9 New Books That Show How Truly Weird Artificial Intelligence Can Be
Janelle Shane, neural net programmer extraordinaire, recommends fiction about the possibilities of AI
I’ve encountered a lot of artificial intelligences, both the ones I’ve trained for my blog AI Weirdness, and the ones I’ve written about for my book on artificial intelligence, You Look Like a Thing and I Love You: How AI Works and Why it’s Making the World a Weirder Place. I focus on the machine learning algorithms that exist today, the ones that sort spam, tag photos, and drive cars. We call them AI, but they’re as different from the AI of science fiction as a toaster is from a person.
In the book, I spend a lot of time explaining why today’s AIs, with their tiny worm brains, don’t understand their tasks or the human world. They won’t be taking over from people, but they also won’t be saving us by questioning bad orders.
In science fiction, though, anything can happen. I’ve been working on my book for two years, and just in that time, a wealth of new science fiction stories have used AI to examine life and humanity. I’d like to step into the world of fiction and talk about some of these stories.
Wanderers by Chuck Wendig
When I mention that this novel includes an AI named Black Swan that generates band names, paint colors, weird recipes, and terrible poetry, readers of my AI Weirdness blog posts about band names, paint colors, weird recipes, and terrible poetry will understand why I was excited about this story. The book is epic in every sense of the word (including being about 1000 pages long), with room in it for every terrible thing you could possibly imagine happening during an apocalypse. But it also has moments of humor and tenderness, and I’m tickled that a goofy AI is part of it.
Exit Strategy by Martha Wells
Murderbot is one of my favorite narrators ever, with its snarky self-awareness and totally relatable social anxiety. (Breq, from Ann Leckie’s Ancillary Justice series is another, published too early for my list, but among excellent company on this list by Tansy Rayner Roberts and Rivqa Rafael.) Among the many great things about the Murderbot books is a keen awareness of what it means to have a mind and body that are owned by a corporation and designed for its purposes. Murderbot is aware of the flaws in its original corporate design and is able to do something about it; today’s AI, often designed in ways that are less than competent, or which prioritize corporate profits over their users’ interests, doesn’t have the ability to fix these problems.
Autonomous by Annalee Newitz
As its title indicates, this novel digs deep into questions of autonomy and personhood. In this future vision of North America, the ability to own and hack sentient AI has lead neatly to the ability to own and hack sentient humans. Any future that has human-level AI intelligence will have to deal with these issues head-on, and Autonomous makes them vividly, warningly real in the parallel struggles of enslaved human and AI protagonists.
Children of Ruin by Adrian Tchaikovsky
Second in the same universe as Tchaikovsky’s award-winning Children of Time, Children of Ruin has minds running on all kinds of strange hardware, artificial, biological, and even somewhere in between. It would be spoiling the surprise to reveal the natures of the minds that encounter and try to understand one another; only know that their differences are delightful and their similarities crucial, even as they push the definitions of mortality in their various ways.
“On the Life Cycle of Software Objects” in Exhalation by Ted Chiang
Speaking of mortality! I move now to a short story, published in Ted Chiang’s astonishing collection Exhalation, which asks tough questions about life and mortality, and our responsibilities toward the lifeforms we create. Just recently an update to one of the most common machine learning toolkits, Tensorflow, made a bunch of algorithms suddenly out-of-date, unable to run on modern systems unless someone goes to the trouble to update them. Ted Chiang’s story vividly anticipates this, asking what will happen to virtual lifeforms as their codebase inevitably ages. It’s a realistic look at what would happen if we did make artificial general intelligences, as opposed to the highly-specialized narrow AIs we have today, and a strong argument against doing so.
“Trojan Girl” and “Valedictorian” in How Long Till Black Future Month? by N. K. Jemisin
A pair of short stories in N. K. Jemisin’s collection How Long Till Black Future Month? looks at possible futures for artificial intelligences that have emerged on their own, and their complex relationships with the human world. Essential are compassion and coexistence, yet humans generally don’t make this easy. The AIs in “Trojan Girl” hide from the human world—in a nod to the pervasive problem of algorithmic bias, the earliest and least creative of these hide behind Caucasian avatars, “a human minority who for some reason comprised the majority of images available for sampling in the Amorph.” The AIs in “Valedictorian” are also separated from humanity—but has humanity walled them out, or walled themselves in? If generally-intelligent AI did someday emerge, would we be able and willing to recognize it?
Prey of the Gods by Nicky Drayden
The theme of recognizing personhood also runs through Nicky Drayden’s novel Prey of the Gods, in which a future South Africa must come to terms with various forms of divinity and magic that thread through the history of the land and its people. At the same time that humans and gods are figuring each other out, sentience is spreading like a virus through their personal assistant bots. Most of the humans don’t realize this, treating their bots like inanimate things. We see some of the same issues today. With today’s AI unable to handle many of the tasks we most desperately want it to do, like voicemail transcription or customer service chats, many companies are turning to a hybrid approach that sometimes substitutes in remote human workers when the AIs falter. Of course, these remote workers sometimes end up being treated poorly by customers who don’t know they’re human, echoing the plight of the bots in Prey of the Gods.
“The Low Hum of Her” from Sooner or Later Everything Falls Into the Sea by Sarah Pinsker
Many stories of clockwork creatures are really stories of AI. This vivid, warmhearted story is set in an alternate version of 1940s Europe where a persecuted man builds his daughter an artificial grandmother to replace the Bubbe who has died. What was first a resented replacement becomes a vital comfort, as the family flees house and homeland. It’s a story of adaptation to new circumstances and new modes of being, and echoes the theme of many of these recent AI stories, of seeking humanity in others.
Love, Robot by Margaret Rhee
The last AI story in this list is actually a collection of poetry. While other stories explore AI minds, Love, Robot focuses lavishly on hardware. With its clockwork and circuitry, its wires and its servo motors, the feel is very retro. The poems deal with the beginning, middle, and end of relationships – the feel is tender and quirky, with the NSFW section particularly so.
Personhood, autonomy, weirdness, and human rights—even if the AIs of science fiction are worlds ahead of the AI of today, they are still entirely relevant to the ways we use these technologies. In a world where screens and oceans can separate us from the people we’re interacting with, or where impersonal algorithms can be used to hide biased decisions behind veneers of plausible deniability, science fiction holds important lessons in recognizing the humanity of the people whose lives we touch.