CROSSPOST: Dan Davies: Meet the New Economy, Same as the Old Economy
Dan’s subhead: “moats and their discontents”. A drum I have been beating for a while: the AI arms race is likely to fatten consumer surplus, not boost national-accounts measurements given the roles...
Dan’s subhead: “moats and their discontents”. A drum I have been beating for a while: the AI arms race is likely to fatten consumer surplus, not boost national-accounts measurements given the roles of platform monopolies, MAMLMs as moat-maintenance, and brain‑hacking ads; thus the most likely scenario is huge capex, tiny measured real GDP gians, and the non‑appearance of any large AI deployment-driven productivity boom.
It’s very nice to see that Dan Davies agrees with me: the overwhelming economic-welfare impact from the coming of MAMLMs—Modern Advanced Machine-Learning Models—is not going to increase measured real GDP, but rather to increase unmeasured user surplus. The main things that providers of “AI”-enabled information and communication services are buying are twofold:
A measure of protection of their existing flow of platform-monopoly profits;
The attention of users that then can be sold to advertisers—but advertising is not itself a final good or service, but rather an intermediate product.
Any effect on measured real GDP from advertising comes from a better fit between the commodities those advertised to purchase and their requirements and desires (a real thing! but hard to measure!). And, in terms of true (as opposed to mesured) GDP-concept output, even that is offset by the brain-hacking: getting people to buy things that or spend their time in ways that they regret later.
But to the extent that our MAMLM ‘bot information butlers can become our proper servants rather than are cruel technoserfdom masters, it has the potential to be marvelous!:
Let me turn it over to Dan:
CROSSPOST: Dan Davies: Meet the New Economy, Same as the Old Economy
moats and their discontents
Dan Davies
<https://backofmind.substack.com/p/meet-the-new-economy-same-as-the> <http://backofmind.substack.com>
I was worrying about this on social media during the week, and thought I might get some interesting discussion … first, have a look at this story about Google’s plans to migrate its famous search product to an AI-driven “intelligent search box”.
Lots of people are very much “gahh, enshittification, I just want my weblinks”, but I’m not - I do think this is potentially very useful. Being able to carry out a semantic search - something like “find me three examples of trade associations making outlandish claims about the economic impact of housing regulation” is a real time saver, and although you do have to do a lot of back-and-forth with the AI to get what you want, it’s less of a pain in the arse than filling your screen with browser tabs and reading the press releases yourself.
However, the question I want to ask is - if we put that kind of issue to one side and assume for the sake of argument that a really useful product will come out of this, how would we expect to see that reflected in the economic statistics?
I’m not sure that we would at all. I like using the intelligent search box, but that’s just unmeasured consumer surplus to me. I don’t like it so much that I’d pay for it, and Google know that if they were to charge for search, they would lose market share instantly.
It also doesn’t seem likely to me that Google could monetise that consumer surplus by raising its ad rates. It’s a monopoly already; it’s squeezing as much out of the advertisers as they have to give.
Google might be able to serve me more ads, if I spend more time in the intelligent search box and less time on other sites. But this is close to a zero sum game, in that those other sites will have fewer opportunities to sell me ads.
So, at best, the increment to Google’s revenue (and therefore, the only thing that has any chance at all of going into GDP) is the extent to which time spent on a Google site with ads is substituted for time spent on other sites which didn’t previously have ads. It feels pretty marginal, particularly since there’s a huge amount of capex needed to achieve it.
I think what’s going on here is that we’re experiencing a less than perfectly foreseen consequence of the big trend towards corporations trying to create “moats”. When a monopolist is earning monopoly profits, then they will invest to protect that monopoly, even if the investment is not one that would be justifiable as a stand-alone. If you are earning profits of 100, and you are faced with the need to make a big capital investment in order to ensure that those profits only drop to 90 rather than to 50, then you grit your teeth and make it.
If a lot of the capex going into AI is going into this kind of investment - negative sum games between monopolists trying to protect their rents - then we shouldn’t expect to see the big consequences that everyone’s been predicting. We’ll just get some more unmeasured consumer surplus, a few marginal players will go bust and the moat will keep getting filled.
Brad here with a few more notes: I think these are worth stressing:
Monopoly pricing is already maxxed out, so better search doesn’t obviously translate into higher ad prices.
Usage shifts are marginal, mostly zero-sum cannibalization of attention from competitors.
Moat attack & defense is very marginal for measured real GDP.
As best as I can see, the prevailing narrative that huge AI capex must herald a visible productivity or GDP boom has no foundation whatsoever. I see paths whereby users avoid getting their brains hacked and benefit massively in terms of the information and entertainment utility they acquire. I do not see anyone able to charge consumers or investing businesses a lot of money for these services.
But then it surprises me that Google, FaceBook, and Amazon are so dominant. So maybe I am missing something big here.
With respect to consumers, I think that Dan will eventually be proven entirely right. With respect to business usage, though, the game is rather different. It’s a rare earnings call these days that doesn’t feature extensive discussion of all the wonderful productivity improvements that are soon to arrive or, in some cases, already have. Some — most? — of this is just corporate pablum, but not all of it. Moreover, there’s bound to be some cost to switching vendors, so it’s not inconceivable that decent profits will be made. Enough to justify current valuations?? Almost certainly not.
I am a worn-out applied mathematician who has grown ancient and addled following the course of "artificial intelligence" literally since the day the term was coined, seven decades ago. Indeed, I became interested in machine learning even before 1956.
Now I make a certain use of LLMs and have no doubt that they make a small contribution toward improving my productivity in some research tasks. But the unstated and very limiting assumption that almost everyone makes is that the "language" involved in "language model" is the common tongue, or perhaps a somewhat specialized variant such as mathematical language or a programming language. Creatures of language that we no doubt are, I nevertheless see it as evident that our production in the economic sense is mediated by language in such sense only to a limited extent; an extent no doubt greater than anciently it was, but yet limited. It would appear to follow that the scope for improving economic productivity through application of LLMs is indeed circumscribed.
LLMs are surely not the only field of application for the current iteration of machine learning (ML) technology. Autonomous self-driving vehicles and self-guiding UAVs represent another field currently in the public eye, of course. Still another is the enabling or improving of a wide variety of industrial processes, as exemplified by https://pubs.acs.org/doi/10.1021/jacs.6c03234 . (Picked at random from the morning's email.)
Yet from what I can tell, the current hyperscaling investment frenzy is directed almost exclusively toward improvement of LLM capabilities, which, as Prof. DeLong trenchantly argues, offers somewhat dim prospects for return. Why is it, I wonder, that the AI industry chooses to focus its investment in this direction, rather in search for fields of application having greater potential?
All of which brings me back to "O'Neil's law of AI," originally formulated about 60 years ago: Those who are most obsessed with *artificial* intelligence seem often rather deficient in *natural* critical intelligence.