When a government's top scientists sit down to formally war-game the future of AI, it's worth reading what they conclude — because governments plan with money, regulation, and power, and their forecasts shape all three. On June 15, 2026, the UK's Government Office for Science, led by Chief Scientific Adviser Professor Dame Angela McLean, published exactly that: "AI Scenarios 2030," five official, internally-coherent futures for where AI lands by 2030, built to stress-test government policy.
Most people will never read a 92-page government foresight report. That's a mistake, because stripped of the policy jargon, this document is one of the clearest, most sober maps of where the AI economy could go — and every one of the five scenarios is really a different answer to the question this whole site cares about: who ends up with the money? Let me walk you through what the UK government actually concluded, the numbers that shocked its own analysts, and what each future means for your job, your business, and your investments.
What this document actually is
First, why take it seriously. This isn't a think-tank op-ed or a vendor's hype deck. It's the official foresight product of GO-Science, produced with the AI Security Institute and the Department for Science, Innovation and Technology, drawing on experts across government, academia, and industry. It's explicitly used inside government to stress-test real policy — and it carries the signature of the UK's Chief Scientific Adviser.
Crucially, the authors are careful about what it is and isn't: "These are tools for exploring uncertainty… They are not predictions." The real future, they say, will likely blend elements of all five and could slide from one into another. That humility is exactly what makes it useful — instead of one confident (and probably wrong) forecast, you get five coherent maps and a method for thinking about which one you're living in. It's the same disciplined "plan for the range, not the point estimate" approach that separates good strategy from wishful thinking, whether you run a country or a company.
The numbers that made the government redo its homework
The UK first built AI 2030 scenarios in 2023. It rebuilt them now because the ground moved so fast the old maps went stale. The specific figures they cite are worth sitting with, because they're the government's own evidence for how quickly this is accelerating:
| Metric | Then | Now |
|---|---|---|
| Autonomous task length (frontier AI, ~50% success, software) | ~4 minutes (Mar 2024) | ~12 hours (Feb 2026) |
| ChatGPT weekly users | ~200M (mid-2024) | ~800M (late 2025) |
| Big Tech AI capex | baseline 2023 | more than doubled by 2025 |
| China's open-weight position | negligible | global leader (post-DeepSeek) |
That first row is the one that should stop you. In under two years, the length of a task a frontier model can do on its own — with even odds — went from four minutes to twelve hours, with early signs of week-long tasks coming. The report is careful to note this is task-specific and measured at partial reliability, not blanket autonomy. But the trajectory is the story: machines are climbing from "helps with a snippet" to "runs a multi-hour project alone." That single curve is the engine under every one of the five scenarios, and it's the same capability leap I traced in the frontier-model coding race. The report also flags the geopolitical turn: after DeepSeek's cheap model shocked the West, China became the leader in open-weight AI, ending America's uncontested lead.
The six forces that decide which future you get
The scenarios aren't random. They're built from six "critical uncertainties" — the variables the government judged most powerful and most unknowable. Understanding these six is genuinely useful, because they're the dials that determine where the money flows:
| Critical uncertainty | The question it asks |
| Capability | How far past today's abilities does AI go, and in which domains? |
| Distribution & access | How concentrated is frontier AI? Open vs closed? Cheap or expensive? |
| Security | Does AI stay controllable? How badly do bad actors exploit it? |
| Adoption | How widely is it used? How much autonomy do we grant? Does the public trust it? |
| Labour displacement | Does AI complement workers, or replace them? |
| Global cooperation | Do nations cooperate, or does competition and militarisation escalate? |
Notice how many of these are money questions in disguise. "Distribution" is really "does the wealth concentrate in a few frontier firms or spread out?" "Labour displacement" is "whose income survives?" These six dials, turned different ways, produce the five futures.
The five futures, from Slow Burn to Take-Off
Here's the heart of it — the five scenarios, grouped by how fast AI capability advances. Read each as a different economy:
| Scenario | Trajectory | The money picture |
| 1. Slow Burn | AI progress slows | Limited adoption, minimal economic uplift, displacement concentrated in a few roles, harms mostly contained |
| 2. Open Frontier | Progress slows but disrupts anyway | Open-weight models spread; modest economic uplift; considerable labour displacement; harder to contain harms |
| 3. Augmented Growth | Progress continues; humans stay "in the loop" | Economic boom, many new jobs created, international standards keep systems mostly secure — the optimistic case |
| 4. Transformation Economy | Progress continues; humans pushed "out of the loop" | Widespread labour displacement, economic tension, and profits "largely accrue overseas" — growth without broad benefit |
| 5. Take-Off | Progress rapidly accelerates | Substantial growth and widespread displacement; safety deprioritised amid race dynamics; misaligned systems pose severe risk |
Sit with the difference between Scenario 3 and Scenario 4, because that's the whole ballgame for ordinary people. Both have AI advancing and the economy growing — but in Augmented Growth, humans stay in the loop, new jobs appear, and the gains spread. In Transformation Economy, humans get pushed out, displacement is widespread, and the profits flow to whoever owns the AI (often "overseas"). Same technology, same growth number — wildly different answer to "did it help me or replace me?" The dividing line isn't capability. It's whether the humans stay in the loop and whether the gains are shared.
The money thread running through all five
Here's what the government concluded that matters most for your wallet, pulled straight from its key findings — and it echoes everything I keep documenting on this site.
The frontier stays concentrated, and that concentrates wealth. The report is blunt: the frontier AI market is expected to stay dominated by a few large firms, so "a large proportion of the gains from AI accrues to frontier firms, to owners of capital invested in those firms, and to those controlling key inputs, potentially contributing to rising inequality." That is the UK government, in its own words, describing exactly the dynamic I laid out in how the AI wealth boom locks ordinary people out and who's really funding it — the sovereign wealth funds. When the official foresight report and the money story agree, pay attention.
But capability commoditises behind the frontier. The flip side, also in the report: while the frontier concentrates, the layer just behind it gets cheap and everywhere — "widely available models embedded across a growing range of use cases." That's the open-weight wave — China's GLM and the free-model movement — and it's the opening for everyone who isn't a frontier lab.
Labour impact is significant but not simple. The government's finding: AI "could cause significant labour displacement by 2030," while also complementing some workers with higher wages and opportunities. Even in mild scenarios, "the nature of work is likely to change, with routine, execution-oriented tasks increasingly becoming automated." That's the exact split I keep hammering — the Gartner data on AI and jobs and which roles AI hits: the routine execution layer erodes, the judgment-and-direction layer gets paid more.
Adoption is uneven, which widens gaps. The report warns that adoption speed varies, so "certain organisations, sectors, or nations capture disproportionate productivity gains." Translation: the early, skilled adopters pull away from everyone else. That's not fate — it's an invitation to be on the right side of the gap.
Why scenario thinking is a tool for you, not just governments
Here's the part I want you to actually steal from this report, because it's the most valuable thing in it. The UK government doesn't use these scenarios to predict the future. It uses them to stress-test decisions — to ask, "if the world turns out like Scenario 4, does my policy still work? What breaks?" That's a technique any business owner or individual can copy for free.
Instead of betting your career or company on one guess about AI, run your plan against all five. If you're building a business, ask: does it survive Slow Burn (AI underdelivers) and thrive in Take-Off (AI explodes)? If you're choosing a career, ask: is it exposed in Transformation Economy where humans get pushed out, or resilient in Augmented Growth where they stay in the loop? This is the same "don't single-source your future" discipline I drew from the Fable 5 shutdown — build for a range of outcomes, not one bet. The government spent a fortune and 92 pages building this thinking tool. You can apply it to your own life this afternoon.
What this means for you
Depending on where you sit, here's the practical read.
If you run or build a business, use the five scenarios as a free strategy exercise. The safest bets are the ones that pay off across multiple futures — and the report's own logic points to the application and deployment layer, since "AI-enabled gains are expected to become the main source of the UK's continued productivity growth." Position to capture productivity gains regardless of which scenario lands, and lean on the commoditising sub-frontier models rather than betting everything on one expensive provider.
If you're thinking about your career, the government just handed you a risk map. Routine, execution-oriented work is flagged as automatable across every scenario — while workers who get complemented by AI see higher wages. The move is to become the human who directs, judges, and stays "in the loop," exactly the skill premium I track in the highest-paying AI-era jobs. The scenarios differ on speed, not direction.
If you invest or just want to understand the board, the report confirms the two-layer reality: enormous, concentrated value at the frontier (great for the owners of NVIDIA, the frontier labs, and their backers), and a fierce, commoditising, US-versus-China contest everywhere else — the same great-power AI race the China strategy piece maps, now echoed by the UK's own analysts. And note the regulatory signal: a government that publishes existential-risk warnings is a government preparing to regulate, the direction I traced in the EU AI Act transparency rules.
The honest take
What strikes me most about this report isn't any single scenario — it's that a cautious, sober government body, not prone to hype, concluded that "AI will have a profound impact by 2030" in every future it could construct, up to and including "serious, potentially even existential harms… without government intervention." When the people whose job is measured skepticism can't build a plausible 2030 where AI doesn't reshape the economy, the debate about whether this is real is over. The only open question is which version of real you get.
And that question — Augmented Growth where humans stay in the loop and the gains spread, versus Transformation Economy where they don't — isn't decided by the technology. It's decided by policy, by ownership, and by who positions themselves to benefit. The government is preparing for all five. The wealthy and the frontier firms are positioned for all five. The people who get caught out will be the ones who assumed the future was fixed and did nothing. You've now seen the same map the policymakers are using.
So here's the question the UK government is effectively asking every department, and that you should ask yourself: if the world in 2030 looks like this scenario, what would it mean for me — and what should I start doing today to be ready for more than one of them?
Source: UK Government Office for Science — AI Scenarios 2030.
