Thoughts in Between

by Matt Clifford

TiB 147: Avoiding an AI race to the bottom; the never-ending Brexit; Fund People, Not Projects; and more...

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

How to avoid an AI race to the bottom

The Thoughts in Between podcast is live and I’m delighted that the first episode is a conversation with Ian Hogarth on artificial intelligence. Ian is an entrepreneur (co-founder of Songkick), angel investor and a deep thinker on AI. He’s well known as the co-creator of the annual State of AI report (see TiB coverage of the 2018, 2019 and 2020 editions) and the author of AI Nationalism - a highly influential essay on the geopolitics of AI.

In the conversation we talk about what it would take to avoid a dangerous “race to the bottom” dynamic in nation state competition to develop artificial intelligence. Fortunately, Ian is a boundless optimist. I find his ideas about how we could reimagine AI as a global public good (and his analogy to Wikipedia) particularly inspiring.

We cover a lot of other topics, including:

I hope you enjoy the conversation - feedback very welcome. If you do, please leave a review on Apple Podcasts or your favourite platform; it’s one of the best ways to help us grow. 

Fund people, not projects: science edition

A recurring TiB theme is how to improve the productivity of science (see, e.g., TiB 92TiB 102, TiB 113), so I’m delighted that super blogger Jose Luis Ricon has started a series on the topic, “Fund People Not Projects” (“FPNP”). The first looks at some programmes that do fund individual scientists and the second at the accuracy of pre-grant peer review (i.e. can we identify good research before it happens?). Both are long but worth your time.

The first post looks at the apparent extraordinary success of the Howard Hughes Medical Institute in funding scientists on the basis on talent rather than a proposed project. One evaluation suggests that researchers funded in this way are almost twice as likely to produce “breakthrough work”. The optimistic hypothesis is that this is because such grants are long and encourage risky, fundamental work, rather than incrementalism (see TiB 103). José examines this in detail and concludes, alas, that it’s not quite as rosy as all that (at least not in terms of HHMI’s causal impact).

I do have some intellectual skin in the game here in that my entire career is built on the FPNP thesis. I think one of the most important arguments in its favour is something José mentions in passing:

Funders don't encounter a pool of projects passively waiting to be funded; rather projects are devised by researchers depending on what they expect will get funded

The same applies to startups. One reason I think funding people pre-company is so important is that the kinds of startups that get built are at least in part a function of founder incentives: if you give people who might not otherwise start a company time and space to do so, you (should) get companies that otherwise wouldn’t exist at all. 

Oh! What a lovely (culture) war

Brexit got done, just about. But it’s not going away, and for reasons that are interesting beyond the immediate issues at stake. We’ve talked before (e.g. TiB 34) about multi-dimensional politics, with the “open/closed” culture dimension supplementing the classic left/right economic one. Crudely, Brexit worked as a bundle of policies: the novel electoral coalition of “anti-elite” voters who voted for it were willing to stomach some economic pain for a cultural victory. But that's also precisely why Boris Johnson's deal is unlikely to be a stable equilibrium.

The deal we have is about the hardest possible Brexit, barring “No Deal”. This creates an unusual situation. Every future UK government now has some enticing low-hanging economic fruit: increase economic integration with the EU! That doesn’t mean the UK will rejoin any time soon. I expect immigration will remain a highly salient issue that prevents the UK going even as far as EEC membership. But it’s difficult to imagine that UK governments can persistently convince the electorate that technical issues like regulatory divergence are important cultural issues when unbundled from immigration.

We only got to such a hard Brexit because of the weird sort of “iterated Hastert rule” that means tiny groups of people can exert enormous control over political parties (See this superb podcast with David Shor for examples in the US Democratic Party). But at some point this dynamic crashes up against the median voter theorem - and the median voter really likes getting richer! As such, I think we can expect the UK’s economic relationship with the EU to be a major issue - and a winning one for ex-Remainer / integrationists - for quite some time. 

Quick links

  1. Forever young. At what age do you hear the songs that have the most lasting impact on you?
  2. Depressing facts. How much does mental illness cost an individual?
  3. Minimalist computation. Lovely thread on hacking everything from Excel to Magic: the Gathering
  4. But how can we be sure? Apparently the famous Dunning-Kruger effect is not real
  5. What's new in innovation studies? A whole rabbit warren of papers on entrepreneurship and innovation from the American Economic Association annual meeting

You know the drill...

Thanks for reading Thoughts in Between. If you enjoy it, please do share it on Twitter / Facebook / LinkedIn - or just forward this email to a friend who might like it too.

You can always email me or find me on Twitter if you have any feedback or questions.

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.