The rise of artificial intelligence (AI) marks one of the most substantial technological shifts of our time. For the global workforce, the adoption of AI brings both promise and peril: new productivity possibilities sit alongside the risk that entire job categories may vanish. This article explores four key themes: (i) which jobs are likely to disappear over the next ten years and why; (ii) what percentage of the world population is expected to lose jobs; (iii) whether new jobs will replace those lost; and (iv) our perspective that in many cases job losses may not be matched by new job creation.
1. Which jobs are likely to disappear within the next ten years — and why?
Taking a task-based view of work helps identify the roles most vulnerable to AI substitution. Jobs characterised by routine, repetitive, structured tasks are particularly exposed. Entry-level customer service, administrative and secretarial functions, basic coding and software maintenance, and junior legal or accounting support roles face heightened automation risk as large language models and decision-automation tools outperform humans on pattern-recognition and rules-based tasks. A number of economic analyses suggest AI’s displacement effects will be concentrated in low- and mid-skill cognitive roles, while complementing higher-skill decision-making work. One major research house projects only a modest, transitional rise in unemployment overall, but acknowledges material disruption for specific occupations as AI diffuses through firms and workflows [1].
2. What percentage of the world population is expected to lose jobs?
There is no single definitive figure, but credible institutions frame the scale of exposure. The International Monetary Fund estimates that about 40% of jobs globally are exposed to AI (around 60% in advanced economies and 40% in emerging markets), with some replaced and others complemented depending on policy and firm choices [2]. If the global labour force is ~3.5 billion, even single-digit percentage displacement would equate to tens to hundreds of millions of jobs at risk. Yet macro-level forecasts from private-sector economists also indicate the aggregate unemployment effect could be relatively small and temporary if economies adapt, even while certain cohorts bear concentrated losses [1].
3. Will new jobs replace the old ones?
Historically, general-purpose technologies created more jobs than they destroyed, but transitions were uneven and protracted. With AI, new roles are emerging in human-machine teaming, data stewardship, AI safety and oversight, and domain-specific expert services. However, skills mismatches, geographic immobility and the capital intensity of AI deployment mean new jobs may not fully offset losses for many displaced workers without targeted reskilling and active labour-market policies. Recent labour-market reads in some economies show no broad “jobs apocalypse” so far; nonetheless, the transition pathways are narrow for mid- and low-skill workers unless policy support scales alongside firm adoption [3].
4. Our view: many jobs will be lost — and not many created.
We judge the next decade will be shaped more by job erosion and transformation than by one-for-one replacement. Exposure is large; the most readily automatable roles are high-volume; and new jobs skew toward higher skills and fewer headcounts. The lag between displacement and creation risks prolonged under-employment. Without robust policy and employer action, structural unemployment could rise in exposed regions and occupations. Social protections, reskilling at scale, and labour-mobility support will be essential to cushion shocks and broaden access to emerging roles [2].
5. Physical AI will affect many more jobs
Beyond software and back-office automation, the advent of “physical AI” — robots and autonomous systems powered by foundation models and simulation — is poised to expand automation into warehouses, factories, logistics and field operations. A leading semiconductor firm has introduced a generative-AI platform (Project GR00T) designed specifically to teach humanoid robots via language and demonstration, bringing manipulation and mobility closer to human-level versatility [4]. Its CEO has argued that humanoids are fewer than five years away from wide manufacturing use, signalling an acceleration window for physical automation [5].
At the same time, global electronics manufacturers and logistics giants are piloting humanoids and advanced warehouse robots at scale, with reports of deployment discussions for production facilities and active trials in fulfilment networks. These programmes aim to automate repetitive pick-and-place, kitting and movement tasks that employ large numbers of workers today [6][7]. Market analyses suggest a substantial addressable market for humanoid robots by the mid-2030s, signalling economic incentives to substitute labour in select tasks once reliability thresholds are met [8].
The implication is straightforward: as physical AI matures, risk extends from clerical and creative cognitive work to a broad array of manual, semi-skilled roles in warehousing, light manufacturing, retail back-of-house and last-metre logistics. For affected workers, reskilling into robot-operations, maintenance and quality roles is possible — but headcounts in these higher-skill jobs are far smaller than the roles at risk without proactive workforce strategies.
6. Acceptance of FSD (Full Self Drive) and robotaxis could displace millions of driving jobs over time
Driving is among the world’s largest occupational categories across freight, ride-hail, delivery and public transport. While exposure differs by country and mode, advances in autonomous driving and FSD-style systems create credible medium-term displacement pressure. Commercial pilots are accelerating: one major robotaxi operator plans its first international service in London in 2026, building on large U.S. operations, and other players are bringing production capacity online [9][10].
On the FSD front, announcements in 2025 pointed to initial robotaxi rollouts with limited fleets and geofenced zones, indicating early demand-side acceptance and supply-side caution as systems mature. In the U.S., a high-profile manufacturer signalled a tentative start to a driverless taxi service, while regulators continue to probe safety performance — a reminder that adoption speed will hinge on demonstrable risk reduction and compliance [11][12].
Freight is a particularly strong near-term candidate for autonomy due to long highway segments and significant operating-cost savings. Analyses indicate autonomous trucking could materially lower per-mile costs, altering labour demand in long-haul segments first, with knock-on effects to regional and last-mile work as ecosystems adapt [13]. Given the sheer size of transport employment tracked by international labour statistics, even gradual adoption may translate into large absolute numbers of jobs restructured or eliminated over time [14].
Our assessment: broad acceptance of robotaxis and FSD-grade systems will likely unfold city-by-city and corridor-by-corridor under tight regulation. As reliability improves and costs fall, displacement could scale, especially in routine, high-exposure driving tasks. New roles (remote fleet supervision, AV operations, mapping, maintenance) will arise, but again with far fewer seats than traditional driving occupations, reinforcing the need for transition plans.
Summary: AI will certainly reshape employment. Roles that rely on routine, structured, replicable tasks are most vulnerable. The percentage of jobs at risk could be material — tens to hundreds of millions globally — though full-scale job loss is not inevitable. New jobs will emerge, but may not match the volume or profile of those lost, especially for mid- and low-skill workers. With the rise of physical AI and the gradual acceptance of FSD/robotaxis, automation pressure expands into warehousing, manufacturing and transport, where headcounts are large. In our view, the next decade is likely to see more job disruption and attrition than seamless replacement. For many workers, the challenge will not only be surviving job-loss, but finding pathways to meaningful new roles.
[1] How Will AI Affect the Global Workforce? — link
[2] AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity. — link
[3] New data show no AI jobs apocalypse—for now — link
[4] Nvidia adds generative AI to power humanoid robots — link
[5] Nvidia CEO: Humanoid robot revolution is closer than you think — link
[6] Nvidia, Foxconn in talks to deploy humanoid robots at Houston plant — link
[7] Amazon rolls out robotic system, begins testing bipedal ‘Digit’ — link
[8] Humanoid robots offer disruption and promise — link
[9] Waymo to launch autonomous ride-hailing service in London — link
[10] Zoox launches first robotaxi production facility — link
[11] Musk says robotaxi service to ‘tentatively’ launch in Austin — link
[12] U.S. probes Tesla FSD for traffic-violation incidents — link
[13] Will autonomy usher in the future of truck freight transportation? — link
[14] ILOSTAT — Sectoral employment statistics (transport and road) — link
beFirstComment