I was thinking today about the subject of artificial intelligence and why people are so scared of it. Actually, let me start by saying that I didn’t believe people really were afraid of AI. I just thought they were making stuff up, or pretending, or telling stories. I didn’t think they were actually scared of it till I heard of Elon Musk’s fear of AI. Apparently he has said that he is pretty afraid of what people will do with AI. Actually he has said we are “summoning the demon” with AI. Apparently a lot of people believe this because there are a lot of movies made and stories told about AI turning against it’s masters and wrecking havoc or enslaving humanity. Basically, the idea is that robots can come up with their own ideas and we might not necessarily like them. But I always believed that AI was purely under the control of human invention because who would build a robot with its own ideas of what it should do, right? Apparently I am wrong. Because a lot of people believe that robots can come up with it’s own ideas.
But when you look at the programming of robots closely you see that there is no real possibility of it coming up with its own ideas unless you it programmed to do that (the infamous chat bot Tay comes to mind). Even when looking at the self-learning of complicated neural network programs such as Deep Mind’s Alpha Go you find that the network is relegated to pursuing one goal only: playing Go. The robot does nothing else and could never do so (without it’s programming being altered). Also, we have the interesting situation of never having a robot go rogue before. Though we see many examples in movies and TV of robots going against their makers, it has never happened in real life.
Perhaps you say because robots just aren’t smart enough? Robots are getting very close to being as smart as humans though they’re not as close to being as flexible and multi-disciplinary as the human mind. That’s part of their programming. Perhaps you may say, Robots just need to be programmed that way and then they’ll be a danger to humankind? it’s possible, but the history of computer programming shows that robots do exactly what you tell them to do and nothing else. They’re really very dumb in that regard. So it’s really a matter of computer programming to get the robot you are looking for. You may point out that robots just need to be a little smarter, but what is smarts really but programming? Even the human brain is a product of design – God’s design. So human intelligence is a type of programming that utilizes neural networks (though different from the ones in man-made deep learning networks). And programming is always limited by the programmer. You know your limits.
So I think robots could be developed that could help us create and do everything on computers that humans already do: just without the work! Is this a bad thing? Possibly. The internet could sometimes be treated as a bad thing and it is, but what would we ever do without it? In the same way, really intelligent robots represent a progress that has never been experienced before, but we can have a taste of what it can do with deep learning and neural net based robots today. It would be a shame if we cut it off while it was still budding.
If any of you are interested in learning more about AI and neural networks, I would recommend the book On Intelligence by Jeff Hawkins. It’s probably one of the best books I’ve ever read in my life and represents a great step forward for human-like intelligence based on the biology of the human brain. If you are interested in learning about the many applications of artificial intelligence (or “machine learning” as it is known), you can read about it here.