The half-life of a miracle
How does extraordinary technology become ordinary so fast?
There’s an Amazon Fresh shop right next to my office. If you haven’t been in one before - I hadn’t - you go in, you pick up the items you want, you tap your card and you walk out. There’s no cashier, no self-checkout, no scanning.
The first time I went, I asked the assistant at the exit where the checkout was. “You can just walk out,” he said. How could I get a receipt? “It’ll be emailed to you,” he said. But I didn’t enter my email? “If you’ve ever bought something from Amazon, your email is connected to your card details.”
The whole thing felt miraculous. A network of cameras figuring out what I was leaving the store with and sending me the receipt using its massive bank of existing data. Streamlined, easy, fast: this is everything that technology promised to be.
But if it sounds too good to be true, that’s because it is. While the cameras try to track your movements, in most cases workers in India manually review the footage and tally up your purchases.
Amazon has not automated cashiers, it’s outsourced them.
And while the company insists that its Just Walk Out tech is powered by AI, they are now walking back on it and bringing self-scanners to their stores.
More human than bot
When the iPhone was first released, one of the most popular apps was a fake pint of beer that reacted to the tilt of the phone (that app was so popular, by the way, that it used to generate $20,000 a day).
The iPhone, and all revolutionary new technologies, feel like magic tricks when we first encounter them.
But sometimes these new technologies are more trick than magic. In 1770, Wolfgang von Kempelen unveiled a machine that could beat any human at chess. The Mechanical Turk, as the machine was known, toured around the world - defeating challengers (including Napoleon) and astounding audiences with its extraordinary capabilities.
Inside the box, though, was not some futuristic robotic technology - but a very good chess player, crammed in and controlling the Turk’s every move.
Amazon’s stores are kind of like a giant mechanical turk: and they’re not the only one.
Earlier this month, at a Tesla event, Elon Musk unveiled his Optimus robot. He predicted every home would have one within a matter of years. These robots, he insisted, would soon be “the biggest product ever”.
After the reveal, these impressive bots worked behind the bar and mingled with the techy attendees. Here it is in action:
If you’ve ever talked to an AI like ChatGPT, or any sort of voice recognition software like Google Home or Alexa, you’ll immediately spot the problem: Optimus responds too fast. Voice recognition just doesn’t happen that fast.
Not to mention, the robot’s responses are quite clearly being invented on the fly, like a human speaks, rather than scripted and read out, like an AI should speak.
It’s pretty obvious that Optimus is being remote controlled, like a sort of robot marionette: which was later confirmed in media reports.
Hype cycle
Innovation always follows the same pattern: what starts as a miracle, soon becomes ‘meh’. In the case of Amazon and Tesla, it was because these technological marvels turned out to be Wizards of Oz, controlled by a man behind a curtain.
But other times, it’s just the passage of time that turns the extraordinary into the ordinary. Flat screen TVs, voice control, 4G: all incredible when I first encountered them, today the standard.
And the ‘half-life’ of these miracles seems shorter than ever. It was just two years ago that generative AI, the most extraordinary technology evolution since the advent of the internet, came onto the scene.
Today, more and more detractors insist that AI is not especially revolutionary, and that its problems (of which there are certainly many) will block its ability to drive radical change.
On AI, it seems we are plunging into the ‘trough of disillusionment’, as named by the US insights firm Gartner in their hype cycle illustration.
But we mustn’t let the limitations of AI blind us to the possibilities. This technology is miraculous, and there is no one behind the curtain.
I remember, in 2020, reading a piece in the Guardian that was supposedly written by AI. I suspected it was some kind of mechanical turk, maybe an old-school chatbot being repeatedly prompted to generate pre-written answers. When ChatGPT was released, I was proven spectacularly wrong: the technology really was that clever, and could generate good, original prose out of absolutely nowhere.
But now, two years later, you’ll find more reports about its problems than its opportunities. It’s a good thing for journalists to be critical rather than enamoured, of course, but a reminder of how fickle we are - and how quickly we adapt to things that once seemed amazing.
Slope of enlightenment
I still think that AI will transform our lives in ways we can’t even begin to fathom. Generative AI - those tools that can write essays and generate pictures - is just a peek into the window of what AI is capable of: mimicking humans, learning rapidly, replicating thought.
I suspect, and hope, it will make everything we interact with smarter, more personal, and more accessible. It’ll augment us and the people we rely on, like our doctors. It’ll make jobs more meaningful for workers, while creating a bunch of new, skilled jobs.
There will be plenty more troughs of disillusionment on the way, and likely some pretty steep ones. The bubble will burst, probably again and again. But none of it will change the ultimate trajectory that AI is on: a technology that will fundamentally change the world.
After all, the dot-com bubble burst at the turn of the millennium, and some thought it was all over. But today, the six largest publicly traded companies in the world are all internet companies.
The test of a technology is whether it is truly miraculous. Bitcoin wasn’t, and that's why it's fading into irrelevance. The ones that are, even if we quickly forget how extraordinary they once felt, are the ones that eventually change the world.
