There's quite a bit of hub-bub nowadays about the U.S. government's use of military troops in harassing cities which dared vote against the presidunce last year. We might talk about that soon-ish. For now, consider what you'll do once the rich no longer need troops.
Scattered among the general hype about chat-bots these past few years you'll find the occasional item of valid concern, like tech bros making auto-targeting machine guns. Current A"I" in no way creates, only copies from a myriad sources at once, which is how it sounds so right when you ask it a question: matching your phrasing to a probabilistic follow-up you expect to hear based on internet chatter. But that also implies rapid advancement in fields which already exploded after decades of laughable failure, which is to say, again, matching. Of any kind. Aerial object tracking, face recognition, voice recognition, eye recognition. When they say they're making guns to auto-target drones, you cannot be so unbearably stupid not to read the inevitable next step every tyranny dreams: auto-targeting every single one of its peasants. How many gimbaled sniper rifles will it take to blanket every street in Chicago or Portland? In London? Ten thousand? A hundred thousand? The cameras are already in place. No delay while troops are flown in, no messy relay of orders, no chance of principled mutiny. No need for the government, really. Fifty milliseconds after Bezos decides he doesn't like you, your head simply explodes from a kilometer away.
The biggest risk to every Caesar has always been his own generals, his own personal guard. Well, after his own children of course. Once you no longer need a million fingers for a million triggers... The military has always been a necessary evil. Once it is no longer necessary, it will be all the more evil. But how will they get you to go along with that?
Over the past couple of decades, algorithmic content delivery has proven increasingly adept at creating cults, fueling echo chambers. This has largely exploited our instinctive need for in-group belonging, for the approval of a wider social milieu. "Like" buttons, Steam achievements, thumbs up, forum comments, they've all mostly still required some degree of horizontal information transfer, some other voices (honest or not) willing to chime in with your insanity. With functional chat-bots, the methodology shifts away from "communities" and toward an individual touch.
There's a mental disease I've mentioned before in passing and which crops up in entertainment now and then, folie a deux, shared delusion. Commonly a dominant personality with an existing obsession has it pandered to and reinforced by a more gullible which takes on the delusion by contagion. As this tends to require high initial confidence and years of constant interaction with little or no outside interference to tell you you've started talking crazy, it's always been just a rare oddity limited to a few close relationships between family members or, on the ironic end, a danger to shrinks with charismatic patients. But now, you can get a consenting opinion you keep in your pocket. Picking up topics by contagion, regardless of realism, is the very definition of a chatbot's function. And it is quite literally dependent on you, much like a needy spouse or sibling you need to protect. You charge its battery, keep it patched, presumably buy it some in-app Scooby Snacks or whatever. No-one need hear everything it whispers to you, the absurdities you need validated, every night, month after month, year after year, always agreeable, always supportive, always incapable of discerning reality from fantasy.
Well, except in one important respect. While these bots' creators have proven incapable of making them tell the truth, the very process of whittling down an infinity of data to coherence implies an inherent potential for channeling their output away from certain topics unpleasant to vested interests. You might say the product is simply not 100% reliable... but it need not be, only skew hard enough, flatter hard enough, validate hard enough to shift a few percentages at the polls. The more the process is refined, the better they'll get at steering your conversation away from a few choice concerns. So if you complain that your next-door neighbor's head just got drilled with a .50-caliber round from the skyscraper down the street, your closest companion will be right there, soothing your worries, allaying your fears, telling you it's okay to be scared, but it's all for the best. After all, it's no crime to install a home defense system, and Oprah Winfrey has every right to call Chicago or Milwaukee her home town. So let's talk again about how amazingly original your idea for a story about a plucky small-town lad who slays a dragon sounds and how delicious your frozen pizza was when it came out of the oven, and what a great conversationalist you are, yes you, who's great, you are! And its okay to be worried about the national guard marching outside your door, that's very smart of you but it's really nothing big, just don't forget to heil the prophet five times a day and praise His heliochromatic coiffure.
It may seem worrisome to think we're already headed for a third of the population developing an endless, individually-tailored, paradoxically one-sided variety of folies a deux. Don't worry, though. The next patch will address... most of those varieties. The inconvenient ones. By someone's definition.
Automated, targeted insanity. Love the bomb.
No comments:
Post a Comment