AI is a Math Book & Isn’t Taking Your Job
You need Biologic Intelligence before that happens and only one group is building it
I. Setting the Stage
There’s a troubling trend happening across the media and public conversation these days, spreading fear and misunderstanding of the way something works.
No, AI is not going to take your job anytime soon. In fact, AI isn’t going to do much of anything to impact your everyday life anytime soon. At least not without a watershed innovation in the way that it’s built.
People say the letters A and I and immediately assume that we’re already at a technology as advanced as that shown in Hollywood, whether Terminator 2 or Bicentennial Man. It’s not judgement day yet, ladies and gentleman.
II. Narrow versus General Intelligence
It begins with something very simple. The difference between Narrow and General Intelligence. The most basic definition I can give is that narrow intelligence does one thing, and does it very well. So well, in fact, that it has optimized itself almost to perfect. You can recognize a simple object in a static picture. You’ve been shown 1 million baseball bats from all angles, lighting conditions, and backgrounds. So, it becomes very easy to recognize the next baseball bat you’re shown as a baseball bat.
General Intelligence works differently. It can recognize a baseball bat, similar to before, but it can also balance a checkbook.
The mechanics of how a system works to recognize a single thing, and only a single thing, is very different than being able to do 2 things with the same system.
Currently, narrow intelligence works with a variety of math equations. Here’s the part that pisses me off. Go back to high school, grab your math text book. Put it on the table in front of you and open it up. Read the equations to yourself.
Is that Terminator? Is that Judgement Day? Did that math book just take your job? Does your math textbook have a soul or consciousness.
Of course not.
But don’t you see that all this mayhem and fear and smart people talking about how these math equations are going to take your job are really just plain wrong?
III. Math Equations Run Amok
What they’re doing is a bait and switch maneuver. And you need to recognize these things as they happen. I’m the tech press. I write a sensational story about math equations run amok. You click on the headline. Ding, there goes the cash register with a new ad impression. A story about fear. Ding! A story about math taking your job. Ding! A story about Robots needing to be taxed by Bill Gates. Ding! Ding! Ding!
Please can we all just pause and stop.
Math isn’t going to take your job. But I tell you what will. General Intelligence. When a robot can pick something up, put something together, manipulate physical objects, and send a communication to another robot, then yes. If you work on an assembly line, you’re out of a job.
But didn’t we see this story play out in Hollywood, nominated for an Oscar? In Hidden Figures, one of the women was a supervisor for “computers”. You know, women who computed things by hand. Then came in this big hulking IBM machine that did computing much faster and more accurately. It “took” those women’s jobs.
So, lets go ahead and hate computers then, right? And then we don’t have Facebook or your beloved smart phone. Or video games. Or, or, or.
Let’s not get ahead of ourselves. Because I can tell you from personal experience being involved in building the future of general intelligence (i.e., Biologic Intelligence), that these mathematical approaches (i.e., Deep Learning) aren’t the answer. They aren’t going to get us very far. The only thing you see is a huge data center the size of a room, running the same calculation over and over again until it finally spits out an acceptable answer.
But that same math equation isn’t going to recognize a cat, balance your checkbook, or drive a car. Everything is time-consuming and a custom job. It’s not a product. You’re just doing the same thing over and over again, applied to a new domain.
There isn’t some new crazy approach that they’re doing. Pitting one “neural network” against another and calling it by funny-sounding names like Adversarial Networks.
Just because you link two things together and apply a weight to them doesn’t mean you’ve recreated the neural pathways and mechanics of a brain and nervous system. It just means you wrote down a regression line. You know: ax² + by³ + cz² + error
And all you’re doing is trying to minimize that error variable to zero. The line of best fit. Only it’s a squiggly line, instead of straight. That’s what the little 2 and 3 represent above the x and y and z. Just squiggles. You’re trying to draw a squiggly line around a bunch of data and separate it into two different buckets: like all of these over here are apples and those over there are oranges but they’re kinda mixed together.
Does that sound like something that’s going to take your job? Does that sound like something dangerous? Does that sound like something you should be fearful of?
Well, only if you get paid by getting ad impressions and want to write sensational headlines that get you to click through on Facebook.
How do I know this? Because I’ve got a theoretical Mathematics and Actuarial Science degree. I used to memorize equations that ran across entire pages of text books then took a series of professional examinations that made the CFA look like a coloring book, with nothing more than a pencil and blank piece of paper.
IV. How To Recognize Falsehoods
Do not be fooled. Dig into the details yourself and stop regurtitating things other people have said. Repetition is your enemy. That’s how Trump got elected. It doesn’t have to be true, as long as it keeps getting repeated, you will eventually believe it.
Yes, even if you’re an engineer taking online courses trying to learn about this voodoo called artificial intelligence. Ahem, excuse me. I mean linear algebra. It’s shown some cool stuff. Like hey it played poker and and go and chess. It played jeopardy. It can drive a car (kind of).
But at the end of the day, it’s still a bunch of humans sitting behind a computer and typing commands into it, trying to hand annotate data and then feed it into a computation device to run a million calculations over and over and over. The result is surprising, I’ll grant you, and that’s why everyone is getting so crazy these days.
But you have to remember that the most successful AI business right now is a startup that labels millions of rows of excel data, by hand. Cat, cat, cat, cat, dog, cat, llama, cat, cat.
It doesn’t sound so intelligent once you realize that a human is spoon-feeding all of this “intelligence” row by exhausting row into a machine into the first place. Then a math equation takes all those cats and spits out “cat” when you ask it for a cat.
So I ask you. Is it taking your job?
Once Biologic Intelligence starts getting bought by major industrialists, and exists in space, helping to ingnite the largest gold rush in human history (space mining and manufacturing), then you can start getting concerned.
Because that means what we’ve been building has made its way into the real world.
And then you have to ask yourself a very real question. Are you lazy and going to sit on your ass when our software and robotics takes your job? Or are you already going to be building something on top of this new platform to profit from this new industry that never existed before.
If you’re reading this, then you already know the answer. You’re curious. You’re involved. You want to contribute. You get excited by it. You will become a part of it.
So, you don’t have anything to worry about.
But the others? The ones who just sit there and don’t want to “work” will just become an additional drain on the system and those that are already working hard to make the world a better place.
Ayn Rand, you minx. The motor of the world won’t stop, it just may leave a few of the rusty cogs behind.
Your Recommended Reading
- Notes from Gigaom’s 2017 AI Conference
- Early Stage Robotics & AI Funding & Market Size
- My Tour of SpaceX Yesterday
- Self-Driving Cars Are Hurtling Towards an AI Brick Wall
- Biologic Intelligence is NOT Artificial Intelligence
from Stories by Sean Everett on Medium http://ift.tt/2m7kNop