What AI Global Leadership should the UK offer?
UK academic research into AI may is but our efforts are dwarved by those of the US and China. More-over Government thinking is akin to that the Alvey Programme which fed the R&D programmes of the US and Japan and helped hasten the demise of the UK computer industry, with software following hardware.
We need to focus on where the UK has a unique leadership position. Brexit has strengthened the position of London as a Centre for global arbirtrage as envisioned by the late Lord Renwick when he opened a debate in the House of Lords in 1997 on the role of the City of London in the world of global electronic commerce.
Vint Cerf picked up on Lord Renwick’s vision when chose “Wrestling with Alligators: Delivering the Digital Promise” – as his theme for the first Renwick Memorial lecture.
Wendy Hall homed in on the UK’s potential leadership role when she chose “From AI to Eternity” as her theme for the second Renwick Memorial Lecture. She reprised the analysis that she and Kieran O’Hara and Wendy Hall published in 2018 as Four Internets: The Geopolitics of Digital Governance.
In 2021 they went into more detail with an introduction by Vint Cerf in Four Internets – Oxford University Press .
There are four visions of the Internet fighting for supremacy. None of them will “win”.
London is unique as the only place you can do business under whichever law, in whichever language you wish.
That puts the UK in a unique leadership position with regard to hosting research into arbitrage between competing visions – including the inter-operability and audit of ever more complex technologies and systems, including, but not just, those which use machine learning – that subset of “AI” which is currently in the headlines.
The Prime Minister’s agreement with the US President to host a summit on AI safety in London is a great move. But it will be a waste of effort unless followed up by plans to host the physical (not just on-line) meetings and secretariats that will be needed to build on what, if anything, is agreed. But the “real” opportunity will be for the UK to host and lead the processes to handle co-operation. cross-fertilisation and arbitrage between what is NOT agreed. That is where the UK, thanks to Brexit, has unique opportunity, including to mix and match the best of thinking from across the divides without the need for industrial espionage.
The Four Internets
O’Hara and Hall identified four visions of the Internet and its regulation.
- The original Silicon Valley vision of an “Open Internet”
- The Brussels Bourgeois Internet: governed by bureaucratic regulation, e.g. GDPR
- The Beijing Paternal Internet: using the powers of surveillance for social good
- The Corporately owned Internet: based on US Supreme Court Decisions, e.g. on IPR
They also refered to the Moscow “spoiler model” and to India as a Swing State.
The UK debate over the On Line Safety Bill can be seen as a struggle between the Beijing Model (adopted by the non-OECD states with half the world’s Internet users) and the other three.
When it comes to AI, the current UK proposals appear to reject the Brussels approach (which would lead to a new over-arching AI regulator), in favour of a compromise between the original “Open internet” approach (light, or no, touch) and the evolving US Commercial approach, with existing sectoral regulators extending their reach to cover the use of AI in the products and services they regulate. The Prime Minister, of course, now under pressure to adopt the Brussels approach, only more so. This has echoes of the Whitehall tradition of gold-plating EU initiatives – which was why so many officials still cannot accept the reality of Brexit. Hopefully reality will intrude now that the Labour Party has accepted that there is no going back. We have to find a new way forward.
Meanwhile Beijing has been outspending the rest of the world in AI R&D and appears to dominate the market for all save English language-based AI applications. And most of the world does NOT have English as its first language.
The UK Opportunity
In his recent landmark speech to the Chinese Communist Party President XI referred to using Hong Kong as a sandbox for innovation and to the use of Common Law for international trade and commercial issues because it was too difficult to get agreement between regulators.
If the UK really does want to take a global lead on AI regulation, as opposed to being piggy-in-the-middle between the EU, US and China, it needs to focus on becoming an attractive location for thought leadership and arbitrage across the competing visions, including using the City of London to host the processes for holding to account those whose actions have caused cross-border harm – whether or not they followed the regulatory rules of their own jurisdiction.
Such as approach is also essential to maintaining the position of London as the world largest Fintech/Cybertech hub outside North America, given that our public R&D support budgets are piddling compared to those of others.
We should focus on hospitality for standards groups looking at inter-operability across and between conflicting visions. Our contribution should be to host the secretariats and information services they need, including keeping track of who audited which version of which module to what standards – mundane, boring and critical. Our research focus should be on how to handle the audit of inter-operability and the processes for resolving disputes.
That should also enable us to avoid taking sides unnecessarily, because the prosperity of London depends maintaining its trusted position as part of the financial services infrastructure of China, trusted by their ruling elite as much as it is by those of the Middle East.
Recognising Intranational as well as International differences
It is also essential to recognising the difference between the agendas of the City of London (with its ongoing global vision) and the City of Westminster (pre-occupied with preserving the “special relationship” with Washington). It should not be either/or, we need both. We need also to recognise that relationships between New York and Washington and between Shanghai and Beijing are equally nuanced.
The difference is that Westminster and the City are within walking distance. It is therefore easier to reach agreement without an e-mail trail – save for those around agreements as profound and meaningless as the Treaty between the Empire and Anacreon which begins the evolution of Isaac Asimov’s Foundation series, or those around the GDPR or the On-line Safety Bill .
Asimov was writing before the term AI was invented to justify the budget for a research programme into Algorithmic programming. Today it would require a massive AI system (or an army of corporate lawyers) to generate the necessary “symbolic analysis” to demonstrate that neither GDPR nor the On-line Safety Bill applies to those whose business models provide the greatest threat to either personal privacy or on-line safety.
I remember Donald Michie discussing the value of modelling Acts of Parliament with Ian Lloyd in the margins of the 1982 Seminar for the UK Technical Press published as Intelligent Systems the Great Opportunity for which I prepared the paper on “Training for multi-career lives” later published as Learning for Change. Over the decades since the idea has been discussed several times.
Perhaps its time has come. If so, I would expect an Anglo-Chinese law firm to lead the way with an exercise to model the inter-play of global regulatory regimes in support of the use of common law, as proposed by President Xi, across all jurisdictions which recognise the supremacy of neither the European Court nor the US Supreme Court.
I would prefer the exercise to be based in London, in co-operation with Hong Kong and Singapore, rather than for it to be led by either of them but that requires identifying those interested in working together to find practical ways forward. as Convenor of the Advisory Group for the London Cyber Resilience Centre, We are hosted by the new City University Multidisciplinary Centre for Cybersecurity and Society. This cuts across Engineering (including cyber), Law, Media, Policy and Bayes Business School. I am therefore exploring the idea of using City University to host industry/academia clubs, akin to those in Cambridge and Palo Alto, to work across boundaries to turn ideas into action with regard the issues of interest to the greatest multi-lingual, multi-cultural financial services, fintech and cybertech hub outside North America.. It also helps that the main campus is between Farringdon Station (where the Elizabeth Line meets Thameslink) and Kings Cross-St Pancras (for Eurostar) – easy to get to from all directions.