Was The UK AI Summit Too Focused On Big tech?

A few weeks ago, discussing the rise of the dollar to financial pre-eminence after Bretton Woods, I mentioned Lord Halifax’ comment ‘they have all the money but we have all the brains’, regarding the rise of American power relative to the declining role of Britain in the post-war period. I normally try not to re-use quotes but this one has stuck with me, and I will at least try to embellish it.

Last week the UK government held its AI Safety Summit, an event that initially built high expectations for the UK’s role as the locus of AI regulation, but that now looks to be bypassed by an avalanche of international regulatory activity, that itself has been left stranded by a wave of investment in AI.

US AI

There is also a sense that the UK’s thunder has been stolen by the US. Notably on Tuesday, the Biden administration issued a directive to US AI model builders that they share AI development and safety test processes with the government. In addition, the UK and US will open AI Safety Institutes.

Whilst the small Bletchley Park invitee list aimed to balance think tanks and universities with select governments (China was included) and AI firms, the overall sense was that the contest between government and AI firms has commenced. It may initially be an uneven one. The AI firms have lots of money and appear to attract lots of brainy recruits (Mistral AI is a recent, noisy example). Governments, it seems, are in short supply of these assets, but they do have laws.

For readers interested in the AI Summit, there is a very useful write up of the main points discussed in the conference sessions. It was an important event given that every time we open our phones, shop online or increasingly, use public and private transport, we are in the grip of AI.

However, granted even that the event was hastily organised, its focus was likely misplaced – it concentrated on the extreme risks from what I might call ‘deathstar’ type AI, that is AI engines that are massively resourced in programming and computer power and fed reams of data (We have in the past referenced these models (The LevALlling)).

Bioweapons

One reason for governments to focus on ‘deadly AI’ is that they have already peered into the pandora’s box of AI in the realms of bio weapons, battlefield implications (Microsoft and Palantir are amongst the most important actors in the war on Ukraine) and security, and they are worried.

There was no better proof of this than China’s representative at the Safety summit voicing concern for the dangers of AI, as his government exercises pernicious social control, by AI, over its citizens, in a way that is clearly prohibited by the EU AI Act.

In sticking close to ‘deadly AI’, governments have at least two prior reference points. One is the defense and aerospace industry where ambitious state targets, large budgets, poles of excellence (Bell Labs, Russian aerospace labs) and the urgency of hot wars and cold wars (the launch of Sputnik 1 was a huge catalyst to American innovation) managed to drive technological advances, but where relationships between defense companies and governments have become too close to be efficient.

Another example is finance, which in its own way became weaponised in the 2000’s (recall Warren Buffet’s comment on financial weapons of mass destruction). For some time, and it may still be true, regulators did not have the resources to properly oversee financial institutions and these institutions became so large and powerful that they effectively captured the political economy. In a world where the large AI firms have promised to share the workings of their models, they have a huge resource advantage over states.

What was most revealing of the thinking behind the UK Summit, was the strange chumminess of Rishi Sunak’s interview of Elon Musk which suggested a prime minister supine to a tech ‘overlord’. There was very little critical thinking in the interview. It would have been better in my view if Sunak had hosted younger, less well-known British entrepreneurs who are applying AI in fields like medicine and education.

The Sunak government (and its predecessors) are a lesson in how to create a society that can be further undermined by AI – in particular the education and health system are not set up to use AI to best effect for students and patients, and the idea (expressed during the Sunak/Musk interview) that the future is one where our friends will be robots is a very bleak one.

If Lord Halifax was in power, he might look at the AI giants and quip ‘they have all the money, and they have all the brains’.