Intelligence Testing Regimes and Theatrics

How shall we ever conceive
However express a
new idea
If we are bound by the categorization
That delivered our problem to us
In the first place
?
-- Stafford Beer

How do you measure intelligence?

You take a tape measure, wrap it around someone's cranium, and jot down the number of inches. Or you weigh their skull and calculate how much brain matter you think it can hold. You set up a booth at the International Exposition of 1884 – step right up, step right up – and charge people admission to have their height and weight and eye color noted, their eyesight, hearing, and posture detected. You devise a test to be administered to millions of Army recruits in the First World War in order to determine their intellectual fitness, their “native intellectual ability”: “Washington is to Adams as first is to ....” You invent a test for children to identity the “feeble minded,” and then you order your test subjects on a scale, sorting them into their station in life. You invent a formula – you divide mental age by biological age – and call it IQ. You propose a "g factor," some math you says points to "general intelligence." You see there's money to be made in these numbers, in a "scientific" system of ranking students, so you churn out innumerable versions of standardized tests, promising schools and businesses that they're easy to administer, easy to grade thanks to a new technology called "multiple choice." You adopt machinery to streamline the assessment and the analysis.

You insist that social hierarchy is natural, intelligence heritable – you say you have the science to prove it.

Your theories of “intelligence” are part of race science, one bound up in classification and taxonomy, in bell curve bullshit, one inextricable from Stanford University and from the school's long history of eugenics. Your ranking and rating based on "intelligence" is not simply about who has access to education or employment; it is always accompanied by programs to regulate immigration and marriage, to control human reproduction, to determine who will live and who can breed.

The history of "intelligence testing" is over one hundred years old, and when you invoke "intelligence," this is the legacy you are a part of.

We understand this. Or we should. We understand that measuring is a misfire. We understand that whatever standardized test scores purport to "count," that they fail to capture a complete picture of someone's capabilities or potential. Education companies have sensed our reticence, so they’ve relabeled their products, obscuring the lineage of "intelligence testing" and offering the more palatable "achievement tests" or "aptitude tests" or "placement exams." “We believe in merit,” they wink.

(That education technology has recently rebranded these products from, say, "learning analytics" back to "artificial intelligence" is a sign that it too has embraced race science.)

Over the last few decades, our politicians and education reformers (and technology companies) have pushed for more and more standardized testing – again, all part of a project to rank and rate students, not to mention punish teachers and schools. Those of us opposed to the ubiquity of testing often point to its racist origins – the biases built into the questions, into the practice itself. We contend that standardized testing is a mismeasurement, to borrow from Stephen Jay Gould, one that reifies intelligence, sees students' capabilities as fixed, and thereby undermines opportunities -- through both institutional judgement and through personal self-doubt. While we have seen some success in making standardized tests optional in some university applications, the practice remains one of the predominant ways in which access to education is managed, by which entry into professions is gate-kept.

How do you measure artificial intelligence?

You take the measurements used to rank and rate humans and ask the machinery to take some standardized tests. You call the machinery's achievements "benchmarking." You devise new tests, but probably do not bother to consult psychometricians; they're busy testing children anyway. You adopt the "g factor," even though it's utterly discredited, and try to sell everyone on the imminence of AGI, artificial general intelligence.