ADAM SMITH CORRECTLY DIAGNOSED THE SITUATION YEARS AGO.

I used to summarize the political economy of antitrust with the following:

To widen the market and to narrow the competition is always the interest of the dealers ... The proposal of any new law or regulation of commerce which comes from this order, ought always to be listened to with great precaution, and ought never to be adopted, till after having been long and carefully examined, not only with the most scrupulous, but with the most suspicious attention. It comes from an order of men, whose interest is never exactly the same with that of the public, who have generally an interest to deceive and even to oppress the public, and who accordingly have, upon many occasions, both deceived and oppressed it.

I wouldn't name the author, rather I'd ask students if anyone had an idea who it was. More than a few for Karl Marx and sometimes I'd get a J. S. Mill, both good choices.  It's in The Wealth of Nations, Book I, Chapter XI.

Widening the market and narrowing the competition is easier with Government Doing Something.

It doesn't matter how good the intentions of the legislators who established the board is, or how much public enthusiasm there is for Wise Experts to carry out the legislative intent.

In principle, the concept of the independent commission is a delegation of powers from the executive and legislative functions of government to subject specialists, holding the ability to implement legislative intent without burdening those legislators with determining what the price of a streetcar ride or the content of a sandwich served to economy class fliers was to be.

Like most principles, the concept sounded better on paper or sketched on a chalkboard than they did in practice.  A left-leaning commentator of the nineteenth century responded to the creation of the Interstate Commerce Commission with an epigram to the effect of the agrarians and the communists getting their laws, and the robber barons keeping their railways.  The same lesson applies to any other independent regulatory commission, or to the more recent administrative state agencies to which we turn.

The problem with any delegation of legislative authority to an agency is that the people most affected by the agency decisions will take the greatest interest in what the agency does.  In the early years, a farmer might have a keen interest in what the rate on a carload of grain is, or an environmentalist in the acid content of power plant smoke.  Over time, though, the regulatory procedures become codified, and boring, and the agrarians and communists of the next generation find other shiny objects to get worked up over, whilst the railroaders or power companies or television stations or air carriers or baby formula manufacturers find ways to get along by going along in such a way as to capture the agencies.  The agencies thus become cartel managers, protecting the incumbent firms, and the best thing to do might be to reduce their powers and banish them forever.

Does anyone seriously think that, should the leading developers of artificial intelligence, or machine learning if you will, will not cultivate the members of any agency Congress or a legislature sets up to regulate their competition?  It doesn't even take the kind of insider majority on the boards, as Mr Slivinski fears.  Wouldn't the management of any business subject to that regulation figure out where the board members play golf, when their birthdays and anniversaries are, what their children aspire to, in order to capture the agency, no matter how stocked with idealists it might be, with kindness?

The post that provided the long excerpt concluded, "The simplest way to reduce the rent seeking might be to generate fewer rents in the first place."  But the basic behavior of regulatory commissions is to generate rents and then dissipate them.  Reason's Tosin Akintola warns what will come next.

There's something disingenuous about sounding the alarm on the existential threat posed by AI models while simultaneously capitalizing on every improvement made in the field. If [Anthropic chief Dario] Amodei and others of his ilk are truly worried about creating a product capable of mass extinction, nothing is stopping them from pausing or slowing development in their own labs.

Instead of doing so voluntarily, they're turning to the heavy hand of government intervention, which, as [Adam] Thierer points out, inevitably "entails foreclosing innovation and competition within a sector," and "that sort of centralized control breeds politicization and cronyism."

Stephen "Vodka Pundit" Green connects the dots.

What Anthropic needs is a moat, and those usually come courtesy of Big Government. Think of taxi medallions, franchise laws, or pretty much any profession requiring a government license. All are examples of regulatory capture, where private interests use government to feather their nests.

So when I see the chiefs of a money-losing industry begging to be regulated, it makes me think less of the dangers of rogue AI's, and more about the dangers of private interests entrenched by government fiat.

Beware AI. But beware even more a "partnership" between AI and DC.

That's the kind of scrupulous and suspicious attention Adam Smith had in mind.  Unfortunately, the True Believers in Santa Claus Agencies Staffed by Dispassionate Experts charmingly keep the faith.

Daniel Kokotajlo, a former researcher in OpenAI’s governance division, told Witt that he’d like to create “an international, open monitoring system” that would act like an air traffic control system for AI development.

Under Kokotajlo’s plan, all AI researchers would be required by law to post information on training runs into a public database that can be monitored in real time.

“If we had this sort of total research transparency,” Kokotajlo told Witt, “then there would be this whole ecosystem of academics, third-party auditors, nonprofits, and rival corporations who would all be able to see what was going on inside the giant data centers of these big companies. Insofar as something scary or problematic was happening, anyone could sort of point it out and start a conversation about it online.”

Sure.  Every rule proposed by a Federal agency is published, meaning available online, in the Federal Register, and in principle each of us is free to check the notices and rules the webmaster has authority to release.  Same problem, at one remove, as the executives of a regulated company having more reason to know the regulator's special days than some random weblogger.  It's transparent in principle, but who stands the most to gain from looking at what's posted.  And "air traffic control system?"  How quickly would the Artificial Intelligence Research Register be using servers and search engines that are legacy compared with what the regulated companies are working on?

There are no solutions, only trade-offs.  A future post will explore reasons to be wary of information companies with relatively free rein to develop new forms of machine learning, or artificial intelligence, or enhanced data scraping, as well as of governments either regulating that development, allegedly in the public interest, or in controlling it.