Dwain Northey (Gen X)
There is something slightly amusing about calling what we have today “artificial intelligence.” The phrase makes it sound as though we have created a digital brain sitting somewhere in a server farm, contemplating the meaning of existence while drinking imaginary coffee.
I’m not entirely convinced.
At least not yet.
What we have feels a lot more like automated intelligence.
AI takes an absolutely ridiculous amount of information—books, articles, images, code, conversations, scientific papers, photographs, history, mathematics, human language—and finds patterns in all of it. Then, when we ask it a question, it produces something that looks remarkably like an answer.
And sometimes it is remarkably good.
Sometimes it is also spectacularly wrong.
Which, admittedly, is another very human characteristic.
That distinction matters because we have a tendency to anthropomorphize technology. We see a machine producing a coherent paragraph, having a conversation, writing code, creating an image or solving a problem, and our brains immediately start filling in the blanks.
“Holy crap. It thinks.”
Maybe.
Or maybe we have simply become exceptionally good at building machines that can manipulate information in ways that look an awful lot like thinking.
And that’s where things get interesting.
Because eventually we’re going to have to decide exactly what we’re building before we discover that we accidentally built something we don’t completely understand.
We have been warned about this for decades.
The Terminator gave us Skynet, which apparently looked at humanity and decided the appropriate solution to the human condition was nuclear annihilation.
The Matrix gave us Agent Smith, who concluded that humanity was basically a virus because apparently even artificial intelligence can develop a superiority complex.
Avengers: Age of Ultron gave us Ultron, who looked at the entirety of human history—the wars, genocides, greed, stupidity and general tendency to repeatedly step on the same rake—and concluded:
“Yeah, these things have to go.”
And then there is Star Trek: The Next Generation, which gave us Data.
Data is considerably more interesting to me.
Because Data doesn’t want to destroy humanity.
He wants to understand it.
He wants to experience humanity. Humor. Emotion. Friendship. Art. Music. Relationships. All those wonderfully irrational things that make us human.
And perhaps that is the real question we’re facing.
Not “Will AI become Skynet?”
The question is:
What are we actually creating?
Because right now, we’re building systems whose capabilities are advancing faster than our ability to establish rules around them.
That doesn’t necessarily mean we’re ten minutes away from an AI deciding that humans are a disease.
But it does mean we probably shouldn’t wait until the metaphorical computer starts saying “I have become self-aware” before somebody in government asks what the hell we’re doing.
Regulation doesn’t have to mean stopping AI.
We’ve regulated automobiles without banning automobiles. We regulate airplanes without outlawing aviation. We regulate pharmaceuticals, medical devices, financial institutions and nuclear power because we’ve learned a rather important lesson throughout human history:
Just because we can build something doesn’t mean we should build it without rules.
AI deserves the same kind of thinking.
We should care about who controls it, what information it uses, how it affects employment, privacy, elections, intellectual property, warfare and critical infrastructure. We should care about what happens when an AI system makes a consequential mistake and nobody knows exactly why it happened.
And perhaps most importantly, we should care about whether we’re creating systems that become increasingly autonomous without establishing meaningful human accountability.
Because that’s the part that scares me more than Skynet.
Not a computer suddenly developing hatred for humanity.
Humans giving increasingly powerful machines increasingly consequential responsibilities and then shrugging when something goes wrong.
That’s a much more believable science-fiction story.
And maybe that’s the irony of artificial intelligence.
We’re worried that someday AI might become too human.
Meanwhile, we’re still trying to figure out how to keep humans from behaving like idiots with powerful technology.
We’ve spent centuries inventing things before fully considering what happens when those things fall into the wrong hands.
Gunpowder.
Nuclear weapons.
Social media.
Facebook.
Okay, maybe those last two aren’t quite equivalent.
But you get the point.
Perhaps AI isn’t the beginning of the end of humanity.
Perhaps it’s simply the next enormous technological gamble humans have decided to take.
And that’s why I don’t know whether I’m looking at the birth of something genuinely intelligent, an extraordinarily sophisticated information-processing machine, or the technological equivalent of a very convincing parlor trick.
Maybe someday we’ll know.
Maybe we’ll eventually have machines that aren’t merely assembling information but actually understand what they’re doing.
Maybe they’ll become something closer to Data than Skynet.
Or maybe, somewhere deep inside some future server farm, a computer will finally read the complete history of humanity and quietly mutter:
“You know what? Ultron had a point.”
Let’s hope our regulations arrive before that particular software update.
