My Big Idea came to me on a soggy August day on Long Island Sound, captive in a lifeless O’Day Mariner, knee to sweaty knee with the houseguest I so wanted to please, sails slopping about uselessly, out of beer and potato chips, at the mercy of the small outboard which—of course—conked out.
During the long embarrassing tow, my guest, who was a physicist, speculated that a “shear pin” in the motor failed, exactly as it was designed to do, to keep the aging and overheated putt-putt from cooking itself to death—a deliberately weak link that breaks the circuit before real damage happens. How brilliant! I thought. What if such a circuit breaker in my brain had stopped me from suggesting Let’s go sailing! on a day clearly meant for an air-conditioned movie theater.
Wouldn’t it be great if automatic brakes in our heads shut us down before we shot off our mouths? Or shot someone else?
Such purposeful failure is routinely engineered into just about everything—by engineers, or by evolution. Sidewalks have cracks that allow clean breaks, preserving the squares when trees uproot them; bumpers crumple so people don’t; eggshells crack easily to allow chicks to peck their way out. Either the eggs fail or the chicks do.
My houseguest, as it happened, had worked on the Manhattan Project, and we both immediately thought: What if a similar safety switch had scotched the bombing of Hiroshima, which “turned people into matter,” as II Rabi later put it, himself one of the many Nobel laureates present at the creation. He was also one of many who was haunted for life by horror and remorse at the terrifying destructive power of the weapons they had built, and the purpose to which they were put.
Of late, prominent creators of AI are expressing horror at the potentially destructive power of their own brilliant tech, which in some sense also turns people into matter, or rather into products in the form of data, vacuumed in and spit out by monstrous machine farms that gobble resources like water and power at a mind-stopping rate, spewing vast amounts of carbon—which is also matter, but not in the form useful for humans.
Some of them are asking for brakes too—at minimum, speed bumps to slow the mad race to create “nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us.” That wording comes from the now notorious “open letter” that put thousands of technologists on record asking for just such a pause. Some are talking of human extinction.
In fact, some of the parallels between the bomb and our new AI brains are uncanny. Before Hiroshima, the physicist Robert Wilson had called a meeting of the bomb scientists to discuss what should be done with “the gadget.” Perhaps they should consider some options, maybe plan a demo or something, before dumping the thing on people—using them as test dummies (rather as AI-driven cars, some say, also use people test dummies). The “father” of the bomb, Robert Oppenheimer, declined to come. He was already caught up in the momentum of the thing, the admitted “sweetness” of the technology, and besides, someone was bound to do it.
Read the full article here
