The most telling phrase in today's AI-jobs debate is: so far.

So far, we haven't seen mass unemployment. Instead, we're witnessing workflow redesign, widespread retraining, and hiring shifts toward AI-literate roles. AI exposes roughly 40% of global jobs to transformation (IMF, 2024), with higher exposure (~60%) in advanced economies—yet large-scale displacement remains limited. Entry-level and routine-task roles show the earliest strain.

But "so far" isn't a strategy.

The trajectory looks concerning. AI absorbs tasks; workers who master AI tools absorb more responsibility; the performance bar rises for everyone. When AI-enhanced robotics moves beyond current niches—global industrial robot stock reached ~4.66 million units in 2024, having doubled over the past decade (IFR World Robotics 2025)—job compression could accelerate.

The economic ground is already shifting. "Services-as-Software" (AI-driven replacement of labor-intensive services) is projected to surge from under $20 billion in 2025 to as much as $700 billion by 2028 and potentially $1.5 trillion by 2035, potentially representing ~20% of IT spend (versus near-zero today; Bernstein/Gartner; HfS Research). Meanwhile, 75% of Global 2000 companies have signaled that legacy "bums-on-seats" delivery models are no longer viable (HfS Research Pulse Study, recent data).

My rough, non-formal estimate: if adaptation lags, 50–80% of roles could face deep transformation, shrinkage, or displacement over the next 5–10 years.

What sets this apart from past disruptions?

Historical shocks were buffered by productivity gains that lowered costs, unleashed pent-up consumption in unsaturated markets, and created new jobs over long (often multi-decade) lags.

This time, the buffer thins dramatically—though unevenly across the globe—because consumption growth faces hard, compounding ceilings:

In mature/advanced economies, material saturation and diminishing marginal utility limit demand waves. Basics and many discretionary needs are largely met; further income/productivity gains shift spending to other categories, but overall expansion plateaus (extending Engel's Law principles beyond food to general consumption patterns in affluent societies, where expenditure shares on necessities fall and aggregate demand growth slows as markets saturate).

Globally, the most binding constraint remains: individual time is strictly limited to ~24 hours per day. This absolute ceiling no AI abundance overcomes. People can't consume infinite personalized services, experiences, entertainment, education, or leisure faster than their waking hours allow. Many AI-enabled "new" demands (hyper-custom content, virtual worlds, bespoke coaching) are highly time-intensive, so demand asymptotes rather than explodes indefinitely. In affluent societies, time becomes the scarcest resource ("time famine" amid material plenty), rushing consumption rather than expanding it unboundedly.

In emerging and developing economies, there's considerably more room for consumption growth—driven by rising incomes, urbanization, demographic dividends (e.g., expanding working-age populations in India/Sub-Saharan Africa supplying nearly two-thirds of new workforce entrants), and unmet needs in basics/discretionary goods/services. This could unleash stronger demand-led job creation, helping offset AI displacement regionally and sustaining higher net gains than in saturated markets (IMF notes lower immediate AI exposure at ~40% in emerging markets and ~26% in low-income countries vs. ~60% in advanced; WEF highlights growing workforces fueling roles in education/healthcare).

However, ecological limits impose a global cap: Planetary boundaries (7 of 9 breached per 2025 Planetary Health Check, including newly crossed ocean acidification; worsening trends in climate, biosphere integrity, land/freshwater change, biogeochemical flows, novel entities) constrain how far aggregate consumption can rise sustainably. AI-driven productivity could accelerate resource pressures (e.g., data centers projected to more than double both electricity demand and water consumption by 2030, risking further overshoot) unless radically decoupled—potentially limiting even emerging-market upside and widening global divides.

Compounding this: AI's rapid adaptability closes the loop quickest—driven by algorithmic progress, compute scaling, and agentic architectures that enable near-immediate workflow integration and self-improvement. Unlike electrification or computing (decades for new industries before further automation), AI iterates quarterly: frontier models improve reasoning/planning/tool-use rapidly, and agentic AI (autonomous systems that plan, execute multi-step tasks, and orchestrate sub-agents) deploys fast. Gartner (2025) predicts 40% of enterprise apps will integrate task-specific agents by end-2026 (up from <5% in 2025), shifting from assistants to proactive workflow partners. Forrester (Nov 2025) highlights role-based AI agents that orchestrate and complete tasks across multiple systems, reshaping workflows and business models. This speed means emerging fields/roles get automated or de-emphasized almost as fast as they appear—e.g., prompt engineering boomed briefly in 2023–2024 then faced partial automation via better models and agent techniques (as predicted by figures like Sam Altman). New "human oversight," "agent orchestrator," or hybrid roles risk similarly short lifecycles as agents gain autonomy and self-reflection.

Together, these shrink—or eliminate—the adaptation window history provided, though with sharper risks in saturated economies and ecological trade-offs globally. Productivity may surge, but without proportional, sustainable new demand (throttled by saturation/time ceilings/ecological boundaries) or sustained job creation (throttled by rapid re-automation), net gains become less assured, and churn more painful/polarized.

It could be a worst-case view—but isn't this the direction of travel right now?

While the World Economic Forum's Future of Jobs Report 2025 projects a net gain of 78 million jobs by 2030 (170 million created vs. 92 million displaced, affecting ~22% of formal employment), it leans optimistic. It assumes productivity surges from AI (86–96% of employers expect major transformation) will drive new demand and jobs through economic growth and complementarity, without addressing potential ceilings on consumption expansion (material saturation in mature economies or the absolute 24-hour time limit per person). It also integrates AI's rapid adaptability (e.g., GenAI surges, robotics at 58%) but frames it as net-augmenting rather than risking quick re-automation of emerging roles. If those ceilings bind or adaptation lags more severely, the high-velocity churn could prove more disruptive than the "net positive" suggests.

Notably, 40% of employers plan workforce reductions in automatable areas while aggressively hiring for AI-related skills. That's not a gentle transition; it's high-velocity churn, growing polarization, and an urgent race to upskill.

What's your take? What am I under- or over-weighting here? And—most practically—how do societies hold together through churn at this scale without leaving large cohorts behind?

Sources:

(1) IMF (Jan 2024): https://www.imf.org/en/Blogs/Articles/2024/01/14/ai-will-transform-the-global-economy-lets-make-sure-it-benefits-humanity

(2) IFR World Robotics 2025: https://ifr.org/ifr-press-releases/news/global-robot-demand-in-factories-doubles-over-10-years

(3) Bernstein/Gartner on Services-as-Software: https://www.investing.com/news/stock-market-news/everything-you-need-to-know-about-the-new-trend-servicesassoftware-4494461

(4) HfS Research on $1.5T opportunity: https://www.horsesforsources.com/sas_1-5-trillion-dollar-opportunity_020725/

(5) HfS Research Pulse Study (recent): https://www.horsesforsources.com/last_18-months_labor_intentive_services_012826/

(6) WEF Future of Jobs Report 2025 (Jan 2025): https://www.weforum.org/publications/the-future-of-jobs-report-2025/

(7) Gartner on task-specific AI agents (Aug 2025): https://www.gartner.com/en/newsroom/press-releases/2025-08-26-gartner-predicts-40-percent-of-enterprise-apps-will-feature-task-specific-ai-agents-by-2026-up-from-less-than-5-percent-in-2025

(8) Forrester Predictions 2026 on AI agents/workflows (Nov 2025): https://www.forrester.com/blogs/predictions-2026-ai-agents-changing-business-models-and-workplace-culture-impact-enterprise-software

(9) Various on prompt engineering evolution (e.g., Sam Altman comments; general 2025 reporting): No single definitive source for full obsolescence, but aligned with industry shifts toward agents (cross-referenced in Gartner/Forrester contexts above)

(10) IEA Energy and AI Report (April 2025): https://www.iea.org/reports/energy-and-ai/executive-summary (data center electricity and water consumption projected to more than double by 2030)

(11) Planetary Health Check 2025 (Stockholm Resilience Centre/PIK, Sept 2025): https://www.stockholmresilience.org/news--events/general-news/2025-09-24-seven-of-nine-planetary-boundaries-now-breached.html (7 of 9 planetary boundaries breached, all showing worsening trends)

#AI #FutureOfWork #Jobs #WorkforceTransformation #Upskilling