Former secretary of state Henry Kissinger, former Google CEO Eric Schmidt, and Cornell University professor Daniel Huttenlocher worked together for three years to write a single article for the August 2019 issue of The Atlantic. In it, they set out to solve the “riddles” of artificial intelligence, which they call an “unstoppable revolution,” but they only managed to mimic other AI sign-twirlers.
Their conclusions read like an encyclopedia entry for “hyperbole”: “The challenge of absorbing this new technology into the values and practices of the existing culture has no precedent. The most comparable event was the transition from the medieval to the modern period. … When the unity of the Christian Church was broken, the question of what unifying concept could replace it arose.”
AI software is more than technology, the authors contend. Its advent will change the meaning of truth as we know it: “The Enlightenment philosopher Immanuel Kant ascribed truth to the impact of the structure of the human mind on observed reality. AI’s truth is more contingent and ambiguous. … How should we respond to the inevitable evolution it will impose on our understanding of truth and reality?”
AI Exaggerations Are Everywhere
Even readers with no technological background or grasp of AI capabilities will surely adjust such claims for inflation. Come on — one moment we’re talking about facial recognition and sorting functions, and the next we’ve got a second Protestant Reformation warping the fabric of reality?
But if you’re paying skeptical attention, you can find the same AI exaggerations everywhere. Kissinger and the rest of the team even admit this in their Atlantic article, noting “public projections of AI have the attributes of science fiction.”
Yet by claiming, as they do, that AI will yield “entirely new ways of thinking” and cause an “inevitable evolution … of our understanding of truth and reality,” the authors fall prey to the very sci-fi anthropomorphism they criticize. Only beyond merely personifying AI, they practically deify it.
Some fair-minded observers have called AI a second industrial revolution: Software now performs complex calculations in numerous industries, and some human-like behaviors can occur at great speed. But before we do Silicon Valley the favor of writing AI’s hagiography, we should be honest about its defects — and theirs.
Show Us the Proof
Here are the facts: Scientists and developers have put forth absolutely no proof that AI technology will ever be successful on a macro scale in noncontrolled environments. Despite its public image, software cannot yet match the versatile skills of even the most primitive mammals in the real world. For all the fictional portrayals of AI software as some autonomous “intelligent” entity, no such entity yet exists.
Oft-cited examples of thinking machines are laughably simplistic: IBM’s Watson can look up answers to “Jeopardy!” questions. Uber’s “self-driving” cars require two human safety specialists at the controls. Honda’s ASIMO humanoid robot can (almost, sort of) waddle around like a toddler. The future is now!
Those whose scientific understanding comes from scrolling through “tech news” believe this unironically. That’s because, while the actual AI explosion is perpetually 10 years away, the AI marketing explosion has already been detonated.
Tech companies constantly create AI illusions for expo demonstrations and puff-piece media. Kissinger and the team unwittingly open with one such illusion: the AlphaZero chess engine, the vaunted exposition of which and subsequent AI stardom was essentially a hoax. Google techs provided AlphaZero a massively more powerful processor than its computer opponent, and moves were limited to one minute. It was a con very similar to Deep Blue, whose handlers reprogrammed the chess engine’s “learning” software after every game in the iconic 1996 series with grandmaster Garry Kasparov.
Robots that make the news, such as the acrobatic “parkour bot” shown off by Boston Dynamics, are always tested in highly controlled environments with scores of human operators. Even Sophia, the android who grabbed worldwide attention for its scripted conversation at Saudi Arabia’s Future Investment Initiative, has social media accounts run by human AI impostors.
That’s how the heralds and press agents of AI operate. It’s all about the show.
Cutting Through the Propaganda and Publicity Stunts
Fittingly, celebrity limelight-seekers buzz around the AI fad, swelling its influence with increasingly grandiose predictions. Steve Wozniak predicts we’ll soon become the “pets” of benevolent robots. Elon Musk says AI is more dangerous than nukes (but has invested billions in “merging” it with human brains). Richard Dawkins, for one, welcomes our new robot overlords. Even Vladimir Putin jumped on the bandwagon, claiming that whoever leads in AI “will become the ruler of the world.”
We might be tempted to view AI’s advertisement avalanche as harmless. After all, publicity and marketing are innocuous forms of free expression and necessary aspects of the modern economy, right? Wrong, Marshall McLuhan would certainly remind us: Mass manipulation by the “ad men” camouflages brainwashing, dehumanization, and exploitation. The goal of AI propaganda is to generate public support for the otherwise unconscionable.
One example is “fauxtomation”: the growing problem of tech companies routing inane tasks through low-income human employees, often in offshore sweatshop-like call centers, while pretending their AI is just that good. It’s another consumerist ploy we’d be fools to fall for. While social media magnates flaunt their advanced “quality control” algorithms in public, they privately hire thousands of content moderators in Manila to sift through hours of illegal porn and violence, while relying on a first-world staff of eager leftists such as Jen Gennai to manually censor and stage-manage the global discourse.
Sorry, Something Went Wrong
That’s not all. Since AI’s economic and social dominance is a fait accompli in the minds of its prophets, many long-game planners have begun to reallocate resources away from human involvement. Why develop new auto safety features if self-driving cars will soon eliminate accidents? Why teach foreign languages in schools if translation software will soon make human polyglots obsolete?
Why study or innovate anything at all if AI will soon handle our thinking for us? That kind of attitude will leave our civilization at the mercy of Microsoft Help. Alexa’s “sorry, something went wrong” could be the motto of our future.
Kissinger and other grave voices warning about AI are right that we’re in big trouble, but it’s not because of some “singularity” or robot uprising. Like our progressive optimism and myopic materialism, our shortsighted futurism has lost all sense of proportion. We’re tearing down the human infrastructure of today to make way for a technocratic tomorrow — and we’ll find ourselves left with the worst of both worlds.