The people’s AI?

The title of Maximilian Kasy’s new book The Means of Prediction cleverly riffs off Marx’s concept of the means of production for th age of AI. These means, Prof Kasy argues, are data, computational infrastructure, technical skills and energy. The core argument is that these means rest in few hands, so the value being created by AI is concentrated in a relatively small number of technology companies and their founders and employees. What’s more, this concentration also skews the kind of AI systems being created: AI is designed to optimise some objective function, as I argued in Cogs and Monsters (and some recent as-yet-unpublished lectures).

This makes a powerful case for democratising AI, the book argues. Unless countervailing power from workers, consumers, and politicians is brought to bear, the technology will create further inequality and will not serve the public good, only private, profit-maximising interest. Kasy convincingly argues for collective means of control, rather than just individual protections such as GDPR: “Machine learning is never about individual data points, it is always about patterns across individuals.” Control over one’s own personal data does not protect privacy as long as similar people share their data.

The book starts with a section explaining AI, followed by a section explaining how economists understand social welfare – this being the approach (as Cogs explains) that is being automated by AI. These are very clear and useful for people who are hazy about either, although as they are so introductory it did make me wonder about the target audience for the book. Having said that, there is such a lack of knowledge among the public and indeed lots of policymakers and politicians that these sections are probably sorely needed.

The final two sections go on to regulatory challenges and the need for democratising the development and use of AI. As Kasy points out, policy choices have always been choices, with winners and losers, and decisions have often involved predictions; after all this is what policy economists have been doing for decades. The increasing use of AI to make decisions automates and speeds up choices – the danger being that it does so with embedded biases and implicit decisions hidden by the technology, and the all-too-common presumption that the machine must be right.

The optimistic take on where we are is that the use of AI to make predictions and decisions will surface some of the implicit assumptions and biases, and so force more public deliberation about how our societies operate and affect different people. The pessimistic take is of course that they simply become more deeply hidden and entrenched. Depending on my mood, at present I think things could go either way. But that open prospect makes The Means of Prediction a very timely book. And – having pondered who it was aimed at – probably one that every official and politician should be made to read as they chirrup about using AI in public services.

71N92DLDMlL._AC_UY436_QL65_

 

When a nearly 30-year old textbook is still the best on the digital economy

There’s a gap in the market. I’ve started developing a Digital/AI economics syllabus for policy MPhil students who won’t all be economics graduates; and of course there are loads of fantastic papers and resources online.

It’s fairly straightforward to write down a list of topics – recognising that one can’t cover anything.Here are my headings for an 8 week Cambridge term – so there’s nothing on crypto for example, as I don’t know enough about it (& am still pretty sceptical). This is the first half – the second 8-week term will cover competition, regulation and trade.

  1. Basics: Information and the economy, why digital is different from the old economy
  1. Implications for organisations: Business models, platforms, ecosystems; auctions
  1. Intangible assets: IP, copyright, data (& privacy), measurement
  1. Inputs: finance, energy and materials, chips
  1. Innovation: clusters and types, direction, policies
  1. The demand side: technology diffusions, ‘free’ services, household production, open source
  1. Impact on labour: task models, AI & jobs
  1. Policies for the digital/AI economy: skills/jobs, industrial policy, digital public infrastructure, ‘the stack’ & sovereignty,

The structure might well shift as I get into details. But naturally I wondered whether somebody had done much of the work for me and written a textbook covering a lot of this. And it seems the answer is no: nobody has done an update of Carl Shapiro and Hal Varian’s fantastic 1998 Information Rules. I just re-read it and of course the examples have dated but the principles have not. The brilliant Daniel Rock told me (via BlueSky) that he still teaches using it.

I wrote Markets, State and People a few years ago as a policy economics textbook because the ones available were unsatisfactory for the course I was teaching then – none took the perspective of the basic co-ordination problems policy seeks to tackle. So maybe it’s me. But I still think there is a gap in the market for a new Information Rules – somebody could surely do this in time for its 30th anniversary?

61atJCARUwL._AC_UY327_QL65_

 

Conference hinterland

I attended recently an excellent NBER workshop at Stanford on the economics of transformative AI. It’s always interesting to hear what books/cultural references people cite in addition to the usual econ papers, and it was an unusually wide range this time – reflecting both the interdisciplinary crowd and the breadth of the issues. Here they are.

Books:

Deep Thinking by Gary Kasparov

Automation and Utopia by John Danaher

The Myth of Sisyphus by Albert Camus

Philosophical Writings by Charles Sanders Peirce

The Limits of Organization by Kenneth Arrow

Brave New World by Aldous Huxley

Movies:

Perfect Days Wim Wenders

Beauty and the Beast

I *loved* Perfect Days. There’s only one of the books I haven’t read (Danaher) although I don’t claim to have got to grips at all with Peirce – on his work, I found The Metaphysical Club very informative. But participants also cited papers from computer science and sociology that I feel the need to read. Sign of a good workshop – a lot of brain space opened up.

Great power transitions and the role of technology

I have mixed feelings about Jeffrey Ding’s Technology and the Rise of Great Powers. On the one hand, it’s an interesting and persuasive hypothesis. He argues that great power transitions such as from the UK to the US around the turn of the 20th century are driven not by the new great power commanding the technological frontier but rather by the extent to which new general purpose technologies have diffused across the whole economy. He contrasts this with the idea – apparently dominant in political science – that it is the country with control of the leading sectors that predominates.

Thus for example it was the US, not Germany, which overtook Britain because Germany commanded the chemistry frontier but the US educated chemical engineers en masse and the new chemical-enabled manufacturing techniques permeated widely. The book looks at the first Industrial Revolution – the UK’s emergence as dominant after the earlier reign of the Netherlands – as well as this 2nd industrial revolution; and also at the failure of Japan to take over from the US in the late 20th century. The book focuses on the importance of developing skills institutions that enable widespread uptake, citing for example the Mechanics’ Institutes in 19th century Britain and the spread of engineering through universities in 20th century America.

This analysis is backed up by detailed case studies – very interesting – as well as empirical work. While economists will characterise transitions involving new general purpose technologies as involving both a period of leading sector change and then diffusion across the economy, it seems very plausible to me that geopolitical transitions depend on the latter. Military and strategic strength depend on robust engineering and production capabilities; leading edge R&D is necessary but not sufficient.

The ‘other hand’ is the writing style. The text is rather repetitive and written in academic-speak. I guess the book is based on the author’s PhD dissertation, but it would have benefited from a rewrite in order not to read like a series of academic journal articles. This is a bit of a shame, as of course the argument is relevant to the relative roles of the US and China now. The book was written before the US started shooting itself in all the feet it could find in terms of sustained technical and economic progress; but in any case the author recommends the US switch its focus to developing the broad skill base needed to enable AI use across the economy if it’s serious about winning the geopolitical contest.

71J5UCb7uWL._AC_UY327_QL65_

 

The end of progress?

Carl Frey’s How Progress Ends: Technology, Innovation and the Fate of Nations is the kind of book that is exactly my cup of tea. I did have the opportunity to read it before publication and part of my blurb comment was: “How Progress Ends is a fascinating account of the way culture and institutions interact with new technologies.” The major part of the book consists of the history of some major technological advances along with some significant innovations in societal models (the American Revolution, Soviet central plannng) exploring exactly this interaction.

The thing that makes institutions and culture matter is that general purpose technologies – printing, steam, electricity, telecommunications – is their disruptive character. The affordances of the technologies enable challenges to the established economic or political order. Sometimes the incumbents can resist successfully – as in China’s ‘reversal of fortune’ following the formation of the Qing dynasty, or in the Soviet elite’s resistance to reform until it was too late. Sometimes the character of technology means political competition enables it to advance faster than if there were political centralisation – and sometimes the other way round.

States can therefore play a decisive role in whether their societies experience and (eventually) benefit from technological progress. The book ends with some reflections about the present. Frey is pessimistic about both the US and China (a bit of an echo of Dan Wang here). In the US he sees the incumbent AI companies and their relationship with the government as freezing out innovation: “Reaping the benefits of technological change requires institutional support to make space for exploration.” In China he sees future innovation as falling victim to cronyism and the assertion of control by the central government. “The decline of either China or the United States is by no means inevitable,” he writes, although one senses he thinks it is.

Who knows. What does seem clear is that the path taken by technology cannot be divorced from the politics, which is highly uncertain everywhere. The historical lessons are well worth pondering. How Progress Ends is well worth reading alongside for example Carlotta Perez (Technological Revolutions and Financial Capital) and Bill Janeway (Doing Capitalism in the Innovation Economy) to reflect on the current moment.

81rAY0c8XAL._AC_UY327_QL65_