James Wallace Harris, 3/13/21
Androids that can pass for humans have become very popular characters in books, movies, and television shows. Klara and the Sun by Kazuo Ishiguro is the second novel I’ve read by a literary writer to explore this theme, the first being Machines Like Me by Ian McEwan. I reviewed it back in 2019. I have to admit that literary writers do a better job at characterization than science fiction writers. Klara is a fully realized being that we come to know from a first person perspective.
And I was impressed with Ishiguro’s science fictional speculation. He develops Klara with a distinctive speech pattern and we learn about Klara’s way of perceiving the world through Klara’s narration. Klara is designed to be a girl robot, an artificial friend, and androids like Klara are referred to as AFs in the book. Other than looks, Klara’s personality has little to identify as female. Klara is a being with growing awareness and Klara’s consciousness grows by observations. We never know exactly what Klara looks like, but since Klara is supposed to be a companion for a teenage girl we have to assume it looks like one.
However, one of my greatest objections to stories about androids passing as humans is giving them gender. For some reason creators of such stories believe AI minds will have gender and that’s illogical. AI minds, no matter what their outer casing looks like will not have gender because they will not be based on biology. Nor will they have human emotions. All of our emotions are tied to our biological subsystems. Writers of these stories seem to assume people will want to buy machines just like themselves. That might be true, but it won’t happen.
Klara starts out life in a store waiting to be bought. I imagined an upscale Apple Store. We slowly learn how Klara thinks about things from its observations. Klara’s vision is broken down into a grid system. Sometimes this appears to help Klara with perspective, sometimes with identifying objects, and sometimes with analysis of the details of specific aspects of what’s Klara sees. If you remember Deckard using the machine to analyze a photograph in Blade Runner, that’s how I envisioned Klara’s visual field. I thought this clever of Ishiguro.
Klara is eventually bought to be the companion for Josie, a young teenage girl around fourteen who suffers from an unnamed medical condition. Josie’s sister died from a similar condition. One of the mysteries of the novel is what they suffer from.
Ishiguro fleshes out this story with many other current science fictional speculations. Some kids, like Josie have been genetically altered (think Gattaca) while others haven’t. Josie’s closest childhood friend Rick hasn’t. The society of this world has also put many people out of work while elevated other humans with high status jobs. In this story, Josie’s mom has such a valuable job, but her divorced dad doesn’t.
Another fascinating theme introduced by Ishiguro is theology. Klara is a Sun worshipper, which is logical since Klara runs on solar power. However, Klara’s simple-minded beliefs are hard to accept. Ishiguro makes his AFs childlike in their thinking. If we ever create AI minds with general knowledge, including a chip with all of human knowledge would probably only add a buck to the cost. AI minds will know all human languages, all of mathematics, physics, chemistry, and biology, all of history, literature, and the arts, all of philosophy, religion, and psychology.
But Klara and the Sun is a story, and for this story, Klara has an innocent faith in the Sun. Giving theology to robots is what made Battlestar Galactica great, and has the same effect here.
Klara and the Sun is really about human relationships, and all the plot complications deal with Josie and her family. Without spoiling the story, it’s what the humans want and fear from machines that make the novel philosophically interesting. Ultimately, I believe Ishiguro rejects our sillier desires for androids that pass for humans such as creating them to satisfy our sexual urges, but even the ones he does suggest seem tied to our basic instincts.
Overall, Klara and the Sun is an enjoyable story, but more and more I’m getting disappointed that writers can’t picture realistic AI beings, or imagine such realistic AI beings’ impact on our society. There are no real reasons to create robots that look like us. There’s no reason to think they will think or feel like we do. And they will certainly have immense intelligences that will dwarf ours. We will probably anthropomorphize them even when real AI beings with general intelligence emerge because that’s our hangup.
But how will AI minds see us? Our behaviors will be inscrutable to them. They won’t be able to imagine love, hate, pain, lust, anxiety, humor, greed, jealousy, pride, and all our other emotions. AI minds will understand us like biologists study all nonhuman living organisms. AI minds will see us through their statistical studies of our behavior and by theorizing on how our behavior coincides with our languages. But could you imagine love, hate, or pain if you’ve never felt them?
Klara believes Josie was helped by the theology Klara imagined, but there is no real reason to believe this is true. If you read this novel, pay attention to Klara’s usefulness throughout the story. Does Ishiguro ever suggest that an AF would be an actual useful companion for teenagers? Or is Klara no more than a sentimental toy?
How will humanity be helped or hurt by AI minds? I think such novels are waiting to be written. So far all the ones about computer overlords or humans passing for human are based on our most basic emotions. Writers need to think outside our brain box we can’t seem to escape.
I actually thoroughly enjoyed reading Klara and the Sun despite my nitpicking about how writers want us to believe we can create androids that will pass as humans.
Other Takes:
JWH
Auxiliary Memory 
Question: do these two literary writers follow the idea of Cartesian dualism, that the mental exists somewhere separate from the body? That the emotions of a robot become some natural consequence of giving them the seed of a mind? Or do they agree that emotions are coded in brains (or in the case of AI, in software)?
Neither book goes that deep. My personal belief is consciousness is completely integrated with the body. That’s why I don’t believe we’ll have brain downloading/uploading which so many people want to believe in. I assume the conscious mind of robots will be just as integrated with their machine bodies. Our emotions are tied to chemical systems within our bodies. Robots won’t have those chemical systems, thus they won’t have our emotions. Robots can have a whole array of sensors that detect aspects of reality we can’t. That might lead to modes of perception we can’t imagine.
It is my understanding that our emotions have evolutionary causes. They fulfill functions in survival, like self-protection, herd mentality, procreation and so on. Robots will not have the same environmental pressures, so there is no reason to assume that they will have emotions, unless we program them to behave in similar ways to us. Maybe there are algorithms that allow an AI to develop emotions based on a determination to survive, so that it feels fear when its robot body is in danger, for example. Anyway, I am interested in this novel. Ishiguro’s other SF novel Never Let Me Go was very good. The SF elements weren’t original but the character writing was very good.
But why program emotions in a machine that mimics humans? Emotions are analog, based on a spectrum of chemical settings. AI minds are digital. Also, I don’t believe we’ll program AI minds. I believe they will evolve through machine learning and future variations of it. If machines evolve the equivalent of emotions, they will not be anything we can comprehend, but be relevant to their digital/mechanical bodies.
In the story, Klara is often anxious, fearful, concerned, worried, etc. There could be digital systems to create similar wariness, but the machines won’t feel them the way we do. It will probably be statistical and intellectual, so that warning subroutines will focus on hunches created by higher rates of statistical input.
I’ve read numerous reviews complaining Ishiguro tackling the SF in only a superficial manner, and also the story of characters to not be deep or whatever. I’ve read interviews in which he seems to want to ask deep questions with this book, but as James wrote, he seems to fail to grasp the subject, and besides, SF authors have been asking these questions more exstensively for decades. I liked The Buried Giant, but I think I’ll pass on this one, because as far as I can tell he has written an SF book for an audience with no knowledge of SF, and for whom these questions might seem original, etc. (Much like that McEwan btw.)
I don’t really understand this criticism, though. I think you could say the same about 90% of the science fiction written. Recently I read Dark Matter and although Crouch gave us the bare minimum understanding of science, even giving his book a title that had nothing to do with the science, I thought it was a story successfully told. Ishiguro could write a Pinocchio story about a robot and do it well. Why expect him to turn out a hard-sf novel?
Yes you are right. I think the main criticism is not that it isn’t hard SF, just that his questions/approach is superficial. I haven’t read the book, but the other criticism boils down to the fact that the story isn’t well told (as opposed to Dark Matter, which I liked as well), but that’s a taste matter obviously.
I have just ordered Klara and the Sun. We’ll see. If it is really Ishiguro’s intention to make profound statements and he fails because his ideas are too simple or outdated, that would be a legitimate criticism.
I wasn’t planning on reading Recursion, to be honest. There’s too many other books that sound interesting!
Recursion was a total fail imo, no loss there. Curious about your take on Klara.
(btw, if you would consider to read Recursion, it would be cool if you kept the plot hole questions of my review in mind while reading and see if some of those questions are unwarranted)
Both Ishiguro and McEwan tell a compelling story. They are literary writers of great skill. But they embed their science fiction in very human stories, both of which I enjoyed and admired. My nitpicking applies to both literary and genre writers. I believe anyone who assumes we’ll create androids that pass for humans isn’t seriously considering the reality of AI.
I think we need to psychoanalyze ourselves as a species for our desire to create machines in our own image.
I saw that this was about Ishiguro’s new one, Klara and the Sun, so I really read no further. I’m in the middle of the Mick Herron spy book series (Slough House) and when I finish Klara might very well be up next. Ishiguro is one of my favorite (top 25 or so) contemporary writers along with Haruki Murakami and so many others. I’ll keep this post in my inbox for future reference. 🙂
I’ll keep an eye out for your review. I was surprised by how much you liked The Ministry for the Future by Kim Stanley Robinson.
For want of any better and less obvious comparison, I can’t help thinking of Philip K. Dick’s “Do Androids Dream of Electric Sheep?” Here, the androids are organic, sentinent and self-serving. They are nevertheless though machines, albeit living ones, because they have been created and their emotional capacity is comparable to them, rather than whether or not they are real or artificial.
What is remarkable about them though, is Dick’s conception of them as being able to be aware, thinking and feeling organisms. However, because they have no empathic relationship with “us”, I think they can be compared to what you call AI minds, even though they are living. In this case, as in Dick’s fiction, because of their detached appearance, their “emotions” are only superficial.
Many androids passing as human stories, such as Do Androids Dream of Electric Sheep?, R.U.R. and Blade Runner use biological methods which I don’t consider true robots or AI. If they are biological they can have emotions. Generally, the authors used these beings as symbolic in some way. In R.U.R. they stood in for workers. The kind of stories I’m considering are ones where we speculate on creating an artificial being with computer technology.
As you know Richard, but probably many people don’t. The androids in PKD’s novels weren’t like those in Blade Runner. They were without emotion or empathy. I always consider PKD using them as stand-ins for sociopathic or psychopathic humans without empathy. PKD didn’t want us sympathizing with them. Blade Runner turned his story upside down, saying androids could feel just like us, asked the audience to care for them.
No, they aren’t true robots or AI, but the theme in “Do Androids Dream of Electric Sheep?”, is that even living androids lacking empathy, are no different to anything mechanical. The point is, that both are artifically created and can’t be expected to have emotions or empathy. In that way, I thought the androids in the novel, were similar to what you said about AI.
Dick did compare the androids in “Do Androids Dream of Electric Sheep” to humans who lack empathy. “Blade Runner” though would never have been conceived without the book for inspiration, but the two are like quite different biscuits and don’t bear comparison.
It should be noted however, that in the Novel he wrote four years before “Do Androids Dream of Electric Sheep”?, “We Can Build You”, there are mechanical robots that do appear to have human emotions and empathy, in addition to one or two real people who lack them.
My point is biology generates emotions. So any android with biological origins should have emotions. Emotions are a byproduct of chemical regulation. There are people who lack empathy, or social awareness, but I’m not sure they lack emotions.
Yes, any living, intelligent things of natural origin probably will have emotions. Biological androids such as those in Dick’s fiction, are artifically created though, and it seems he was speculating that like machines, they would therefore perhaps lack empathy and emotions.
It would be interesting to give test readers Klara and the Sun, Machines Like Me, Do Androids Dream of Electric Sheep?, The Moon is a Harsh Mistress, When HARLIE Was One, the WWW Trilogy by Robert J. Sawyer, Galatea 2.2 and some other AI books and see which are liked best.
John Sladek’s two volume Roderick novel [now a whole book], is about a mechanical robot that does have emotions and probably empathy [similar to those in “We Can Build You”, but which are imitations of actual people]. I’m not sure what the author was trying to say regarding AI, whether it was just satirical [which it was] or that it’s human capacity was false, but we are supposed to sympathise with the robot.
Love to read. 🙂
„I have to admit that literary writers do a better job at characterization than science fiction writers.“ — I am not convinced. I think Ishiguro jumped on the bandwagon of what is popular right now. The book didn‘t add anything new for me. And as an example—I think that Ann Leckie did a stellar job with the character development of her main (AI) character in the Imperial Radch trilogy.