Half Decent But Half Decent Is Not Good Enough

Scotland AI Strategy

The Productivity Problem

Scotland's productivity gap isn't new. Output per hour has trailed the UK average for years, and the UK itself sits behind France, Germany, and the US. The AI strategy acknowledges this. It explicitly says AI should "close the productivity gap." But acknowledgement isn't a mechanism.

The actual productivity levers in the document are remarkably narrow. You've got the SME adoption programme (500 firms engaged so far out of roughly 370,000 Scottish businesses), modular training courses, an AI Leadership Academy that hasn't launched yet, and a standardised readiness tool. These are all sensible inputs. However, none of them operates at the scale required to shift a national productivity number or anywhere near.

This is the blindingly obvious core issue.

Scotland's productivity problem isn't primarily a technology problem. It's a structural composition problem.

The Scottish economy is weighted toward sectors with historically low productivity growth. Public services, which are way too big for the size of the population, hospitality, tourism, food and drink, and small scale agriculture. The strategy's sectoral focus lands on healthcare, financial services, advanced manufacturing, space tech, and semiconductors. These are high productivity sectors where Scotland already performs reasonably well. The sectors dragging the average down are the ones where AI adoption could make the biggest relative difference, and they barely feature.

Tourism is the perfect example. It's one of Scotland's largest employers, overwhelmingly SME and micro business, labour intensive, and operating on thin margins. It's also a sector where AI can deliver immediate productivity gains: automated enquiry handling, dynamic pricing, review-driven marketing, and operational scheduling. The productivity uplift per business might be modest in absolute terms, but multiplied across thousands of operators, the aggregate impact on national productivity would be significant. The strategy ignores it entirely. How can something of this scale be ignored is beyond my limited brain

The Growth Problem

Scotland's growth challenge is different from its productivity challenge, although they're obviously connected. Growth requires either more people working, or the people working producing more, or entirely new sources of economic value being created. We don't have the infrastructure or services to handle a significant population increase, so we need to increase production or create new value.

The strategy's growth thesis rests on three pillars: attracting inward investment in AI infrastructure, growing a pipeline of AI startups to "unicorn scale valuations," and commercialising university research more effectively. Okay, each is worth a look-see.

Inward investment is real. The Lanarkshire AI Growth Zone (£8 billion private investment), the North Ayrshire AI park (£15 billion), the CoreWeave facility (£2.5 billion) These are genuinely large numbers. But data centre construction creates a one time capital expenditure spike, not sustained GDP growth. The permanent employment from a hyperscale data centre is typically 30 50 people. The 3,400 jobs cited for Lanarkshire include construction and ancillary roles. The strategy doesn't distinguish between temporary construction employment and permanent high value roles. That matters when you're claiming growth impact, and it is something that happens all the time. The renewable industry is the screaming example, and Aberdeen is the screaming proof.

The startup pipeline is where growth should come from, but the strategy's ambition outstrips its toolkit. Scotland has roughly 296 AI focused companies. The strategy wants "unicorn scale valuations" but is piloting an accelerator, not funding one at scale. Compare this to France's €2.2 billion AI investment plan or even Ireland's targeted approach through Enterprise Ireland. The funding gap between aspiration and commitment is substantial. And there's a retention problem the strategy barely addresses: Scottish AI companies that do scale tend to be acquired by or relocate to London, the US, or elsewhere. Wordsmith AI reaching $100 million valuation is celebrated, but the strategy doesn't ask whether the value created will stay in Scotland. I suspect it will not.

Research commercialisation is Scotland's business frustration. World class research, weak translation into commercial outcomes. The strategy proposes a "Venture Creator" pilot and references the Scottish Spin out Report. These are incremental improvements to a system that has been underperforming for decades. Edinburgh's informatics department is genuinely world leading. But the gap between academic paper and revenue generating company remains ocean scale wide, and the strategy doesn't fundamentally redesign the incentives that keep it that way.

What's Missing

Three things the strategy needed to say but didn't.

First, AI driven productivity gains in low productivity sectors matter more than AI innovation in high productivity sectors. To me, this is obvious, but please tell me I am wrong. If you want to move the national number, you need to raise the floor, not just polish the ceiling. A food tour operator in Edinburgh using AI to cut admin time by 10 hours a week contributes more to Scotland's productivity statistics than another fintech startup in the same postcode. The strategy is fixated on the glamorous end of the AI economy and neglects the mass market adoption that actually shifts aggregate productivity.

Second, growth requires demand, not just supply. The strategy is almost entirely supply side: build infrastructure, train people, fund startups, and commercialise research. But Scotland's domestic market is tiny. The strategy says almost nothing about how Scottish businesses AI companies or AI enabled companies in other sectors will access and win in international markets. There's no export strategy embedded in the AI strategy. Scottish Development International gets a passing mention.

For a country of 5.4 million people, international demand generation isn't optional. It's the whole game.

Third, the public sector can't be both the safety net and the growth engine. The strategy wants AI to simultaneously improve public services, build public trust, maintain Fair Work principles, and drive economic growth. These objectives will conflict. AI that genuinely improves NHS productivity will reduce headcount. AI that transforms local government services will eliminate administrative roles in communities where those are the primary employers. I view the reduction in the sectors as a huge positive and should be a detailed objective with KPIs and responsibility to make sure it happens, allocated to named individuals

The strategy doesn't acknowledge these trade offs. It treats all outcomes as simultaneously achievable, which suggests either the hard conversations haven't happened or the political will to have them doesn't exist. I strongly suspect both, but certainly the second.

The Verdict

This strategy will produce some good things. The SME adoption programme will help hundreds of businesses. If implemented well. I say it is due to the history of previous digital awareness programs. The data centre investment will create real infrastructure. The regulatory positioning I also worry about. Aligning with Europe may just be aligning with more slow/no growth.

Will it close Scotland's productivity gap or shift its growth trajectory? No. Not at this scale, not with this funding, and not with these priorities.

The strategy is a half decent government document that describes a desirable future. It is not an economic transformation plan. The difference between the two is specificity about trade offs, honesty about constraints, and funding commitments that match the stated ambition. On all three counts, it falls way short.

The most telling line in the whole document is the £23 billion GDP uplift figure. That's the number Scotland could capture with the right investment and leadership. The strategy itself doesn't provide either at the level required to get there.

I should not let this stuff wind me up, but it does. AI is the opportunity not of a lifetime but of a millennium. Scotland really has the assets to grasp it and transform a fast failing country. The ironic thing is, it is not technology that will fail us, but the ability of human leadership to grasp the opportunity at the right level and execute it.

Pete