26 July 2023

The AI delusion: Britain can’t wish itself into a ‘global leadership’ role


In a former life I once found myself managing an AI / machine learning (ML) team for a big American web firm. I took my colleagues round our internal customers asking whether they wanted AI/ML. ‘Oh, yes,’ they replied, ‘so badly!’. The minority wanted it because they thought if they could lift X output by Y% it would deliver Z extra profit. The vast majority, however, just hated their jobs and wanted us to magic them away.

That experience came to mind reading the recent report from the Tony Blair Institute: AI Promises A World-Leading Future of Britain – just the sort of title a chatbot might have come up. It too has the distinct whiff of magicking away the long-running and complex problems of the British state by simply summoning the AI genie. 

And if the technology isn’t up to the task just yet, then just double down! After all (to quote the report):

Machines with the ability to outperform humans will have capabilities that cannot yet be imagined.


Relatively simple algorithms, powered by enormous quantities of compute power and data, are producing models that can already surpass human thought across a range of cognitive tasks.


…our current revolution is the first in history to automate cognition itself.

This all sounds absolutely marvellous, but hold on a moment. Before waxing lyrical about the supernatural powers of AI, perhaps take a moment to read AI security boffin Nicolas Carlini’s ‘predict what ChatGPT can actually do‘ pop quiz.

Beyond the questions about what the tech can and can’t do is a broader issue about national policymaking. Every country is informed by its priors. Indeed, Finland’s ‘National Readiness’ courses which the TBI want us to take up are informed by the two Russo-Finnish wars between 1939 and 1944 – as such, they stem from existential fear of invasion by their neighbour.

Estonia’s digital state is designed to enable the state and nation to float off the territory, taking national records, patrimony, citizenship, government, elections and all into exile and embraced by the citizenship on that basis. Israel’s Silicon Wadi is based on their military-industrial surveillance complex.

British priors remain as follows: Britain is a great nation, Britain deserves to lead, Britain must lead, Britain is nothing if not leading.

Rather than reflect on those priors, the TBI report is written backwards from one of its crossheads – ‘A Golden Opportunity for Global Leadership’. Can Britain lead? Does it have the capability? These are unquestions: not just unasked, but seemingly unaskable.

So we have the American research organisation Darpa, blindly stolen to become Aria, despite the fact Darpa’s key AI work was done in the 1980s and early 1990s as part of the Strategic Defence Initiative. It was the product of lavish funding, autonomy, a long-term outlook and benign neglect by the political class. In Britain, meanwhile, we will prosper with direct prime ministerial oversight of research capabilities. Really? The older I get, the more I appreciate Mrs Thatcher: her fantastic European Single Market, her hatred of governments picking winners…

Of course, AI will be transformative in many small ways, and perhaps large ways – not always for the good. The US and China have first mover advantages. The correct, humble response is to develop a ‘second-mover strategy’. Yet for the authors of the TBI report, AI is not merely technology but a ‘golden opportunity for global leadership’. Winners, therefore, must be picked.

A UK powered by AI can be one in which every child has an AI mentor catering to their individual learning pace while crafting a curriculum that aids comprehension; in which automated language translators ensure the preservation of Welsh; in which advances in AI-assisted drug discovery lead to cost-effective health care; and in which faster grid electrification produces renewable energy for every household.

When presented with such a golden basket of transformation, the wise man asks ‘how?’ and the cynic ‘will it have surfball?’.

In TBI-land, the UK just needs one more turn of centralisation, one more lick of reform for its institutions, one more push for academic excellence (but not from existing academics, silly) and if we can just ‘do AI’ to the institutions of the state hard enough, all will turn out well. Number 10 is the state institution most mentioned, a sign of the dirigisme that underpins the whole report:

This is a technology with a level of impact akin to the internal combustion engine, electricity and the internet, so incrementalism will not be enough.

Again, this sounds terrifically insightful, except that we actually did adapt to the combustion engine, electricity and the internet incrementally.

If you build capability, patiently, strategically, then opportunity, unexpected and glorious, will burst from it – as AI itself did. Saying we will be world leaders is easy, saying we must be world leaders is just daft. Getting there is hard. This report would benefit from being stripped of declamation and bombast back to the smallest, easiest programme of capacity development, all the unsexy stuff.

Is there, perhaps, a hint of regret from Mr Blair, the dominant British political figure of the commercial internet era, that he didn’t secure global leadership for Britain then? A regret that he didn’t send an email or use a mobile phone until out of office?

You might say so, I couldn’t possibly comment…

Click here to subscribe to our daily briefing – the best pieces from CapX and across the web.

CapX depends on the generosity of its readers. If you value what we do, please consider making a donation.

Gordon Guthrie is a Research Fellow at the Scottish Government under the First Minister's Digital Fellowship Programme, studying how the state creates and runs digital services. You can follow his work at https://digitalpolicy.substack.com/

Columns are the author's own opinion and do not necessarily reflect the views of CapX.