Words matter, says every client in every meeting.
Words matter, say hack-assed creative directors, without every bothering to read the words.
Words matter, as another billboard goes up that hasn’t a soupçon of stopping power, charm or wit.
Or even coherence.
Words matter.
And then we bring on legions of machines that don’t understand words, but merely select them based on a set-of-pattern-matching probabilities that are most likely to give you a set of words that look just like a set of words you saw not long ago.
Words matter.
But not enough so that we select them carefully.
Most of what we read and hear and say has the originality of an in-flight ad for a credit card offering points. You can’t walk out on a plane. But no one can bear to listen.
In fact machines don’t understand words—they don’t think after all—they just predict which one should come next. It’s like looking at Muybridge stills to learn how a horse gallops.
There’s a lot you can, or a machine can, infer from the above. But as Angus Fletcher points out in his book “Primal Intelligence,” (I’m quoting from the Wall Street Journal review you can read here)
AI can steal datasets from stories and recycle them into plots but, Mr. Fletcher is keen to remind us, the computer is still not reading the story. We are.
We are the ones taking the word scrambles and lacing meaning into them. Mr. Fletcher uses moving pictures to drive the point home: If you ask AI to generate an image of a running horse, it will do so with bits and bytes of data, still images that it pushes together for us to view.
It’s our brain that sees the horse run; we write action and story over bald data.
The subhead of the WSJ review of Primal Intelligence is something the propagators of trillionaire-backed AIs are hopeful that we’ll ignore or forget.
In other words, while AI can spit out words, it takes human understanding to select the right ones. Man’s search for meaning is not algorithmically based.
Here’s an example from “1929,” Andrew Ross Sorkin’s great book on “the greatest crash in Wall Street History,” at least until the trump dump, which is (my guess) coming soon.
Until Herbert Hoover, the periodic collapses experienced by the amerikan economy were typically called “panics.” We had:
The Panic of 1837.
The Panic of 1857.
The Panic of 1873.
The Panic of 1893.
The Panic of 1907.
The Panic of 1920.
Hoover was scared that using the word “Panic” would create a panic.
So, though while the NY Stock Market lost 25% of its value in just one day, though unemployment rose to one-person-in-four out of work and industrial output fell by half, Hoover didn’t want people to panic.
Hoover chose to use the word Depression.
Depression.
Here’s Sorkin:
And that didn’t quite work out.
Words matter.
Even more, so does thinking. Which machines cannot do. sam altman’s purported singularity or “super-intelligence” are just marketing hype.
We’re 30 years into the cellphone era, and about 70% of calls drop and about 2% of people understand their bills. We’re five years into the bot era and we can’t get a problem solved. And merry-izing the tone of voice and employing human-like interjections don’t fool anyone but the bought and paid for.
It’s the South Sea bubble.
The Tulip bubble.
The Credit Mobilier bubble.
It’s the 1929 Market.
Ponzi-flated for the benefit of a dozen men.
A dozen men constantly raising more money.
Marketing-hype designed to separate the gullible from their money.
It’s all as real as the word Intelligence in Artificial Intelligence.
Going back to the Journal’s review of Primal Intelligence:
Having defined intelligence as “logic,” we’ve inadvertently turned education into endless data mining.
Logic, reduced to its basics, is math—it’s a formula that operates only when there is data to plug in. That’s the whole mechanism behind algorithms and computer science.
We’ve become so accustomed to thinking of logic as the primary good that we miss the obvious problem: Data, by its definition, is information that is known. It already exists.
You can mix and recycle to make new tweaks, but it cannot lead to innovation—and it can’t tell you what to do when the future is unknown.
We’re using the wrong tools for the wrong job, getting the wrong results, and concluding there’s a 2.0 that will ameliorate that.
There’s a word for that kind of thinking.
But this is a family blog.
No comments:
Post a Comment