Tyler Cowen Grief

I read Tyler Cowen almost every day, he is probably my single favorite source of novel information. But I recently noticed some patterns in his thoughts on three disparate topics that struck me as problematic and I am lumping together in one grievance post.

The AI technology comparison set. In his existential-risk post¹ he ranks today's AI alongside fire, the printing press, and fossil fuels. These are huge and conceivably warranted, but I was immediately confused how/why he would leave out electricity, the computer, and the internet. Fire and the printing press are distant enough that "nobody could have predicted their impact" is trivially true, and that unpredictability is the whole basis of his radical-agnosticism argument. Electricity and the internet are recent enough to be measurable yardsticks, and by that yardstick it seems much more resonable to list today's LLMs as genuinely useful tools rather than a rewiring of civilization. The fire comparison only holds if you assume the AGI that has not arrived and treat its progress as inevitable rather than something that could stall. Grant the AGI extrapolation and he may well be right. But that is a bet on a trajectory, written as if it were a description of the present. Maybe recent legible technologies are bad analogies precisely because AI's impact is unknowable, so fire is the best comparison, though I think fire is more legible than electricity in many ways. But if the future is so unknowable, radical agnosticism and "immense benefits, take the plunge" makes just as much sense as doom. I'm not saying he shouldn't be optimistic, just that he needs a better reason than this.

"The left," "the right," "woke." Cowen would never accept any of these labels for himself. His views are too nuanced to reduce to a team. He is right about that, just like every person he labels would also be right to defend their individuality in the same way. In the wokeness column² he writes that "the social energies of the American left have moved... into plans for concrete action," handing an obvious abstraction a single will it cannot wield. He concedes the assassins' connections to the left "can be debated," and assigns collective responsibility anyway through the vibe of "left-wing rhetoric." A category that fits no real person can be blamed for anything, and I'm certain he would agree that this same generalization is used to great detriment by many other pundits and political manipulators and in much worse faith. It is disappointing to see him voice the same tired (and lazy) crutch which splits any "group" into a "positive side" (good woke: gay rights, anti-racism) and an "unreasonable side" (bad woke: cancel culture, DEI), which makes the category mean good-things-I-endorse plus bad-things-I-don't, which means it doesn't even function as a generalization. His "I don't let the right off the hook either" does nothing but extend the two-team ontology while pretending to stand above it. Of course aggregates can be real (he's an economist!) even when no individual matches the stereotype, but there is a difference between aggregate-as-measurement ("X% of left-leaning voters polled...") and aggregate-as-agent ("the left's energies moved into plans for violence"). He is performing an intellectual disservice to data-appreciating readers like myself which struck me as beneath my expectation of him.

UAPs. Cowen puts extraterrestrial origin at one to ten percent and rejects the alien-bodies claims outright³. He has even argued the real secret is mundane government surveillance capability, not aliens. But I think he accepts the government's general premises too readily, jumping too quickly to agree that the sensor data is "real and not a mistake of instruments." Yet amateur astronomer acquaintnces of mine who would love it to be aliens, keep finding parallax and lens artifacts. This reflects a more general gripe about when non-experts take strong positions on things. I gave Cowen's take I think more credence than it deserved; he seems to have come around because of multi-sensor correlation which should not be waved off, but the correlations require study on a case by case basis, and when one of my astronomer friends did this he found the evidence for a trend was not that great. "I can't immediately explain this video" is the foundation of conspiracy nuts, but not a strong reason to drop one's skepticism.

I guess my grief at large is that I trust Cowen a lot and have recently recognized that what I took for good-taste skepticism is more selective than I'd hoped, and he does not seem to notice. None of this makes him less worth reading, if anything it just means he deserves even closer reading.