How Likely is a Robot Invasion?

Original article was published by Christopher Adam on Artificial Intelligence on Medium


How Likely is a Robot Invasion?

The rift between robot and human is shrinking at an alarming rate. From self-driving cars that make less mistakes than we do to smart speakers that can pick you out from a crowd of people! These innovations in smart computers have opened up the doors to a completely new realm of possibilities, some of which can help billions. There’s no doubt that AI is the future- It’s just that…

You can’t help but wonder…

Does that future include humans?

When Will the robots take over?

Now slow your roll there a little, maybe we’ve all just watched The Matrix one too many times. Surely there’s no way that computers would be able to be better than people. I mean, we made them… Right?

Well yes and no.

How AI works

Although humans made AI, we don’t actually know how it works. You see, using math, engineers make machines learn through this thing called a neural network. A neural network is a series of mathematical equations that takes an input, say a picture of Chris Evans, and converts it to an output, Captain America.

This is done at the most simple level, by taking these input values (because everything in the computer is actually stored as a series of numbers) and passing them through oftentimes thousands of layers of nodes. The computer then assigns a random level of importance to each node, we call weight. After calculating with all these weights in mind, the computer spits out an output. The engineer, or another program, would then compare this output with a prelabeled output for the given input. If the output is wrong, the computer recalculates with another series of random weights until it finally gets it right or the engineer stops it. Basically, the computer makes random guesses until it gets the right answer.

So how does this relate to the impending robot invasion?

Well you see, with all of these random guesses, we don’t actually know what the computer is doing to get to the right answer.

We don’t really know how it works.

Don’t believe me? Check out these Facebook Robots that made their own language to communicate without the creators understanding.

https://www.youtube.com/watch?v=QaoDXYYtgK0

But even with these terrifying developments, you don’t have to beg for Siri’s mercy quite yet.

General vs. Narrow AI

When we think of Robot Masters, what we’re thinking of is called General AI (AGI). General AI is AI that, “allows a machine to apply knowledge and skills in different contexts.” (1) Basically, it lets the program think for itself. This differs from Narrow AI (ANI) which can only perform one specific task.

The thing is, AGI doesn’t exist yet.

Current technology simply isn’t enough to support AI on that scale, not to mention the moral implications of conscient robots. I mean, do they have to pay taxes? What if a robot kills someone, do you punish it or the creator? Do we have to pay THEM? We can’t even deal with climate change as a society, much less living robots.

So Will the Robots Take Over?

No.

Well, at least not for another 40 years 😉 (2)

References

  1. https://www.springboard.com/blog/narrow-vs-general-ai/#:~:text=Narrow%20AI%20(ANI)%20is%20defined,and%20advances%20in%20that%20spectrum.%E2%80%9D
  2. https://research.aimultiple.com/artificial-general-intelligence-singularity-timing/