Matt's Thoughts In Between - Issue #19

AI nationalism - the next big idea in geopolitics

I've discussed Ian Hogarth's idea of AI nationalism before in Thoughts in Between and this week Ian published a superb essay that lays out his argument. It's a must read.

The core thesis is that artificial intelligence is going to open up a new dimension of geopolitical competition, in which a nation state's ability to build and deploy AI will be key to its ability to compete and thrive. There are a number of important parts to this. I'm particularly interested in the points that Ian makes about the strategic importance of redistribution (a favourite TiB topic), and the idea of "AI client states" - the idea that countries without their own strategic AI capability will have to align themselves to one of the emerging AI superpowers. Ian points to Zimbabwe's deal with Chinese AI company Cloudwalk, which feels like the thin end of the wedge.

My only point of (mild) departure from Ian's arguments come in the section on British politics. Ian - rightly, I think - points out that nationalisation may become a key tool in the policy arsenal of governments who want to win the AI nationalism race. But he then dismisses the idea that a future Corbyn government might be interested in this space. I'm not so sure. I understand from people close to the Labour leadership that they are genuinely engaged in radical ideas in tech policy. Which raises the interesting possibility that - whether you view the prospect of the the UK nationalising the most important AI company in the world with excitement or horror - influencing Labour's tech agenda may right now, oddly, be one of the most leveraged activities in world politics.

Superintelligence, Chinese edition

One of the other major themes of the essay is, inevitably, the huge role of China. The indispensable source on the topic of Chinese AI remains Jeff Ding, who whose work we've discussed before. His excellent newsletter last week included a translation of an important piece on AI safety and ethics by Zhao Tingyang, a prominent Chinese philosopher. It's worth reading the whole thing.

As far as I can tell Zhao might be termed a Chinese Nick Bostrom, with a dash of Yuval Harari. His essay is an important read, not least as an antidote to the tendency in the West to caricature or polarise Chinese thinking on AI. Zhao covers a number of important topics: what are the threats of superintelligent AI? what are the risk of autonomous weapons? how will advanced AI change the meaning of life?

Interestingly, like Ian Hogarth, Zhao ultimately comes down on the importance of global cooperation and governance if we want to avoid the most dystopic outcomes from advanced artificial intelligence. Zhao calls for "a world constitution and a world political system". This doesn't make me very optimistic; our track record is not good. Ian points to the fascinating Baruch Plan as a role model. But, of course, that plan failed. Will we do any better with AI?

Fraud, damned fraud and unhappy families

One of the big tech stories of the week is that the founder and CEO of Theranos, Elizabeth Holmes, has been charged with fraud and faces jail. If you don't know the Theranos story, the short version is that in 2003 Holmes (then aged 19) founded an apparently revolutionary medical technology company that claimed a breakthrough in non-invasive blood testing. Fifteen years and $700m of investment later, it seems that almost all the claims the company made were false. This tweetstorm lays it out well.

Theranos' spectacular failure has prompted critiques of Silicon Valley and its willingness to fund, well, bullshitters... but Theranos doesn't really fit that pattern. Most of its funding came not from VCs, but from wealthy families, including those of Rupert Murdoch, Carlos Slim and US Education Secretary Betsey DeVos. And its board was full not of prominent West Coast names, but rather the DC establishment: Henry Kissinger, James Mattis, William Perry, George Schultz and others. (If you want a truly bizarre story about Schultz's involvement, read this extraordinary piece on his relationship with his grandson, who worked at Theranos)

What can we take from this? To start with an easy one, governance matters - surprise! - and retired diplomats may not be best placed to provide it to a medtech company. More optimistically, it's impressive how well investigative journalism worked (You can read the book by John Carreyrou, the journalist who broke the story, here). And, in a surprise turn for Rupert Murdoch as the good guy, it's impressive that despite having $100m invested, he refused to kill the WSJ investigation that ultimately killed the company. Hurray for good journalism!

Quick links

Tell no one. What are the secrets we keep the closest?
How to steal a career. Extraordinary thread on a bizarre art fraud.
Everything you think you know is wrong. Stanford Prison experiment edition - another classic psych study bites the dust.
Incentives matter. And so does the power to control the weather (thanks Arnaud for the link), in case you weren't worried enough about Chinese technology
No conflict, no interest. How Elon Musk's Tesla bought Elon Musk's Solar City. Incredible - and troubling - story.

Your feedback

This is the 19th edition of Thoughts in Between - thanks for continuing to read! Next week I'm going to include a short reader survey to try to learn more about how I can make TiB better. In the meantime, if you have feedback or links to share, just hit reply. And if you like this, please forward to a friend and encourage them to subscribe.

Until next week,

Matt