversight, human oversight that keeps people able to understand the system, notice automation bias (over-reliance on its output), override it or stop it (Art. 14(4)), appropriate accuracy, robustness and cybersecurity (Art. 15), automatic logging of events (Art. 12), a provider quality-management system (Art. 17) and conformity assessment. An Annex III system is not high-risk if it poses no significant risk of harm, for example a narrow procedural or preparatory task that does not replace human assessment; systems that profile people are always high-risk, and a provider relying on this exception must document it and register (Art. 6(3)). Deployers of high-risk AI must use it as instructed, assign competent human oversight, monitor its operation, keep logs for at least six months and report serious incidents (Art. 26); employers must inform workers' representatives (Art. 26(7)). Public bodies, private providers of public services, and deployers of credit-scoring or life and health insurance pricing systems must carry out a fundamental-rights impact assessment before first use (Art. 27). Providers must run post-market monitoring (Art. 72). A deployer that puts its name on a high-risk system, substantially modifies it, or changes its purpose so that it becomes high-risk takes on the provider's obligations (Art. 25(1)). A substantial modification (Art. 3(23)) of a high-risk system needs a new conformity assessment, unless the change was pre-determined and documented at the first assessment, as with planned continuous learning (Art. 43(4)). Providers of general-purpose AI models (from 2 Aug 2025) must keep technical documentation, have a policy to comply with EU copyright law including text-and-data-mining opt-outs, and publish a sufficiently detailed summary of training content (Art. 53). Research, testing and development before a system is placed on the market or put into service is outside the Act, except testing in real-world conditions (Art. 2(8)). Since the 2026 Omnibus, the Art. 4 AI-literacy duty is an obligation of effort (take measures to support literacy), not of result. Fines reach EUR 35 million or 7% of global turnover for prohibited practices.

  • 2024-08-01 — Entered into force
  • 2025-02-02 — Prohibited practices (Art. 5) and the AI-literacy duty (Art. 4) apply
  • 2026-07-27 — Omnibus softens Art. 4: providers and deployers must take measures to support AI literacy; no specific level must be guaranteed
  • 2025-08-02 — General-purpose AI model obligations apply; governance and penalties regime in place
  • 2026-08-02 — Transparency duties (Art. 50) apply: disclose AI interaction, label synthetic and deepfake content (marking for generative systems already on the market: 2 Dec 2026)
  • 2027-12-02 — High-risk obligations for Annex III systems (e.g. hiring, credit, education, essential services) - moved from 2 Aug 2026 by the 2026 Omnibus
  • 2028-08-02 — High-risk obligations for AI in products regulated under Annex I

Last verified 2026-10-06 · official text

Worker consultation on workplace technology · EU member states

AI Act Art. 26(7); national co-determination law, e.g. Germany BetrVG s.87(1) no. 6, Netherlands WOR art. 27

Introducing systems that can monitor or assess employees usually requires informing or obtaining the consent of works councils or employee representatives, depending on the country. Plan this before a pilot, not after.

Last verified 2026-10-06

References

  1. Stanford Digital Economy Lab (Erik Brynjolfsson, Bharat Chandar, Ruyu Chen). No Widespread Displacement, but the AI Employment Gap for Young Workers Has Widened to 19%. Stanford Digital Economy Lab. 2026.
  2. The Budget Lab at Yale. Tracking the Impact of AI on the Labor Market. Yale University. 2026.
  3. David H. Autor, Frank Levy and Richard J. Murnane. The Skill Content of Recent Technological Change: An Empirical Exploration. The Quarterly Journal of Economics 118(4). 2003.
  4. Tyna Eloundou, Sam Manning, Pamela Mishkin and Daniel Rock. GPTs are GPTs: Labor market impact potential of LLMs. Science 384(6702), 1306-1308. 2024.
  5. David Autor, Caroline Chin, Anna Salomons and Bryan Seegmiller. New Frontiers: The Origins and Content of New Work, 1940-2018. The Quarterly Journal of Economics 139(3), 1399-1465. 2024.
  6. Erik Brynjolfsson, Danielle Li and Lindsey Raymond. Generative AI at Work. The Quarterly Journal of Economics 140(2). 2025.
  7. OECD. OECD Employment Outlook 2023: Artificial Intelligence and the Labour Market. OECD Publishing. 2023.
  8. World Economic Forum. The Future of Jobs Report 2025. World Economic Forum. 2025.
  9. Robert C. Allen. Engels' pause: Technical change, capital accumulation, and inequality in the British industrial revolution. Explorations in Economic History 46(4), 418-435. 2009.
  10. Pew Research Center. U.S. Workers Are More Worried Than Hopeful About Future AI Use in the Workplace. Pew Research Center. 2025.
  11. European Parliament and Council of the European Union. Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Official Journal of the European Union. 2024.
  12. Ingka Group. AI and Remote Selling bring IKEA design expertise to the many. Ingka Group newsroom. 2023.
  13. Reuters. Ikea bets on remote interior design as AI changes sales strategy. Reuters (via Yahoo Finance). 2023.
  14. Retail Gazette. Ikea owner Ingka to cut 800 roles as CEO says it has "grown too complex". Retail Gazette. 2026.

Further reading

Sources last verified 2026-10-09.