Thoughts in Between

by Matt Clifford

TiB 124: How to create tech optimism; the future of AI; ideological warfare via TikTok; and more...

Welcome new readers! Forwarded this email? Subscribe here. Enjoy this email? Forward it to a friend.

GPT-3: a remarkable glimpse of the future of AI

We’ve talked several times about GPT, OpenAI’s vast language model - most recently just a few weeks ago when they announced GPT-3, the latest version. This week they made some of its capabilities available in beta and it set tech Twitter alight. Here’s a good, short and non-technical introduction. Some uses cases are truly impressive. Try this example of GPT-3 explaining a dense and complex idea. Or this one of it converting plain English into a prototype design for an app. Or here, where it answers questions about rockets in the style of Elon Musk or philosophy in the style of Aristotle (It has limitations too, of course - as do we in interpreting it)

GPT-3 feels like an important moment in the history of technology. Arguably the most magical thing about technological progress is what’s sometimes called “combinatorial innovation”: once a new way of doing things exists, it can be repurposed and combined with other “building blocks” to create new products and services. Today’s GPT-3 use cases may seem like mere toys, but now these capabilities are available they can and will be used in ways that we cannot imagine today. 

Anders Sandberg has a good thread on this (the replies are worth reading too), in which he compares the state of AI today to the Homebrew Computer Club of the 1970s. GPT-3 may end up being one of the foundation stones of a "comprehensive AI service ecosystem”. We’ve talked before about the idea that the future of AI may be one of networks of services rather than super intelligent agents. If that’s right, we’ll likely look back on the release of GPT-3 as a primitive but seminal moment. 

TikTok, China and ideological conflict

TikTok, the wildly popular social video app owned by Chinese company ByteDance, has been banned in India - and faces possible bans in the US, UK and elsewhere. It’s the latest in a series of incidents that some analysts have called a “tech Cold War”. How much should we worry about TikTok? As so often, Ben Thompson has the best analysis; he asks why any country would allow an opaque algorithm controlled by an avowed ideological opponent to determine what tens of millions of their citizens see.

If that framing seems overblown, it’s worth reading an extraordinary speech by John Garnaut, one of the best networked Western analysts of China, reprinted in Bill Bishop’s superb Sinocism newsletter last year. Garnaut argues that we tend to systematically underplay the role of ideology in explaining the actions of China. He sees an essential continuity between Stalin, Mao and Xi in seeing writers and artists as “engineers of the soul” in the service of the Party.

Does this apply to consumer tech companies too? The answer appears to be yes - much more than you might expect. Fergus Ryan has a good short analysis of the political background of Bytedance. As he notes, it’s deeply entwined with the CCP and even has a party committee as part of its governance structure. It would be comforting to believe that the West in engaged in purely economic competition with China. The truth, alas, appears to be much darker and more dangerous. 

Apollo, technological optimism and Moore's Law

Longtime TiB readers will recall that back in January, along with my friend Arnaud, I started a reading group on the topic of “Organising Genius”. The pandemic forced a hiatus, but I’m delighted that we’re now restarting remotely; the next session is on Silicon Valley and MIT. You can learn more and sign up here. This week Ryan Khurana has an excellent essay on the Apollo Missions, the focus of our last pre-lockdown session. It’s based on Charles Fishman’s One Giant Leap, which I also recommend. 

It strikes me that Apollo has two important lessons for governments thinking about how to invest in technology today, (as for example the UK Government’s plans for ARPA - see previous coverage). The first is that, if successful, the cultural impact of a successful “moonshot” will far outweigh the short-run economic impact. As Khurana notes, Apollo shifted the cultural association of the idea of “technology” from the violence of the B-29 bomber to the optimism of the moon landing, as well as attracting countless students into science. 

Second, Apollo demonstrated the ability of a government to drag technology from the future into the present through visionary procurement. NASA’s demand created the market for integrated circuits - and ensured that the price fell by 90% in five years. Apollo effectively kicked off Moore’s Law before Moore even coined the term. On both counts, something analogous is possible - and desirable - today.  

Quick links

  1. Getting on a bit. What do leading ageing scientists think about ageing?
  2. You spilled a bit. Innovations from VC-backed companies apparently have more positive spillover effects than those from their non-VC-backed counterparks (Correlation != causation, etc)
  3. Unemployable? Corporate recruiters don't like hiring good founders.
  4. Capitalism, alone. Why does household size shrink as countries get richer?
  5. Borrowed time. Interesting thread on the history of usury laws - why did the Catholic Church ban interest in the fourth century?

APOLOGY: One of the most clicked links last week was to Fredrik de Boer's essay on what Twitter has done to journalism. The piece seems to have been taken down, but you can still read the cached version here.

Your feedback

Thanks for reading all the way to the end. If you like this, please help grow the readership: forward it to a friend or - even better - share it on Twitter or LinkedIn.

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.

I’m always happy to hear from readers: feel free to hit reply or message me on Twitter.

Until next week,

Matt Clifford