In a move that can only be described as politely panicked, over 100 artificial intelligence scientists, including several who have designed the very systems currently threatening to overthrow your job—and perhaps your sense of reality—signed an open letter pleading with governments to take regulation of AI a little more seriously and a little less like a Netflix plot device.
The letter, released Tuesday by the AI Policy Institute, was endorsed by some of the most prominent researchers in the field, such as Geoffrey Hinton, who is often referred to as the godfather of AI, presumably because “Frankenstein’s uncle” did not look as good on a nameplate. These experts are urging governments worldwide to develop dedicated institutions which could, with any luck and moderate funding, shape the direction of AI development in a way that does not end in an accidentally robotic coup or the collective deskilling of humanity.
The signatories emphasize that artificial intelligence carries a “staggering” array of risks, which include deception, manipulation, job displacement and, for the cherry on top, potential human extinction. The letter delicately suggests that leaving tech companies to self-regulate would be roughly as effective as letting kindergartners grade their own behavioral reports. It also notes that these tools are advancing so rapidly that they now require “regulatory institutions on the scale of existing government regulators for medicine, aviation, or nuclear safety.” Which is another way of saying that if AI could fly, it would already have asked for airspace over your local elementary school.
In case anyone missed the urgency, the experts clarify that no such institutions currently exist. At all. Which explains why many are concerned that AI regulation at present relies heavily on press releases, vibes and occasional gentle nudges from overworked senators.
This call to arms—or at least to bureaucracy—discourages laissez-faire optimism about AI’s trajectory and suggests that a bit of old-fashioned rule-making might be in order before we all find ourselves outwitted by a chatbot named Clippy 2.0 with a passive aggressive streak.
Because when the machines come knocking, it would be nice if someone official was home to answer the door with a clipboard and a policy.