How Eugenics Shaped Statistics

Exposing the damned lies of three science pioneers.

Nautilus

In an amusing twist, Richard Lewontin's landmark 1972 paper—which showed that the concept of race has no biological basis—uses Shannon entropy in its arguments rather than conventional biological statistics. Though this is not a knock on statistics, one can obtain the same result using conventional statistics.

@mehluv @viz

@dialecticalmusings Finding connections between modern tools and past eugenic ideas isn’t hard, but that alone is a weak argument. Tools like statistics and logistic regression are general-purpose and also enable beneficial things like vaccines and medical decisions. While it’s important to acknowledge problematic histories, to imply that tools are inherently racist or “baked” with those values is patently false and this line of argument is a waste of time.
@dialecticalmusings we need to understand that causality is complex and making causal claims is a bad idea unless there is a NEED to do so. we can oppose eugenics today and slavery today, without making essentialist arguments that not just are bad logic, but also bad politics.

@anandphilipc

I didn't make those arguments.

In fact, I had already answered a more or less similar question from you earlier.



RE: https://app.wafrn.net/fediverse/post/25fe80e1-fe4d-461b-85f8-484790925b8c
@dialecticalmusings yeah i was commenting on the verge article. we discussed this.

@anandphilipc

I have not watched “Ghost in the Machine”, but I disagree with your interpretation of Valerie Veatch’s argument (though the article doesn’t do a great job of bringing it out either).

What is artificial intelligence? We know what “artificial” means, so the question boils down to “what is intelligence?”. Galton and Pearson had a deep-rooted belief that people of various races were fundamentally different in quantifiable ways (in simple terms, they were racist). This belief influenced how they defined, and more importantly, how they quantified, “intelligence”. This carried over into the characterization of artificial intelligence. And into the architecture of machine learning. So… (Valerie Vaetch is saying) eugenics and racism are shaping how artificial intelligence is being defined, developed, and deployed at present.

(Check out this paper for an illustration of how Pearson’s beliefs influenced his actual work.)

You can also look at this from a social-historical point of view. As Timnit Gebru and Émile Torres show in their TESCREAL Bundle paper, Elon Musk, Peter Thiel, Jaan Tallinn, Sam Altman, Dustin Moskovitz, Vitalik Buterin, Sam Bankman-Fried, Marc Andreessen, Bostrom, MacAskill, Kurzweil, and other driving forces of the artifical intelligence movement are adherents of (at least one of) Transhumanism, Extropianism, Singularitarianism, Cosmism, Rationalism, Effective Altruism, and Longtermism (TESCREAL). All of these ideologies are derived from the eugenics movement. Their ideologies are foundational to their work on AI. These ideologies shape the kind of questions they ask, the direction of the answers they go for, the data they gather, the datasets they curate, the methodologies and tools they choose/reject, the problems to which they apply AI as the solution, etc.

So yes, Valerie Vaetch is certainly showing that this development is historically contingent, that contemporary AI permeating our day-to-day lives is a descendant of colonialism, imperialism, capitalism, racism, and eugenics. But her primary concern is the present—having demonstrated that eugenics and racism (and colonialism/imperialism and capitalism) are woven into the warp and weft of today’s AI, she is asking what we are going to do about it.

At least that's my interpretation of her work.


#ArtificialIntelligence #AI #TESCREAL #ValerieVaetch #GhostInTheMachine #eugenics #racism #colonialism #imperialism #capitalism