Our human brains are unfathomably sophisticated at performing a huge range of tasks -- enough to make us the ultimate apex predators of the animal kingdom, despite having no useful physical advantages over all the rest. But one thing we are almost universally incapable of doing is grasping probabilities.
■ If you properly shuffle a deck of cards, that exact order of cards has never happened before and will never happen again. It’s not even close, either: The odds of predicting any exact order of 52 cards are completely unfathomable. Even if you were to shuffle merely a single suit of cards in that deck, there are more than 6.2 billion ways it could come out, or close to a unique possibility for every single person on Earth. Add in three more suits, and there’s no analogy anywhere in the natural world that our brains can map to those kinds of odds. And it involves a set of inputs and outputs that anyone can easily hold in a single hand.
■ When people serve up hypotheses about the prospects of an AI-driven apocalypse, it’s important to bear in mind that the outcome certainly is possible. But the sequence of events that might deliver us there is long and involves a chain of probabilities that rapidly take the forecast to the realm of the incomprehensible.
■ That sort of uncertainty tends to be paralyzing. But we don’t have to think about the wild scenarios before we give closer thought to the outcomes we can comprehend. Just like a simple deck of cards contains vastly more possibility than we can even remotely grasp, circumstances as they already exist today contain far more than enough possibilities for bad outcomes that we should examine -- soberly and thoughtfully -- without waiting to ponder worst-case “Terminator” scenarios.
■ In fighting one of the worst human foes to ever curse the earth, Dwight Eisenhower observed that “[E]very battle plan comprises merely an orderly commitment of troops to battle under the commander’s calculations of desirable objectives and necessary resources, but always with the certainty that enemy reaction will require constant tactical adjustment to the requirements of the moment.”
■ We need that same limberness of mind today. The worst could come to pass, or so could the best. So could trillions of possibilities in between.
■ But we ought to start with comprehensible battle plans in mind for harms we can be certain will come to pass: Spouses trying to manipulate partners, candidates trying to destroy rivals’ reputations, criminals seeking to fleece victims, and bullies trying to exact pain on classmates. For each of these and many other obvious scenarios, we should be asking, “How could these tools be used for harm, how can the harm be prevented or mitigates, and who is legally or morally responsible for doing something about it?”
■ And we can’t try to solve every problem one-by-one: Like shuffling only a single suit of cards, the number of possibilities are still beyond comprehension. But we do need to start somewhere. Only when we break the big, highly uncertain problems down into smaller, highly likely ones will we start to know if we’ve got what it takes to make Eisenhower’s “tactical adjustments to the requirements of the moment”.