Technologies of Individualization Are Technologies of Inequality

The 74 published an op-ed this week, penned by an AI industry consultant, invoking Sputnik and claiming that "AI education is the new space race." I'm going to set aside its problematic call for "AI literacy" for another day; and although it's really the main gist of the author's argument, I'm not going to engage too deeply either with its xenophobia. It's not the Soviets who pose the threat (LOL?), but the Chinese apparently; but the gist of the argument is similar to the one made in 1957: the American education system must invest in rigorous science and math education for the sake of national security. Or at least invest in "AI literacy." (Also LOL.)

The author points to the National Defense Education Act, which was passed in response to Sputnik and marked an unprecedented injection of federal dollars into the US school system; and she argues we need that again, but for AI. So I guess I'm sidestepping too how utterly tone-deaf this whole article is considering the Trump Administration is actively trying to dismantle the Department of Education and axe funding for all sorts of scientific research initiatives in and adjacent to K-12 schools and universities.

The passage of the NDEA, as I argue in Teaching Machines, was a boon for the nascent education technology industry as it generated an incredible amount of interest and funding from the government and from philanthropies for education R&D, for all sorts of content, curricula, and gadgetry. And as I've said before, Sputnik was, in its original instantiation, already "the Sputnik moment for AI," and the passage of the NDEA initiated a great deal of interest and investment in AI in general and in education – that is, in intelligent tutoring systems.

But to talk about the changes in education in postwar America solely as a response to Sputnik, solely as a boon for math and science education, and then to call for changes to education today so that we can echo that very "arms race" isn't just bad history or bad policy or bad metaphor – although the latter should give you a clue that it is, indeed, bad: a "weaponized" vision of education is not a virtuous or just vision of education, folks.

As Jim Wynter Porter argues in a 2018 article in Multiethnica, one of the "underexamined aspect of NDEA legislation was its incentives for curricular stratification by 'ability.' About 50% of its total spending was for increased 'intelligence' testing," used to group and track students based on "their natural, individual differences." [emphasis mine]

Wynter Porter argues that the passage of the NDEA – and specifically its support for intelligence testing – must be viewed alongside that other momentous governmental intervention of the 1950s: Brown v Board of Education, the Supreme Court decision that ended de jure school segregation. The discourse around the NDEA's passage, he argues, "worked to justify durable structural changes that remade or maintained divisions by 'race' (and so 'race' itself) via racialized tracking." That is, while race-based segregation was ostensibly dismantled post-Brown, much of it was then rebuilt through tracking and through discrimination based on "individual ability" – ability as decreed by intelligence testing, a practice with its own racist history.

Tracking students into different classes and into different schools – dividing students into "bright, medium, and dull" as Wynter Porter puts it, into Honors Math or AP History – has come and gone and come again, has been rebranded and technologized in the decades since the Cold War.

That individualization and "intelligence" work hand-in-hand should be no surprise to those familiar with education technology – its practices, its history – and the ways in which ed-tech has, on one hand, promised a technologically-enabled meritocracy, serving all students at the level and pace they need, while at the very same time re-inscribing educational inequalities through an algorithmic sleight-of-hand that never has to mention race or racism. (And by re-inscribing educational inequalities, I should add, by extracting data from students and schools and dollars from communities that might better be used to build human capacity rather than buy gadgets and software.)

"The new psychometric individualism could still accomplish the work of 'race'," Wynter Porter contends, "not only because of the neo-hereditarian principles around which it was structured, but also because 'race' was embedded latently in the tests, in the structures of schooling itself, and indeed in the broader knowledge production practices and social matrix that surrounded them."

"Individualized learning," particularly when enacted through technologies that defy scrutiny and accountability through algorithmic decision-making, echo the legacy of "separate but equal." Individualization surely has great appeal – it is a core tenet of American ideology, after all; and most teachers today who embrace it do not see themselves as reinforcing discriminatory practices. But in the school and in the classroom, individualization operates through the establishment of a hierarchy based on "ability," often determined by standardized testing – a way to avoid saying "IQ" perhaps while continuing to practice ranking based on "intelligence."

This ranking is at the heart of artificial intelligence too – not simply the rating and ranking of machine over human but the ranking and rating and sorting of humans.

AI determines access to content and to opportunity and – this is crucial – to human support. AI technologies offer a less valuable, less human educational experience through the ranking and rating of "intelligence."