AI and Work in 2026

The Signal:
Across seven Fortune 500 role profiles tracked in 2024 and 2025, AI absorbed the lowest-value work inside each role while the human responsibility set expanded upward. No portrait shows elimination. Every portrait shows transformation.

• Junior Consultant, Mumbai: AI handled approximately 60% of analyst tasks by mid-2024, including document review, desk research, first drafts, and baseline models. The firm did not reduce headcount. Analysts two years out started leading projects that previously required a senior manager. Base pay rose 28% in two years. Draup found AI-engineering wages in India rose 35% between 2023 and 2025. ‍
• Mid-Career Paralegal, Cleveland: Caseload is up 40%. Headcount is up 20%. Document review is 3x the 2022 level. AI-assisted review cuts human error 20 to 25%. Risk, Compliance, and Governance postings now list 3.64 additional skills, mostly about working with AI, not being replaced by it. ‍
• Radiologist, Rural Midwest: A critical-access hospital with no on-site radiologist before 2024 now operates with a part-time radiologist two days a week. Imaging volume doubled. Correctly identified abnormalities rose 6 to 26 percentage points with AI in the loop. The role did not disappear. It became possible where it could not exist before. ‍
• Customer Support Agent, Manila: A chatbot handles 40% of tickets end-to-end and another 35% with a human in the loop. The remaining 25%, angry, confused, grieving, unusual cases, drive most of retention. Seats are up 15% since 2023. Pay rose. The work shifted toward what the firm calls emotional quality. ‍
• Financial Analyst, London: In 2022, a typical analyst's week was 55% data gathering and 45% analysis. By 2024, AI handles most of the data gathering. Each analyst now covers 30 to 40% more companies. The FCA's 2025 monitoring found no net reduction in qualified analyst roles. AI-fluency premium is 22 to 26% in London, Frankfurt, and Amsterdam. ‍
• Small-Town Accountant, Eastern Pennsylvania: Net income is up 22% in two years, the best stretch of her career. She absorbed clients from a regional firm that closed in 2024. Contract and specialist hiring is up 17.3% year-over-year across the sector. ‍
• New-Graduate Software Engineer, Atlanta: Starting salaries at a Fortune 500 consumer-goods firm are near $118,000, roughly 15% above the prior year. Firms leading AI deployment grew entry-level hiring in 2025 and again in 2026. AI governance and model-risk skills are growing 81% year-over-year.

What It Means:
Draup identifies this as the workload iceberg effect. AI takes the tip: the 2.2% of visible routine tasks that represent only 11.7% of wage value. The human role, judgment, oversight, client relationship, regulatory accountability, keeps the iceberg. The absorbed tasks are the most automatable, not the most valuable.

Strategic Implications:

The Signal:
The standard displacement argument assumes that when AI raises worker productivity, firms reduce headcount to maintain output at lower cost. The evidence from Fortune 500 hiring data does not support this assumption. In every portrait studied, productivity gains converted into expanded output rather than reduced workforce.

Key Data Points:

• The paralegal firm's caseload grew 40% while headcount grew 20%. Productivity went to volume, not layoffs.
• The Manila support center's seat count grew 15% after chatbot deployment. Easy tickets moved to automation; hard tickets moved to humans.
• The financial analyst now covers 30 to 40% more companies with the same headcount. Firm output expanded.
• The accountant absorbed clients from a closed competitor. Her capacity increased through AI tools. She took on more, not fewer, clients. Net income rose 22%.
• WEF projects 78 million net new jobs globally by 2030. The Yale Budget Lab finds no aggregate employment disruption through end of 2025.

What It Means:
Demand for expert services is more elastic than first-order displacement models assume. When AI makes professionals faster, clients do not receive the same amount of work at lower cost. They receive more work, broader coverage, and services they previously could not access. AI relaxes the supply constraint. Demand expands to fill it.

Strategic Implications:

The Signal:
Global unemployment hit 4.81% in 2024, the lowest in the World Bank and ILO dataset going back to 1991. The two prior shocks were the 2009 Global Financial Crisis, which peaked at 6.41%, and COVID-19 in 2020, which peaked at 6.59%. Both recovered. The AI era began not at a fragile high but at a 34-year low. The decline did not pause when AI went mainstream in 2022. It accelerated.

Key Data Points:

‍

• High-income economies, the most AI-exposed, with large professional services, finance, legal, and technology sectors, averaged 4.37% in 2024, below their 2019 pre-pandemic baseline of 4.76%.
• Not one high-income economy shows a rise in unemployment after November 2022, the month ChatGPT launched.
• Every country above 20% unemployment in 2024 sits in Sub-Saharan Africa. Eswatini: 34.64%. South Africa: 32.28%. Botswana: 23.81%. These figures reflect governance and structural economic conditions that long predate AI.
• MENA has run above 9% for three decades through every technology wave. East Asia has held near 3 to 4% the whole time. AI did not put these lines where they are.

What It Means:
The countries most exposed to AI automation have the tightest labor markets in the dataset. This is the clearest empirical test of AI displacement available over the past three years, and it does not support the claim. The AI risk map in most forecasting models is actually the formality map: higher-income countries show higher measured unemployment because they count informal workers as jobless. That structural measurement difference has nothing to do with AI.

Strategic Implications:

The Signal:

The prediction that AI would suppress wages by reducing the scarcity of cognitive labor has not materialized in the data. Across every portrait studied and in aggregate US earnings data, workers whose roles intersect with AI are seeing pay rise faster, not slower, than their peers.

Key Data Points:

‍What It Means:
The countries most exposed to AI automation have the tightest labor markets in the dataset. This is the clearest empirical test of AI displacement available over the past three years, and it does not support the claim. The AI risk map in most forecasting models is actually the formality map: higher-income countries show higher measured unemployment because they count informal workers as jobless. That structural measurement difference has nothing to do with AI.

Strategic Implications: