I spent the start of the week at the imposing Stormont building in Belfast. It has been the centre of power in Northern Ireland for nearly one hundred years, and hosts Northern Ireland’s Assembly, which notoriously now has not sat for over a year, something the vast majority of people in Northern Ireland would like to see rectified Notwithstanding this the local economy is strong (unemployment is 2.7%) and this weekend there will be a US led investment conference in the city, which amongst many other things, demonstrates the economic dividend of falling geopolitical risk.
Unfortunately, Northern Ireland is very much the outlier here, as geopolitical risk is on the rise across the international spectrum. One of the sharpest reminders that we are in a ‘war by other means’ environment is the ongoing ‘Chip Wars’, the latest salvo of which involved the Chinese government banning the use of iPhones by government employees (presumably as a form of leverage over the USA at the same time as Huawei is launching a new flagship phone).
EU AI Act
Despite this I have found the lack of interest by European firms (large and medium sized) in geopolitics to be striking.
In markets, geopolitical risk has rarely been a major factor over the past forty years, owing to the tailwind of globalisation (no wars amongst democracies and emerging markets have been on a path to growth). The events that signalled the cracks in globalisation – Brexit and the Trump trade war – were risks that investors needed to measure. In addition, as inflation has risen, the absence of quantitative easing has made markets more sensitive to geopolitical risk.
In today’s multipolar world where strategic competition is increasingly the norm, geopolitical risks loom large– from the risk that Donald Trump comes close to re-election, to the drive of Chinese technology into Europe (BYD cars have arrived) to the re-alignment of the European defense, aerospace and greentech industries owing to the war in Ukraine. Many of these risks are events that require a response and readjustment, others are more existential.
One important source of risk is the way governments react to landmark technologies such as Artificial Intelligence (AI). The ways in which the use (and abuse of these technologies) is framed and watched over can determine the shape of the industries of the future.
As an early mover on regulation, Europe’s response to AI is the EU AI Act, whose main contribution is to divide the application of AI into four pillars from harmless to very harmful (akin to China’s use of AI for social credit scoring). The fact that Europe has moved first sets the bar for other regulatory initiatives on AI, notably that being pursued by Chuck Schumer in the US and Japan’s G7 level ‘Hiroshima’ paper on AI.
One area of vibrant debate, and some confusion, that has opened up is that way in which companies – software, media, data firms – who do not necessarily ‘generate’ AI content formally but are at the centre of many of its processes, are treated. The extent to which they will be policed by the AI Act is not clear and we have not yet had detail forthcoming from the EU. Indeed, it may well be up to individual companies to try to take the lead within their own industries for setting out the ways in which they will treat and use AI (telecoms, retailers with large datasets for example).
An additional source of confusion lies in the use of AI by start-up firms, especially out of France (ironically), which is striving to become an AI innovation centre (it has a rich labour market here and the backing of key entrepreneurs). Their argument is that the Act might constrain innovation by new firms. One response to this may well be a ‘sandbox’ system where young companies are allowed to innovate in a regulatory sandbox, to a certain level of commercialisation. A related issue that is bubbling up is access to models (and data) for researchers, and whether we might begin to see more EU coordination on this.
We also need more detail on governance. It may well be that there is a top-level AI Office at the EU level, with then individual state AI Offices, or to make matters more complicated some countries may argue to have an AI office for sub-sectors (AI in telecoms, AI in banking for instance). This is not yet resolved.
A sure sign that AI is becoming the terrain of competition between the large nations is that the UK is trying to become the ‘locus’ of AI regulation, in the sense that it is signalling that it wants to be the coordinator of international regulatory activity, and in the same way as Geneva hosts the WHO, WTO, it wants to be the locus of AI standards.
The AI Summit at Bletchley (Nov. 1st and 2nd) will be an important milestone here. If the UK was to physically host an international regulatory body, why not have it in Northern Ireland?