A revolutionary step forward in real life artificial intelligence is making Sci-Fi movies seem more and more like future documentaries. A team of researchers from the University of Cambridge have developed a mother AI (artificial intelligence) that builds her own baby AIs using mechanized blocks.
What’s more, each new generation is a little more advanced than the previous ones. The study has pretty much proven that Hollywood AIs aren’t really all that fictional as these new robots evolve on their own, exactly like humans and animals.
And they’re better at it than we are. The mother robot will look for the best traits and skills in her children, and once she finds them, she actively chooses which ones to give to future children and which ones need to be modified so that future children will be better at performing certain tasks. She is not a slave to genes that determine such things independently of her will.
For their study, the researchers from Cambridge set the mother robot to build children five (5) separate times and noticed that all of a baby’s best traits were passed down to its siblings. The fifth generation baby robots were capable of performing tasks twice as fast than their siblings from the first generation.
The findings were published in the journal PLOS One and concluded that each of the baby robots carried a certain “genome” that was made out of different genes. As this genetic information passed down from generation to generation, it deleted or merged with other genes so that each new generation would be better than the previous ones.
While we’re not likely to start hearing robot rights debates anytime in the near future, the findings come to light in a very interesting social context. A few weeks ago, Tesla Motors CEO Elon Musk, Professor Stephen Hawking and more than 1.000 other field experts signed an open letter warning about future killer robots.
The letter asked governments to put a ban on autonomous weapon development as such a practice can only lead to a global arms race where the Ais are the only winners. The scientists made sure to stress that they’re not against the development of autonomous robots, just the development of militarized AIs.
But at the same time, one of their arguments was that humans would lose a significant amount of control over essential combat decisions if the world’s governments started sending AIs on the battle field.
And they fear that it would be a wide-spread issue. They argued that nuclear weapons aren’t very accessible because they’re costly and their materials aren’t very easy to find, but killer robot AIs, on the other hand, would be cheap to make and require raw materials that are very easy to find. The scientists said that every major military power in the world would begin to mass produce them and use them for nefarious purposes.
While their letter may have sounded a little paranoid at first, the new development certainly gives it some weight. What if fifth generation militarized AIs decide that all humans do is kill one another, and that the planet would be better off without us on it? Could we really blame them for their point of view if the only thing we exposed them to is war?