What Could AI Do to the Isle of Man Economy? Three Scenarios

The short version

Anthropic's research institute published the most rigorous economic model yet for what AI could do to a national economy. It models three scenarios for the United States. We've adapted it for the Isle of Man.

The headline findings for the IoM:

• Even modest AI adoption lifts GDP by ~2% and barely moves unemployment — but cognitive workers (68% of the island) see almost no wage gain while everyone else benefits
• Substantial adoption lifts GDP by ~11% — transformative for an economy that's been flat since 2016 — but unemployment triples from 0.6% to 2.0% and cognitive wages actually fall
• Extreme adoption lifts GDP by ~50% but creates 5.6% unemployment, a level the island has never experienced, while the labour share of income drops from 55% to 44%

The island's structural differences from the US — higher cognitive workforce share, tighter labour market, heavier services dependence, smaller population — amplify both the opportunity and the disruption.

Explore the full interactive model

The paper behind this

In September 2026, Korinek, Jones, Sacher, Cotter, and McCrory published Economic Scenarios for Transformative AI through The Anthropic Institute. Rather than a single forecast, it builds three internally consistent scenarios with different assumptions about AI capability and adoption speed, then traces the economic consequences of each.

The paper models the US economy. We needed to know what it meant for the Isle of Man.

Why the IoM is different

The Isle of Man is not a miniature United States. The structural differences are large enough that the US results can't simply be scaled down.

Every one of these differences changes what AI does to the economy. A higher cognitive share means more tasks in AI's path. Tighter labour markets mean less capacity to absorb displacement. A flat economy means AI could be the only growth driver — raising the stakes considerably.

Three scenarios, three futures

Modest: "AI as a better spreadsheet"

AI affects 22% of tasks, 20% adoption rate, 30% productivity gain per affected task. Half the impact is automation, half augmentation. New tasks replace half the displaced work.

Noticeable but not transformative. The key signal: cognitive workers — 68% of the island's workforce — see almost no wage benefit while everyone else gains.

Substantial: "real disruption, real growth"

AI reaches 35% of tasks, 40% adoption, 45% productivity gain. Three-quarters is automation, one-quarter augmentation. Only 25% of displaced work is reinstated.

This is where it gets serious. GDP growth of 11% would be extraordinary for an economy that's been flat for a decade. But cognitive wages fall, the labour share of income drops from 55% to 51.5%, and unemployment triples. The IoM has ~260 unemployed people today. Tripling that means ~500 additional workers competing in a market with no experience managing reallocation at scale.

Extreme: "enormous wealth, profound inequality"

AI affects 55% of tasks, 60% adoption, 80% productivity gain. 90% automation, zero reinstatement of new tasks.

GDP nearly doubles. But cognitive wages fall while other wages almost double. The labour share drops to 44%. And 5.6% unemployment — nearly ten times the current level — would be a genuine crisis for an island whose support services, retraining infrastructure, and housing market are all calibrated for near-full employment. The social infrastructure simply doesn't exist for it.

What this means for the Isle of Man

Three things should concern policymakers:

1. The majority of workers lose out relatively, in every scenario. With 68% of the workforce in cognitive roles, even small cognitive wage declines affect most households. The benefits flow disproportionately to non-cognitive workers (who are scarce) and to capital.

2. Small unemployment numbers mask real disruption. Going from 0.6% to 2% sounds modest. But the IoM has never had to absorb occupational reallocation at this scale. There is no institutional memory of how to manage it.

3. The upside is genuinely transformative — if captured. An economy that's been flat for a decade gaining 11% of GDP would change the fiscal picture entirely. The question is whether the gains are distributed in a way that maintains social cohesion, or whether they concentrate in capital and a shrinking pool of high-demand occupations.

What this actually means for IoM jobs

"Cognitive workers" is an economic abstraction. Here's what it means in practice for the island's actual sectors.

eGaming (15% of GDP)

eGaming companies employ compliance analysts, platform engineers, data scientists, marketing strategists, and licensing specialists. Almost all of these are cognitive roles. In the substantial scenario, an eGaming compliance analyst currently earning £45,000 could see their real wage fall by 1-2% by 2030 — not because they're fired, but because AI tools handle enough of the routine compliance checking, transaction monitoring, and report drafting that the scarcity value of those skills drops. Meanwhile, the customer support and facilities staff at the same company see their wages rise. The firm's output goes up; the distribution of who benefits shifts.

Financial services and insurance (30% of GDP)

Trust administrators, fund accountants, actuaries, underwriters, regulatory compliance officers — the island's largest employer sector is overwhelmingly cognitive. These are precisely the roles where AI is already demonstrating capability: document review, risk assessment, regulatory reporting, client correspondence. In the substantial scenario, a trust administrator's wage pressure comes not from being replaced, but from AI making a junior administrator as productive as a mid-level one, compressing the experience premium. The firms become more profitable. The question is whether that profit flows to workers, shareholders, or lower fees for clients.

IT and software development

Software developers, systems administrators, data engineers, and IT project managers are among the most directly AI-affected occupations globally. On the island, they work across every sector — eGaming platforms, banking systems, government digital services. Paradoxically, IT workers may experience both the largest productivity gains and the most wage pressure. AI coding tools dramatically increase output per developer, which is great for the individual — until the market adjusts and firms realise they need fewer developers to build the same thing. The extreme scenario's 3.2% cognitive wage decline hits this group hard.

Public administration (IoM Government)

The Isle of Man Government is the island's single largest employer. Civil servants, policy analysts, administrators, social workers, teachers — a substantial share are cognitive roles. Government doesn't respond to wage signals the same way private sectors do. A trust company can cut staff when AI handles compliance; government tends to redeploy. This could make government a buffer against displacement — absorbing workers other sectors release — or it could mean government becomes relatively more expensive as private sector productivity races ahead while public sector wages stay flat.

Care, hospitality, and trades (non-cognitive)

Here's the counterintuitive finding: care workers, hospitality staff, electricians, plumbers, and construction workers are the winners in every scenario. These are the non-cognitive roles that AI cannot easily automate. With 68% of workers in cognitive jobs, the remaining 32% become increasingly scarce. In the substantial scenario, non-cognitive wages rise 17.8%. A care home worker or electrician could see meaningful real wage growth precisely because everyone else is competing for the cognitive jobs that AI is squeezing.

The practical implication: the island's workforce strategy should focus less on "teaching everyone to code" and more on understanding which cognitive roles gain value from AI and which lose it — and ensuring the social infrastructure exists for workers who need to move between them.

What we don't know

We've been transparent about the data gaps. Six key assumptions lack Isle of Man-specific data:

• Labour share — adjusted from a nominal 35% (distorted by corporate domicile profit booking) to 55%, but could be anywhere from 45% to 60%
• Capital supply elasticity — no IoM measurement exists, and this parameter is consequential
• Job flow rates — the IoM publishes no quit rates or occupational switching data
• AI diffusion — no IoM-specific AI adoption survey exists
• SOC mapping — UK and US occupation classifications don't map perfectly
• Small-economy dynamics — concentrated employer markets may not behave like the competitive model assumes

These numbers should be read as "if our assumptions hold, this is the direction and rough magnitude" rather than as forecasts.

Explore the model

The full interactive page includes quarterly projections through 2030, side-by-side IoM vs US comparisons, the complete methodology, and downloadable data. The Workforce Resilience Index and AI Job Guides provide the occupation-level detail that complements these macro scenarios.

The model adapts Korinek, A., Jones, C.I., Sacher, S., Cotter, T., & McCrory, P. (2026). "Economic Scenarios for Transformative AI." The Anthropic Institute Working Paper No. 2026-02.