Writing in 1958, the great Canadian economist John Kenneth Galbraith coined the phrase “the conventional wisdom” to describe “the ideas which are esteemed at any time for their acceptability.” The phrase immediately leapt to mind upon seeing Philipp Staab, Marc Pirogan and Dominik Pietron’s article on German technological sovereignty (paywalled, sadly). Their article serves as yet another reminder of how thoroughly the preceding conventional wisdom of the past 40 years has been displaced by a new set of ideas. The conventional wisdom in terms of economic policy has gone from emphasizing markets, free cross-border flows of everything and limited state economic regulation (a phrase which, granted, means something different in Germany than in the United States) to a focus on state economic intervention and the promotion of state sovereignty, in both its digital and economic flavours. For long-term economic-policy watchers, it’s been quite the ride.

Might we also be witnessing a change in the conventional wisdom surrounding generative AI, and the data-driven economy generally? A few pieces this month suggest that bloom may be coming off the generative AI rose, at least in some circles. In Rob Lucas’ article, Unlearning Machines, he turns his eye to the dark social side of machine learning. The article also doubles as a book review of Matteo Paquinelli’s The Eye of the Master: A Social History of Artificial Intelligence, which sounds like a fascinating read. Meanwhile, self-described artists/programmers Ting-Chun Liu and Leon-Etienne go deep into the large language models, highlighting, among other things, how these models often draw on “the same few datasets, models, and algorithms,” biases introduced by image labelling (“text-image pairs”), “questions about the models’ tendencies to default to learned patterns,” and the biases that come with coding images. There’s a lot here; check it out.

Then, focusing on the material aspect of the data-driven economy, tech podcaster Paris Marx interviews Sébastien Lehuedé, a lecturer in ethics, AI and society at King’s College London about “how to stop a data centre.” Not the kind of thing you talk about unless you’re very concerned about how the data-driven economy is developing.

It’s fascinating to watch one form of the conventional wisdom supplant another. But it’s a whole other thrill to be present the moment a new conventional wisdom comes into being. Those of us at the 2018 Internet Governance Forum (IGF) annual meeting in Paris witnessed just that, when French President Emmanuel Macron, in his keynote address, highlighted three models for global internet governance. His typology – Chinese authoritarianism, the US free market model, and EU human rights-focused governance – has become the dominant way of thinking about internet governance.

(As speeches go, it was an all-timer. Most speakers at events like this pander to the assembled great and good. Not that day. Instead, Macron deployed his Obama-level oratorial skills to promote the inevitability of state internet regulation, to an auditorium filled with people holding very strong opinions about the inadvisability of such regulation. This was politics on the highest difficulty setting. The moment he finished, a colleague turned to me and said, “We have to talk about that speech.” We did, and the speech inspired a co-edited volume, by Natasha Tusikov, Jan Aart Scholte and yours truly, on the state’s role in internet governance.)

We weren’t the only ones paying attention. Anu Bradford, promoting her latest book, Digital Empires: The Global Battle to Regulate Technology, on the TrustTalk podcast, unpacks these models and their consequences in a podcast. Though, as a citizen of a smaller country, I’m often left wondering about how other regions are regulating the digital economy, along the lines of Universidad de los Andes professor Jean-Marie Chenou’s work on varieties of digital capitalism (also paywalled – sorry). Maybe there’s more than three ways to regulate big tech? Time to revise Macron’s conventional wisdom?

I’ve long thought (as have many others) that the problems with recommendation algorithms stem primarily from companies’ business models and the incentives they face as for-profit corporations. So it was fascinating to read the Knight First Amendment Institute’s account of how the non-profit BBC tackles recommender algorithms. They face all the same technical problems with these systems. But, as you might guess, it matters what you’re optimizing for, even if one’s choice of goals is always open to debate and difficult to realize.

As always, we have several other fascinating articles and videos for you, including Melanie Brusseler and Matthew Lawrence on the political economy of the energy transition, a discussion with Jostein Hague on his book, The Future of the Factory: How Megatrends are Changing Industrialization, and a book-length examination of Decidim, a collective platform. Will any of these books and thinkers spark a rethink of the conventional wisdom in their subject areas? Stranger things have happened, and it’s only February.

-  Blayne Haggart