Three New Stories and some Repetitive Essays
23 September 2020
I have to admit that the only reason I purchased this book was so that I could read the three robot stories that aren’t in any of the other books that I have, and that is quite annoying because there are a bunch of stories that appear in the other books meaning that I pretty much skipped half the book because it ended up that I would be simply rereading something that I have read multiple times before. Still, as I mentioned, there were three stories that I hadn’t read, as well as a collection of essays that Asimov wrote on robots, one of them back in the 50s, and the rest of them in the 80s.
As he says, most of his robot stories deal with conflicts in the Three Laws of Robotics (something that he drums on about repeatedly, as well as the fact that he is the first person to have ever coined the term robotics, though he also points out that this is probably the only long-lasting contribution that he has ever made to humanity). The first few stories weren’t so much like this, but when he and Campbell sat down and nutted out the rules, they did open up lots of opportunities to write about how they interact, and the problems that arise when these laws come into conflict. Then again, the entire legal profession builds itself around the fact that laws can never be hard and fast, and there are always exceptions, and loopholes, and ambiguities that can be exploited. As some have said, every time you attempt to plug a loophole in the law, a hundred more open up.
One interesting story has a robot named Rambo (Asimov suggested that all robots have names that start with R, meaning that no human has a name that starts with R). This quite clearly tells us when the novel was written because, well, the word Rambo only entered the English language after the character appeared in the film First Blood (which I have to admit is actually a pretty good film). I sometimes wonder if this was intentional, namely because Asimov is demonstrating how language changes over time, and that you can actually date a story based upon the words that are used, even a word that is as innocuous as a name (my English teacher once said that the name Shane didn’t appear until after the film of the same name, though the internet suggests that he may have been wrong).
With the essays, it is interesting to see how dated that they are. Okay, we did have a rudimentary form of the internet back in the 80s, but Asimov was writing as if robots needed to have all of their thinking power inside of their units, but this is no longer the case, with wifi and with the internet. In fact, with the development of the cloud, processing power is stored elsewhere and software accesses this power remotely, meaning that robots don’t need to have all of the processing power inside of them. In fact, I suspect that a lot of automation is done this was these days, and that driverless cars would also be using this technology (though it doesn’t solve the problem of what would happen if the network went down).
Yet there is also the question of whether robots can ever think and react like humans. Sure, we have machine learning, and some of the methods are designed to mimic the way the human brain works, yet the catch is that human brains don’t think in binary – we think in different ways – computers simply come down to thinking in terms of 0s and 1s. Another thing is that you have to tell computers everything that it needs to know. For instance, if we put a cup on the table, we know that this cup will be there when we return (unless something happens otherwise, such as our housemate puts it in the dishwasher). This needs to be programmed into the robot, as well as contingencies (if it is not there, somebody has moved it – yeah programming computers comes down to a lot of if/then statements).
It is interesting to see how Facebook developed the reactions that exist beyond simple likes. We were discussing this in one of our Machine Learning classes, how it is a way to teach computers what makes us sad, what makes us laugh, and so on. Yet, I’m still not convinced, that we all of this data being passed through Facebook’s servers, that a computer will learn to be able to respond to a joke or even be able to create one themselves. Another thing is that the first time we hear a joke we consider it funny, but as time goes on, and we continue to hear it, it ceases to be funny – can a computer be trained in that method as well, or is it the case that if a computer learns that something is funny, then it just laughs whenever it sees that joke, without realising that the joke has ceased to be funny years ago.
It is interesting reading this book, and the essays, after two and a half years of computer science, and halfway through an AI subjects. I suspect that a lot of developments came out of Asimov’s theories, but there were a lot of things that he couldn’t speculate on because, well, he was a chemist that liked writing Science-Fiction. Personally, computers tend to be reactionary, and can really only react to things that it is told to react to. Okay, they can search, but once again the parameters must be given to it. On the other hand, one can argue that the same is the case with us. However, our brain is able to take in an awful lot more information, whereas computers must be instructed to take that information, and has to be specific as well. Sure, we do have advanced machine learning algorithms, but the reason is that people have already created them. Mind you, as Asimov suggested, even when robots to replace humans, they also tend to open up a lot more jobs that humans are required to do.