The AI Con by Emily M. Bender

Author:Emily M. Bender
Language: eng
Format: epub
Publisher: HarperCollins
Published: 2025-05-13T00:00:00+00:00

Extraordinary Claims, Extraordinarily Thin Evidence

In order for their story to hold water, and keep commanding policymaker attention, the Doomers have to convince their audiences of imminent artificial intelligence. Some of the Doomers speak of an AI system that becomes “smarter” than its designers and thus is able to design even more advanced systems, which design even more advanced ones, quickly overwhelming what we can even comprehend. This will result in a system that we can’t hope to control. Geoff Hinton,41 one of the so-called “Godfathers of AI” whom we met in Chapter 1, framed his Doomerist concerns to CNN journalist Jake Tapper by saying “there are very few examples of a more intelligent thing being controlled by a less intelligent thing.” We wonder when the last time was that Dr. Hinton suffered from any kind of food poisoning, and if he then decided that bacteria are more intelligent than humans. Meanwhile, a common refrain of the AI Boosters is to imagine systems that are “smart” enough to solve our problems for us (cure cancer! solve climate change!). And both the Doomers and the Boosters are asking the general public, investors, and policymakers to make decisions based on their extraordinary claims.

Extraordinary claims require extraordinary evidence, so you might hope that the Doomers and Boosters would go above and beyond in carefully substantiating the basis of their storytelling. But you would be disappointed. In order to evaluate “intelligence”, we’d first need a clear and operationalized definition of the concept, along with a compelling narrative of how it relates to the Doomer/Booster claims. Then we’d need a way to test for it, such that we can show that the test is actually measuring the property of interest. We saw in Chapter 2 that, throughout their history, measures of “intelligence” in humans have been based not in sound science but in eugenics and racism, so that’s already a bad start.

This isn’t a new problem: people have been pondering it since at least the 1950s, when Alan Turing wrote42 in his paper “Computing Machinery and Intelligence” that the question “Can machines think?” is too vague, as there are not sufficiently precise definitions of “machines” and “think”. His solution was to replace that question with a “closely related” one, which has since come to be called the Turing Test. The popular understanding of the Turing Test is that if a computer can fool a person into thinking it’s a person, that computer has “passed” the Turing Test and must therefore be considered intelligent.

In fact, Turing’s “imitation game” was a bit more convoluted. He set up a scenario in which an “interrogator” communicates with two other participants via teletype. In one version, the participants are a woman (trying to help the interrogator get it right) and a man pretending to be a woman (and trying to fool the interrogator); the interrogator’s job is to determine which participant is the man and which is the woman—a much more gender-bending version43 than is usually discussed. In the second version, the participants are a man and a computer pretending to be a man.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.