Bregman has fully embraced late stage capitalism

Of course, several people sent me Rutger Bregman’s video; "My secret goal, of course, was to provoke a rant from you" someone wrote, haha. Let's go, I guess.

I always find it difficult to write about Bregman; I’ve written before about his sexist views on vegetarianism, his book, and his lack of system-level critique (all in Dutch). Inevitably, people will say: you're just jealous of his success. Sure, who wouldn't want to sell a hundred thousand books?

Let's start with a compliment though: now that he's accepted baldness and ditched the combover, he looks good! Not only his hairstyle (and a fine mustache) deserves a compliment, but the video too; the comparison to Al Gore is super clever, well-executed, and highly effective.

So much for the positive stuff.

Now for the content—nothing surprising there; he hits all the talking points of the hype, saying exactly what the AI companies keep repeating: AI is amazing, "here to stay", and we must embrace it because it's a tremendous opportunity, but be careful... it's also dangerous so we must keep it under our control (EBOLA!!).

Below I take apart a few detailed comments, but the most important thing to note I think, is the striking absence of Bregman's claim to fame—"taxes, taxes, taxes". He does talk about the enormous profits of AI companies, but makes absolutely no mention of the fact that this would mean, in his previous thinking, that they'll have to pay more taxes. Maybe to fund, I don't know, a universal basic income or something? He does mention taxes as a means by which people could hold rulers accountable in the past (no workers, no taxes, no castles), but it is clear he no longer believes taxation of the rich to be a path forward. You could argue that AI companies would need to compensate for all the taxes people no longer pay. For now we do still have democratic rule that could enforce this, but no, Bregman has lost faith in governments' ability to do so. That is a very interesting shift, which he himself does not mention.

So what is going on there? Knowing the stuff that is being said in tech circles, we see clearly that Bregman has fallen into the narrative of AI abundance, everything will be solved by AI, but we have to control the AI, such that is works for us. AI is going to give us "Universal basic wealth", says Bregman, echoing a key talking point of the tech bros, and of course a key phrase of himself, but modified. No longer will we need to control the wealth of rich people, we need to control their technologies. If only he would talk about that change of heart, this could get us some insights.

Let's now dive into a few specifics (or not, click here if you want to skip to the end)

Individual arguments

Bubbles build infrastructure -> No, unlike train tracks or internet cables, data centers only last a few years!

But it's worth a hundred million billion -> That's mainly due to the construction of those data centers, says even Deutsche Bank—not exactly left-wing, is it? [Growth is] "not coming from AI itself but from building the factories to generate AI capacity" (ibid.)

Just look at what it can do, it's no parrot -> Bender, McMillan-Major, Gebru, and Smitchell's argument isn't that the output is wrong! If you're going to discredit people by name and with photos in your video, have the decency to actually read the paper itself. One danger is precisely that the output seems correct, and that people fill in the rest: [[1]]

humans are prepared to interpret strings belonging to languages they speak as meaningful and corresponding to the communicative intent of some individual or group of individuals who have accountability for what is said.

Bender wrote about this herself last month, not in response to this video.

Further on, Bregman falls right into that trap when he calls a chatbot "devious" and "cunning", which is, of course, exactly the kind of intent Bender is talking about.

Chomsky—did I mention that I'm very left-wing and that he's my hero—said that it can't make moral judgments, and look, LLMs can win math competitions, make diagnoses, and program -> Those aren't moral tasks...?

Also, his argument—which goes back to that of Joseph Weizenbaum—is not that computers can't do it, but that they shouldn't (“A computer can’t be held accountable, therefore it must never make a management decision”)

But the AIs are now coding themselves! a little while later I have never programmed myself -> Okay, maybe you shouldn't say too much about programming then? Just a thought—stick to what you know, or (see above) at least do some proper research, for example on model collapse.

But then there's the more high-level criticism: just like the AI-Delta plan people in the Netherlands, Bregman assumes there's such a thing as "our values," but that’s nonsense:

They assume that "we" Dutch people all share the same values (namely: democratic ones), and that an AI can reflect those. But in a pluralistic country like the Netherlands, there are diverse, contradictory, and conflicting values. To preserve such a society, we don't need more technology but more humanity, more understanding for each other, and more willingness to listen. (translated from Dutch)

In summary

Most interestingly, we see that Bregman has moved from thinking that governments can and should fix problems (with taxes) to thinking that companies can do so. Specifically, Anthropic. It will not be lost on the astute observer that the video—as is customary nowadays with the hypers—criticizes OpenAI but expresses admiration for Anthropic (which, according to Bregman, kept Mythos out of the hands of the "bad guys"). My god, have they cleverly positioned themselves as "good" AI! The race is not won on better AI in the sense of performance, but in the sense of (perceived) morals.

He has fully embraced capitalism, which of course is not a surprise given that his School of moral ambition also states this, implicitly. [[2]]

I'll wrap this up with a compliment, because what Bregman is doing is just brilliant. He dismisses people like Bender—who would wholeheartedly agree with him on the power of the oligarchs—as naive and misguided, and thus manages to eat his cake and have it too.

Notice also that his aesthetics have entirely taken the form of anti-vaxxers's, the "I am giving you the real truth that others won't"-vibe, a bit like Sabine Hossenfelder's increasingly over the top and now openly hostile science communication video's.

In Dutch we say "Talk left, get rich on the right" ("links lullen, rechts zakken vullen"), but this calls for a whole new saying: "bashing left, cashing right". He keeps saying (nominally) left-wing things, but also sucks up the tech-bro right, and his hyping totally helps their story of dominance, which is he is so worried about.

Notes

[[1]]: And many other dangers, such as environmental damage and exploitation

[[2]]: Also contextualized well in Dutch by the always sharp Bo Salomons