AI – and politicians striving for relevance

Give some small credit to Britain’s embattled PM Rishi Sunak.  At least he is trying to define what being centre-right means in these days.

One of his latest moves is to push Britain as a centre of global excellence for the regulation of artificial intelligence (AI).  Last week he assembled the great and good for the world’s first global AI safety summit. 

In many ways, it went as you might have expected.

First, the participants roundtabled the risks and opportunities and struck a balance of approximately 99% risk and 1% opportunity.

Secondly, Rishi got a little rough handling from the media.  There may have been a little professional jealousy for his stunt interview with Elon Musk and his failure to speak plucky British truth to Elon’s global power.

Thirdly, the US made it even more irrelevant by issuing its own regulation a couple of days before the event (thank you Joe).

The White House executive order may not be great regulation but, unlike the global talkfest, it is helpful in clarifying the fundamental issues with regard to what may be the most consequential technology of our age. Which might just be essential to solving so many of the problems we’re grappling with.  Like say energy efficiency and decarbonisation. 

Using national security powers (worth thinking about that as the game goes on), the order will require “companies developing any foundation model that poses a serious risk to national security, national economic security, or national public health and safety [to] notify the federal government when training the model, and … share the results of all red-team safety tests.”

OK. And then:

“ … The National Institute of Standards and Technology will set the rigorous standards for extensive red-team testing to ensure safety before public release.” 

Hmm.  This does bear some resemblance to the overall Chinese approach, depending on how it’s applied. Of course, we hope and expect the US to apply a somewhat lighter hand.

Because, unlike China, every time the Americans try to slow down ideas and their application, they are likely to apply a different risk assessment to the work migrating elsewhere.  

Still, the potential for the Biden administration to bring things to a juddering halt is on the table. And Joe Biden was one the men who kicked off the encryption wars.

Tech bosses want to be seen going along with the flow, perhaps because they really believe that this time is different and AI could take over the world without us knowing, or perhaps because regulatory intervention now could put one or the other of them in pole position.  So we find Open AI’s Sam Altman preaching for a system which might just consolidate and enforce ChatGPTs early proprietary lead, while Facebook’s Mark Zuckerberg is promoting a freer and more flexible open-source approach which seems to fit better with his current ambitions.

But perhaps it is worth asking just how much we need the bureaucrats deciding whether something is safe in advance.

Brian Williamson – a London-based New Zealand technology consultant – points out that historically the introduction of general purpose technologies like steam and electricity was not regulated by specific legislation, nor made subject to assessments prior to release.  In his words “innovation without permission is an enabler of progress”.

Any such permissions are bound to retard progress because they cannot take account of the uncertainties: of the path of technological progress itself; of the complementary innovations to be generated; or of the users’ responses.

He argues that the benefits are maximised and the risks minimised with regulation on a neutral basis between AI and existing applications and implemented primarily by adaptation of existing mechanisms in response to actual impacts.

Let’s see.  At best, light touch implementation of Joe Biden’s executive order could achieve this desirable state of affairs. At worst, it serves as an entree to a mind-bogglingly costly global regulation on the lines of the Paris climate treaties.

The sceptics among us might expect a more normal global pattern to emerge, of regulated activity migrating to safe havens, and a handful of countries which are unusually open to change – Israel, Kenya, Taiwan for example – serving as the laboratories which provide us with early signs of the benefits which come from allowing risks to be taken and consequences to be borne. New Zealand need not apply.

Meanwhile, Rishi will find that to profit fully the UK will still need to be more attractive as a location than the US, not more strictly regulated. And NZ’s new PM Luxon may want to reflect on the old business principle that the more you focus on minimising the disruption from innovation, the slower the realisation of its benefits.