Thoughts in Between

by Matt Clifford

TiB 210: AI comes for the illustrators; Zero COVID and rebellion; the End of History; and more...

Welcome new readers! Thoughts in Between is a newsletter (mainly) about how technology is changing politics, culture and society and how we might do it better.

It goes out every Tuesday to thousands of entrepreneurs, investors, policy makers, politicians and others. It’s free.

Forwarded this email? Subscribe here. Enjoy this email? Forward it to a friend.

AI comes for the illustrators

It's been an extraordinary week for (visible) progress in AI. OpenAI showcased its "DALL-E" model, which takes natural language captions and turns them into images. There are some pretty incredible examples in this thread or try "rabbits attending a college seminar on human anatomy", "a city on mars" or "an elephant tea party on a grass lawn". Meanwhile, Google Research released a paper on their latest large language model work. This tweet highlights some of the most impressive results (The Guido / Python one seems particularly amazing to me).

What does all this mean? Most commentators think these are big achievements and several say they are ahead of what we might have expected a year ago. The scaling hypothesis strikes again! It's probably not going to be a fun decade to be a commercial illustrator, at least. Indeed, there are some good thoughts here on why it looks like white collar work will be easier to automate than blue collar work. I also enjoyed this thread on what we're learning about artificial creativity (in short: "creativity is mostly combinatorial or extrapolative, and comparatively easy. It's precision that's difficult") and this on GPT-3 trying to figure out similes.

Depending on who you believe the latest demo shows either that AGI is near (which it might be - Metaculus's forecast for AGI moved forward eight years in the last two weeks!) and that it's a long way off (also possible). As usual, Gary Marcus is the most strident critic, but see also this thoughtful thread. My own view is that AGI in the next decade is far more likely than almost anyone in the "normal" world believes, though any prediction should have large error bars. In the nearer term, it's further evidence (see TiB 196206) that we're moving towards a world where the ability to run - or at minimum access - large machine learning models is going to be economically important. This thread is excellent on the topic. If progress in AI isn't already occupying a decent share of your brain space, I think it should be.

RELATED: Thread of "big AI art projects"

"We absolutely cannot abide any hunger"

A few weeks ago I linked to this (fascinating) thread that asked, "Why is China still pursuing a zero-COVID policy?" I noted that some of the answers were interesting, but none of them very satisfying. What seemed peculiar then seems borderline disastrous today. The scenes coming out of deeply locked-down Shanghai (try this one) feel like the gravest breakdown of what you might call China's national "message discipline" that I can remember. It’s striking how these videos so perfectly chime with various canonical depictions of technological dystopias (drones fly in to enforce compliancerobots deliver lockdown messages).

The most interesting commentary I've seen on the broader implications is this thread by Naomi Wu. Do read the whole thing, but the crucial tweet is this one:

We absolutely cannot abide any hunger*. You can do a lot to Chinese- you've seen us put up with a lot, but hunger we will not. Whether it's an epigenetic legacy I can't say but the memory of famine is firmly burned into the Chinese collective consciousness

Wu says that every nation "has its buttons you do not push if you don't want people in the street" and China's is hunger. Read that, then take a look at this.

Does this mean that the CCP is in danger of losing its grip? Almost certainly not. But it's a useful reminder that the path from impregnability to vulnerability can be a very short one. I'm reminded of this essay by Tanner Greer, "Fissures in the Facade", that we discussed back in TiB 96. Greer argues that there are few narratives in China that are truly dangerous to the CCP - and certainly not the ones that Western liberals might think or hope - but the one that is goes something like:

The Party is a racket ...The Party exists to make sure their kids have a spot at the front of the line no matter how much more your kids deserve it ...You live a slave so that someone else’s children can get ahead

It's hard not to see some echoes of that idea in the anger out of Shanghai. Most likely things will calm down swiftly, but it is a fissure in the facade, no doubt.

The End of the End of History (again)

It must be intensely irritating being Francis Fukuyama. You write one of the great, genuinely prescient works of geopolitical analysis and yet are forever doomed to be the punchline of jokes made by people who didn't get further than the title... Nevertheless, it's worth thinking about why capital-H History feels so obviously in full swing, after 20 years or so in which things really did seem more settled. Bruno Maçães (see previous coverage) has a good new essay on this, "A War of World Building" in which he attempts to explain the re-emergence of History and point to its new battlegrounds.

The core argument goes something like this: the promise of the decades after the collapse of the Soviet Union was that the common goal of prosperity, facilitated by neutral rules and open networks, would bring about the end of History. Competition between states persisted, but (rhetorically at least) it wasn't zero sum. The dominant narrative was of a journey to abundance, enabled by the new economy of zero marginal costs and ideas shared at internet-scale. What changed? Macaes argues that the (re)fracturing of the global order isn't some reassertion of the baser elements of human nature; it's a technological moment with two distinct prongs.

The first is what I've called the stubborn persistence of the physical. From the pandemic to Ukraine, it's become increasingly obvious that much of what is important cannot be virtualised or pulled into the world of zero marginal cost. Control of the physical therefore becomes (again) the top priority of geopolitics. The second is that increasingly the "rules of the game", far from being neutral, are embodied in specific communications networks that can be weaponised: "states can now fight one another not by winning in a direct battle but by setting the rules under which other states must operate" - see, for example, SWIFT (TiB 50 and 204), Huawei (TiB 72), the semiconductor supply chain (TiB 161), etc. It's History, but not quite as we know it.

RELATED: Adam Tooze on what Ukraine tells us about the End of History

Quick links

  1. An Englishman's home is his... meme stock? Striking data on house prices.
  2. The world's highest ROI investment? The extraordinarily low cost of vaccine prototypes. Huge, if true, as the saying goes.
  3. Men and women are different... in that they are differently psychologically affected by luck in tennis. A puzzling finding.
  4. Silicon Crimes. Adventures in semiconductor espionage.
  5. When not to have a heart attack. Fascinating thread on a study of doctor behaviour, patient mortality and machine learning.

If you're still here...

... thanks for reading. Thoughts in Between is a labour of love, but a full refund is available on request.

Please do share with a friend if you think they might enjoy it.

Until next week,

Matt Clifford

PS: Lots of newsletters get stuck in Gmail’s Promotions tab. If you find it in there, please help train the algorithm by dragging it to Primary. It makes a big difference.