So, Sam Altman, the boy wonder over at OpenAI, dropped another one of his sermons from the digital mount. Calls it “The Gentle Singularity.” Gentle. Like a velvet-gloved eviction notice, I suppose. He says we’re past the event horizon, the takeoff has started, and it’s all “much less weird than it seems like it should be.” Speak for yourself, Sammy. The inside of my head after three days on cheap rye is less weird than the promises you guys sling around. My cat coughing up a hairball is less weird.
He reassures us that robots ain’t walking the streets yet, people still die, and we can’t pop over to Mars for a weekend bender. Thanks for the bulletin. For a moment there, I thought my bookie was a T-800. Glad to know the Reaper still makes house calls the old-fashioned way. But don’t you worry your artificially intelligent little head, because we’ve apparently built systems “smarter than people in many ways.” My landlord’s dog is smarter than half the people on my block, and it doesn’t need a server farm cooled by the tears of unemployed poets. These “hard-won” insights like GPT-4, he says, will take us very far. Yeah, far from a paycheck, for a lot of folks.
This AI, this digital messiah, is gonna shower us with “gains to quality of life.” Whose life, exactly? The quality of my life usually improves with a full bottle and an empty answering machine. Altman’s talking “faster scientific progress,” a future “vastly better than the present.” Every prophet with a new gimmick says that. Remember “Plastics, son, plastics”? Now it’s “Algorithms, schmuck, algorithms.” He figures ChatGPT is already “more powerful than any human who has ever lived” because hundreds of millions rely on it. Hell, hundreds of millions rely on coffee and desperation to get through the day, doesn’t make the barista a god. Power ain’t just about reach; it’s about what you do with it. Can this ChatGPT feel the raw, bleeding edge of a broken heart? Can it write a sentence that makes you want to either kill yourself or run naked through the streets screaming with joy? Didn’t think so. It’s a tool. A complicated hammer. And we all know what happens when you give a toddler a hammer. Or a CEO.
The timeline is a real knee-slapper. 2025: AI doing “real cognitive work.” So, my job’s gone next year then. Good to know. 2026: “novel insights.” Probably insights into how to more efficiently lay off the remaining humans. 2027: robots doing tasks in the real world. Fetching your slippers, Sammy? Or maybe operating the automated bulldozers that level the slums to make way for more datacenters? The man needs another drag from this cigarette, just thinking about it.
But hold on, don’t slit your wrists just yet. Because even in the 2030s, “People will still love their families, express their creativity, play games, and swim in lakes.” Well, ain’t that a relief. For a second, I thought we’d all be uploaded into some digital nirvana, our consciousnesses frolicking in fields of pure data while our bodies composted. Glad I can still take a dip, assuming the lakes ain’t been privatized and renamed “Lake OpenAIada, brought to you by Abundant Intelligenceâ„¢.”
And here’s the kicker: “intelligence and energy… going to become wildly abundant.” Wildly abundant. Like cockroaches in a flophouse. And “ideas, and the ability to make ideas happen.” Sounds like a recipe for a whole lot of bad ideas happening very, very quickly. Remember that startup that wanted to disrupt shoelaces with blockchain? Imagine that, but supercharged. The mind, or what’s left of mine after last night’s fifth of bourbon, boggles.
Altman admits we go from being amazed to wondering when AI can do more, like write a novel or cure diseases. He calls it “how the singularity goes: wonders become routine.” I call it the human condition: never satisfied, always jonesing for the next hit. Give us a miracle, and by next Tuesday we’re bitching about the maintenance costs. He mentions “a larval version of recursive self-improvement.” Larval. That’s a good word. Something squirming, getting ready to burst out of a chest cavity you didn’t know you had.
Then there’s the self-replicating hoohah: robots building robots, datacenters building datacenters. “If we have to make the first million humanoid robots the old-fashioned way, but then they can operate the entire supply chain…” Yeah, and then they can operate us right out of the picture. Sounds like a great way to ensure the only thing “recursively self-improving” is the unemployment line. And the cost of intelligence will “converge to near the cost of electricity.” Fantastic. So my thoughts will be worth about as much as leaving the bathroom light on. I can already feel the wealth trickling down, can’t you? Trickling like a leaky faucet in a condemned building.
Sure, “whole classes of jobs going away,” he says, breezy as a summer fart. But the world will be “so much richer so quickly that we’ll be able to seriously entertain new policy ideas we never could before.” Like what? Universal Basic Scraps? A lottery for the right to operate a food truck that serves the robots? The world gets richer, alright. The same way a casino gets richer. Not by spreading it around.
He throws us a bone, claiming humans have an advantage: “we are hard-wired to care about other people… and we don’t care very much about machines.” Some days, looking at the news, I wonder about that “hard-wired to care” part. Seems like a lot of us are hard-wired to climb over each other for the last scrap of meat. And not caring about machines? Wait till your AI therapist starts charging by the minute for its “empathy” subroutine. He even brings up some subsistence farmer from a thousand years ago looking at our “fake jobs.” Hell, that farmer would look at Sam Altman’s job and probably offer him a shovel, tell him to do something useful for a change.
And the solution to all this potential chaos? Oh, it’s simple. Just “solve the alignment problem.” Make sure these super-brains act towards “what we collectively really want.” Who’s “we,” kemosabe? You and your board? The Davos crowd? Some global committee of thumb-twiddlers? “Collectively really want.” Last time a bunch of people collectively wanted something with that much fervor, they invented religion, or war, or reality television. All top-shelf ideas. He even admits social media is “misaligned AI” because it just gets you to keep scrolling. If that’s misalignment, I’d hate to see what a truly pissed-off, misaligned superintelligence cooks up. Probably turn the entire internet into a non-stop loop of cat videos and existential poetry, just to watch us squirm.
Then, after we’ve solved that little “alignment” pickle, we “focus on making superintelligence cheap, widely available, and not too concentrated.” Good luck with that, pal. That’s like saying you’ll make nukes cheap, available at Walmart, and ensure no one builds too big of a stockpile. Power doesn’t like to be shared. It likes to sit in its gilded cage and make pronouncements.
“We (the whole industry, not just OpenAI) are building a brain for the world,” he proclaims. A brain for the world. My own brain is a damn nuisance most of the time, filled with bad bets, worse women, and the ghosts of a thousand hangovers. The world doesn’t need one brain, especially one cooked up by a bunch of guys who think “disruption” is a virtue. It needs a stiff drink and a long nap, if you ask me. And “intelligence too cheap to meter”? Heard that one before, about atomic power. My electric bill still reads like a hostage demand.
He ends with a prayer: “May we scale smoothly, exponentially and uneventfully through superintelligence.” Smoothly. Uneventfully. Like a greased pig sliding into a sausage grinder. Right.
Look, maybe these code-jockeys will birth a new dawn. Maybe they’ll cure cancer, solve fusion, and teach us all to knit sweaters for the robot overlords. Or maybe, just maybe, they’re brewing up the biggest, fanciest, most technologically advanced way for humanity to shoot itself in its collective foot. Again. The only thing I know for sure is that all this talk of gentle singularities and abundant intelligence is making me thirsty. This old dog isn’t learning any new tricks from a microchip, not when there’s still whiskey that needs drinking and bad poetry that needs writing. The “long arc of exponential technological progress” he talks about? From down here in the gutter, it looks suspiciously like the same old circle.
Time for another glass. Or maybe the whole damned bottle. To the future, whatever the hell it turns out to be. Just try not to spill too much of it on us regular folks on your way to utopia.
Chinaski out. Probably for a refill.
Source: The Gentle Singularity