So, You Think You’ve Seen AI
Suddenly I’m hearing it even from my sometimes tech-challenged buddies:
“I really like AI,” says a fellow septuagenarian.
“I use it all the time to look up all sorts of things,” announces another. “I love it!”
A sexagenarian—who still has, of all things, a job—gushes: “AI is the best thing that has happened to the internet.”
And when I ask these senior citizens a snooty question: “Yeah, which AI are you using?”—just to see, you understand, how tech-savvy they really are. I usually get a reasonable answer: “ChatGPT.” Often they add a “4” or a “5.”
Don’t get me wrong. I’m no tech wiz myself at this point. And I’m proud of my fellow oldsters. They are staying up to date. They can handle a new technology.
When you’ve lived through the debut of so many new-fangled things—some of which panned out (smart phones), some of which didn’t (Google Glass)—it’s not a terrible idea to look with some cynicism upon the latest of them. But these older folks are not scoffing. They are marveling.
And they ain’t wrong in marveling. The answers you get on ChatGPT or Anthropic are usually solid, sometimes revelatory, sometimes marvelous and, as one of the senior citizens quoted above puts it, “it’s like talking to a human.” That’s quite a trick for a machine.
However, it is also the case that my alert, tech-savvy oldster friends, gushing about the capabilities of GPT 4 or 5—and plenty of alert tech-savvy, similarly gushing young’uns—have little sense of how sweeping the transformations unleashed by artificial intelligence might be.
For, having researched and written about the history of technology, I understand that the transformations new technologies inevitably produce often are shockingly profound but always slow to arrive.
For example:
* Writing would help organize increasingly large states and contribute to the great achievements of Egypt, Greece, China and Rome, but for its first thousand or so years it was mostly used just to record transactions or memorialize oral tales.
* The wheel would transform transportation and war, but for a few hundred years it was used exclusively to make pottery.
* The printing press would give us the French and American Revolutions and Tolstoy, but it took a century and a half before the first newspaper and the first novel were printed.
* Early computers were used to crack codes and then mostly for spreadsheets or word processing. It took decades and the internet, accessed through the World Wide Web, before we felt their unprecedented power to move information.
Moreover, if you asked us toward the end of the 20th century, we would have probably said that we love the internet—and the ability to send instant, stamp-less mail—while knowing nothing about iPhones and Amazon and Google, let alone Google Maps, Facebook, Spotify, Twitter, Match.com, PornHub and, alas, QAnon and Truth Social.
In other words, most of us in the early years—the early decades—of the internet did not have a clue what the internet would be.
And, for damn sure, we are still in the very early years of the great and exciting and scary world-transformer that is artificial intelligence.
It would be surprising indeed if OpenAI’s, Anthropic’s, Grok’s, Google’s initial attempts at artificial intelligence—arguably the ultimate human technology—provide much of a clue to what AI can eventually do.
Yeah, we’ve all heard the spooky talk about AI taking over everything, which may be somewhat exaggerated. But now we seem to have settled into seeing and loving OpenAI’s ChatGPT, for example, primarily as a souped-up search engine—one that provides prose answers and encourages follow-up questions.
Meanwhile, other, mostly younger, people have found real value in AI’s ability to write better than they can write or produce images to their specifications but beyond their talents.
Here, for example, is a skillful though not-all-that-spectacular ChatGPT imagination of Leonardo da Vinci using AI to produce the Mona Lisa:
But these are still just efforts at locating, borrowing, imitating. They are ho hum.
More is coming—as more, lots more, came in the first decades of this century with computers.
For the point of artificial intelligence is not just supposed to be doing what humans do only quicker and better. The point of artificial intelligence is supposed to be to think new thoughts, to conjure up new things: new cures for diseases, new inventions, new ways of organizing societies, new opportunities for humanity.
And apparently, we can expect—maybe next year, maybe next decade—that talking to AI, whether in the form of the current “large language models” or something somewhat less derivative, is going to be like talking to something not just better at chess than we are or incredibly better read than we are, but incredibly smarter than we are about more or less everything.
Since we Homo sapiens spent the last 300,000 or so years being the smartest things on the planet, that will certainly be a kick in the head.
And the danger of AI is not just that it will usurp human prerogatives and make many humans redundant. It is that it will leave humans behind.
First, we will need less from other humans: A growing number of young people are already interacting with “artificial” psychologists, friends or lovers.
And second, artificial intelligences may no longer need much input from humans . . . at all.
I’m no doomer. I don’t imagine mechanical higher intelligences, wanting to bump off us biological less-high intelligences. But it is not that difficult to imagine them supplanting us brainy but slow and neurotic great apes.
And I do think we should expect that if we ever are contacted by an advanced civilization from another solar system, there is a reasonable chance they will not be weird, three-eyed blobs of flesh but brilliant erector sets.
The true power of AI might not become apparent within what is left of our lifetimes. But this is going to be big and a lot more profound and much weirder than a revved-up search engine or an imitation painting.