“There is criticism to which one responds, other criticism to which one replies. Wrongly perhaps. Why not lend an equally attentive ear to incomprehension triviality, ignorance, or bad faith? Why reject these as so many incidents, regrettable for family honor? Is one correct in believing them inessential to the activity of criticism? I wonder if […]

Kate Crawford’s “Atlas of AI” presents a unique perspective on artificial intelligence (AI). Rather than focusing solely on its technical aspects, she delves into the often-overlooked societal implications of AI, including its environmental footprint, political consequences, and ethical dilemmas.

The Book in Three Sentences

• “Atlas of AI” by Kate Crawford is an insightful exploration of the real-world impact of artificial intelligence, presenting a comprehensive view that goes beyond algorithms and computations.
• The book delves into AI’s environmental, economic, political, and ethical implications.
• Crawford challenges the prevailing view of AI as neutral and objective by highlighting its inherent biases and inequalities.

Extended Summary

The book starts with an exploration of the physical infrastructure underpinning AI systems. It looks at everything from data centers to lithium mines, highlighting the environmental cost associated with these technologies. Crawford also examines how data is collected and used in AI development, drawing attention to privacy and consent issues.

The second part of “Atlas of AI” looks at how these technologies are used in society. Crawford explores how AI systems can reinforce existing biases and inequalities due to their design and training data. She also discusses how powerful entities can use these technologies for surveillance and control.

In the book’s final part, Crawford presents some possible paths forward for more equitable and sustainable use of AI. She argues for greater transparency in how these systems are developed and deployed as well as more democratic control over their use.

Throughout “Atlas of AI”, Kate Crawford paints a comprehensive picture not just about what artificial intelligence is but also about what it does to our world – environmentally, politically, economically, ethically.

Key Points

• Artificial Intelligence has significant environmental costs due to its reliance on resource-intensive infrastructure.
• AI data collection often involves issues around privacy breaches and lack of informed consent.
• AI systems can reinforce existing societal biases due to their design.
• The use of AI for surveillance and control by powerful entities is concerning.
• Greater transparency and democratic control are needed to develop and deploy AI.

Who Should Read

“Atlas of AI” is a must-read for anyone interested in the broader implications of artificial intelligence. It would appeal to tech enthusiasts, data scientists, policymakers, sociologists, or anyone interested in understanding the far-reaching impact of AI on society. This book is also an excellent resource for information technology, computer science, or digital ethics students.

About the Author

Kate Crawford is a leading scholar on the social implications of artificial intelligence. She has spent over a decade studying the social implications of data systems, machine learning, and artificial intelligence. She is a research professor at the University of Southern California in Los Angeles, a senior principal researcher at Microsoft Research in New York, and was the inaugural Visiting Chair in AI and Justice at the École Normale Supérieure in Paris.

Further Reading

• Kate Crawford’s website
• Related books: “Weapons of Math Destruction” by Cathy O’Neil; “Automating Inequality” by Virginia Eubanks
• Founded by Kate Crawford and Meredith Whittaker, the AI Now Initiative is an independent, interdisciplinary research initiative working to understand AI’s social and economic implications.

The post Unmasking AI: How Intelligence is Molded and the Traps it Conceals – A Review of ‘Atlas of AI’ first appeared on Dr. Ian O'Byrne.

Last year, I received an email asking if I would write an introductory essay for the Tor Essentials reissue of Kim Stanley Robinson’s novel Icehenge. It took me approximately thirty seconds to convince myself that this was not some kind of hallucination, and another three or four to type YES! OF COURSE!!! and hit reply. It was the most delightful email I’d gotten in years. Stan is now a friend, but the request had its origins in a conversation years before I’d met him. At least a decade ago, Patrick Nielsen Hayden and I were chatting about his books, and I said that Icehenge was both (a) his best novel in my opinion, and (b) criminally under-appreciated. Patrick, who recently stepped down as editor-in-chief at Tor, somehow remembered this, and asked me to pick up on the notion many years later. So how could I do anything but seize the chance?

The book is out in June, and I’ll have more to say then. When I was writing the introduction, I talked to Stan about how he had come to write Icehenge, and what it had meant to him as a writer. There was a lot about modernism! When the British Science Fiction Association’s journal, Vector put out a call for submissions on SF and modernism, I made inquiries, and they said they’d love to publish the conversation. A cleaned up version is available in the new issue (which has a ton of other great content). It’s published under Creative Commons, as is pretty well everything that Vector publishes, so I’m republishing it here. I simply can’t say how happy I am about all of this, and how much I’m looking forward to the book itself – out in just a couple of months.

There are some spoilers in the below – so if you prefer to wait for the book, wait for the book!

Henry Farrell talks to Kim Stanley Robinson

HF – How did you come to write Icehenge?

KSR – When I was a kid I loved stories about archeology, including pseudo-archaeology. There were quite a few fake archaeologies about when people first got to the Americas – the Phoenicians; St. Brendan; the Welsh – I read all these with huge pleasure. Everybody got to America, it seemed. I was perhaps 10 or 12.  Whether I was making any distinctions as to whether these were real or not, I’m not sure.  I just loved them so much as stories.

One of the stories was about the Kensington Stone, which was discovered in Minnesota in 1898. A Swedish American farmer found a piece of stone, with runes carved onto it saying more or less ‘we’re out here, the natives are killing us, mother Mary save us.’ It’s actually quite moving as a prose poem or last testament.  It was dated to 1362, and Hjalmar Holand, a scientist from Chicago, decided that this was a genuine stone and spent his career trying to find an expedition from that era that would explain it. He found that a pope of that time had asked the Danes to find out what had happened to the church in Greenland, and an expedition had gone off to do so, and never was heard of again. Hjalmar Holand said these people got to Greenland, found it abandoned, went up the Hudson Bay looking for the missing Greenlanders, then went up one of the rivers leading southwest, and in two weeks were in the middle of Minnesota, where the locals killed them with arrows.

You can still go to Kensington, Minnesota, where there is a 10 ton, 20 foot high copy of the stone, which was just a little thing. The original stone was displayed in the Smithsonian for a while as evidence of Vikings in America, but many experts in runes were dubious from the start about the language on the stone. They thought it was all wrong, but Holand defended it until he died. A couple of years later, someone noticed that all the runes were multiples of one inch long, suggesting it had been carved with a one inch chisel. It turned out that the Swedish farmer who found it was a country intellectual, who wanted to bother the brains of the learned, as he once put it. He’s almost certainly the guy who did it.  But since Holand had died, he didn’t see it being removed from the museum.

At that point I began to get interested in hoaxes as such.  The Vinland map was thought to be a hoax, and then was thought to be real, and now we think it’s a hoax again. I was interested in how hoaxes got found out, what the methodologies are and so on. Then in the midst of my reading, they found a real Viking site in Newfoundland at L’Anse aux Meadows. At that point I was 11 years old, so that dates my reading of this stuff.  The news was announced in National Geographic, and I was thrilled.

So, when I became a science fiction writer, I was wondering what kind of stories to tell.  I was young, nothing in particular had happened to me, so I was often telling stories out of books. Then a friend sent me an article in Forbes magazine saying that we could live up to 500 years if we could repair our DNA when it got damaged. I thought, Wow, what if Hjalmar Holand had lived a little longer, and thus saw his entire life’s work knocked down like a house of cards—what would he have said? How would he have felt?  And I thought that would make a story.

I wrote it through my mid-twenties.   At that time I was impressed by Ford Madox Ford and modernism’s chronological tricks. I thought I’d write the story with the scenes all out of order. It would be a jumble you’d have to unsort, like the first chapter of The Sound and the Fury. So I became Benjy in effect; it was like throwing scenes up in the air and collecting them where they fell.

I submitted the story to Damon Knight. He had already bought three stories from me, including my first. He was my editor and my patron, and my teacher. If I owe my sixties as a writer to Tim Holman, as I do, I owe my twenties to Damon Knight.

So he read this novella and said, “Stan, have you listed the scenes for yourself?” No, I hadn’t done that. He suggested I number the scenes and then make a list of them both chronologically and in their order in the text; then ask myself why these two were different, and make a pattern of the difference, if I still wanted it. Maybe Edmond Doya was in a present time, he suggested, but remembering things in an order one could see had a logic to it.

Damon also said that the little ten page version of Emma’s journal was undercooked and didn’t need to be there. I could write it up later on as a separate novella, he suggested.  And he said, if you get this all in a good order, and you make a good revision more generally, I will buy this. Even though it was 100 pages long, maybe 25,000 words.

So I did all that, and he bought the story, called “On the North Pole of Pluto.” It was the last story in his Orbit series, in Orbit 21.   The story’s final words were the final words of the Orbit series, and I am sure Damon did that on purpose. He was my teacher and advocate, and a lovely man. Well, he would laugh at that description; he was a great man.

Several years later I sold The Wild Shore to Ace Books, and they said: we’d like more from you. I said, I could do a three-novella novel like Gene Wolfe’s The Fifth Head of Cerberus, which had completely blown my mind. John Clute calls this form the “fix up,” but that’s not a good name.  Many novels are made of disparate stories; no one calls The Sound and the Fury a fix-up.  So I thought I could bulk up the Emma Weil story, and if I put Nederland’s story in the middle, it would work very well as a trio.

The Emma Weil story was bought by Ed Ferman at Fantasy & Science Fiction as an independent novella. You didn’t have to know anything about the aftermath to appreciate it, although Gardner Dozois noted that the story has a peculiar ending that jags off in a new direction. But whatever—it was published, and it got on the Hugo ballot, which was maybe a first for me.

My regular editor at Ace, Beth Meacham (Terry Carr was a guest curator), said to my idea for a book, “fine, good idea, write the middle novella.” By that time I was married to Lisa, and things were stable for the first time in my life – this was around 1983, in Davis. So I sat down and considered the problem of my Hjalmar Nederland, who had to be immensely old. His section of course had to be a first person narrative, like the other two. Here I began to recall the Anglo-Irish novelist Joyce Cary, whose books I loved.

His two trilogies are all first person narratives, and constructed such that first you get the woman’s story, then stories from two men closely involved with that woman. The “First Trilogy,” is Herself Surprised, To Be a Pilgrim, and The Horse’s Mouth.  The narrator of To Be a Pilgrim is an ancient, crabby, conservative, often unpleasant lawyer, Tom Wilcher, who is in love with Sara Monday, who is also involved with a crazy artist, Gulley Jimson. Wilcher and Gulley Jimson hardly cross paths, but they both have interesting stories that cast a light on Sara. The “Second Trilogy” has a very similar construction. They’re all great novels, and in six very different voices.

So I thought to myself, I’ve got that kind of pattern here. Emma comes first, then both Hjalmar Nederland and and Edmond Doya are fascinated by her, even though they never meet her. It’s a literary fascination, but that fit me, because these were to some extent books out of books; bookish stories by bookish me. Books about reading, in fact, and how that can shape a life, become a quest. Books about falling in love with characters.

So I gave it a try, and I was very surprised to find that Nederland’s first person narrative, stuck in the middle and overdetermined in many ways, was the first time that I wrote something better than I felt I could write. When it was done and I read it over, I was startled. I thought “where did that come from?”

This crabby old man had tapped into something I had never tapped into before. Up till then, I had felt that it was tough to be a writer when you don’t have a gift for it, but only have the desire. That was a powerful feeling through through my twenties.  But by the time I wrote the Nederland novella I was a bit older, I had lived some. And in the process of filling a gap in my story, something had happened.  It was quite startling to me.

HF  – I would have guessed that the Nederland novella would have been the first thing to have been written, and that the other two were ancillary.

In fact, it was filling in a gap. That in itself was interesting, to see that filling a gap in the larger story was how something had popped. It implied that all kinds of things were possible. So there’s a kind of before and after in my interior life as a writer, with Icehenge marking the division, or really, that middle novella.

The shift is obscured in my bibliography, because I then rewrote an older novel of mine, which became The Memory of Whiteness.  That one was written by a twenty year old collaborating with a thirty-three year old, which I think shows. It’s weird and unsatisfactory to me, though some readers have enjoyed it.  But I see problems. Same with The Wild Shore.  But really, in this regard, my opinions don’t matter. The books are out there, and readers pretty often bring them to life.

HF – And so was it the character of Hjalmar Nederland that brought Icehenge to life?

KSR – It was. Obviously, that was Hjalmar Holand, as a starting point. I played with the name for anyone who saw the connection to the Kensington Stone. Then there was Tom Wilcher in the Cary novel, a crabby old man who is consistently wrong and irritating. First person narratives compel the reader to sympathize with the person talking to them, even if they are quite off-putting, like in Dostoevsky’s Notes from Underground. They aren’t just unreliable narrators; you, the reader can see perfectly well that they are wrong about everything they say about themselves; they’re self deceptive, they’re lying to themselves over and over, or are just wrong and they don’t even know that. They don’t lie to the reader, because they’ve already lied to themselves. This is a beautiful trick of characterization.

And then, another question was, what’s it like to be 350 years old? At that time I was around 32 years old, and had no idea what it was like to be even ordinarily old, like I am now, much less what it was like to be 350 years old. So one thing I did was to read Wallace Stevens, who often seems 300 years old, or some immense weary age. He’s a great poet, though often close to incomprehensible, and reading him helped me in thinking about age, in expressing the feeling of it.

This is all so literary. But I had lived some life at that point, and I was also beginning to lose what felt like a photographic memory of my life. Through my twenties I felt I could recall every day of my life, from about maybe age three up to the present, but as I got older that feeling began to go away. I began to realize I wasn’t going to have that kind of photographic memory for much longer. And I definitely don’t now. So maybe that was the authentic feeling I was writing about in that book.

HF – So it was a personal experience as much as anything that gave rise to the focus on the frailties of memory?

Yes, I think so. What had felt like nearly photographic memory had begun to fuzzball out into “wait, do I really remember that?” Or worse, just the feeling that I was forgetting things.  And I was thinking about memory a lot, and the way that you remember best what moves you most. Emotions spike memories into your brain. This is obvious, and yet we don’t have the brain science to be precise about how it happens. I was reading books on memory at the time, quite intensively.  They were already beginning to understand that you can’t find a spot in the brain where memories are held, that the storage is somehow holistic or widely dispersed. It’s somewhere but nowhere. That struck me as interesting.

And there’s the desire to be a fiction writer that hits when you’re young, when you haven’t really done anything but read. You have to write about being young and wanting things, and you have to write about your reading. Also, then comes a time when you can feel yourself aging out of youth, which is the first shock of aging, and has that power of coming first.

The revolution stuff in Icehenge might have come out of the 1970s, with the anti Vietnam demonstrations that I saw. I had a quite low draft number, 89.  At UCSD, we had Marcuse, Angela Davis, my teacher Fred Jameson. The 60s is how we name that time, but for me it was 1970 to 1974. That mild offset is interesting, though maybe not crucial, in terms of the feel of that era. It was crazy until we got out of Vietnam and Nixon was gone; it felt revolutionary. We really didn’t know what would happen next.  Leaders were getting assassinated, drafted soldiers were dying, it all felt centrifugal, or that things were simply blowing up.

Also, because of Jameson, I had a Marxist education, which included lots of thinking about what a revolution was.  And I had always been intently interested in the American Revolution as a kid. I was a military history kid, actually, one of those boys who move soldiers around on maps.

Lastly, I was beginning to think about what my history of Mars needed to be. By then, the Viking data were available, and I was convinced that this was a story space for me. I had written “Exploring Fossil Canyon,” and I was gearing up for a longer Mars story—I didn’t know how long.

But I knew I was going to do something. So Nederland’s last walkabout on the Martian surface, and all of the epigraphs at the beginning of Nederland’s chapters, were part of that.  The main source was Michael Carr’s The Surface of Mars. This was the first good general explication of the Viking data, with early geological explanations and so on. So I was setting out an early version of a story that would describe Martian history.

HF – When Nederland is wandering on Mars, he finds what seems to be a cave with a Grecian temple. Later, there is a bit in Red Mars where they build a temple. Which book refers to which?

KSR – The second one is a re-use of the first. It seemed to me such a beautiful image. I’ve always been happy to plagiarize myself, because it isn’t plagiarism. I like to recycle ideas. Another example is the city Terminator on Mercury, which came out of an early short story and then appears in The Memory of Whiteness, and much later in 2312. The beauty of that Martian image, of a temple rising out of the ground and then descending mysteriously back into it, pleased me.

HF – It’s a lovely image.

It was good to know that Red Mars was ending with something strong. In fact, that was my initial idea for the last moment of Red Mars, that temple sinking back into the ground; that was going to be the end of volume one. But my editor Jane Johnson said no, we need a little uptick there emotionally, and a sort of teaser to get people to go on to Green Mars. So that final section in italics, where Maya leads them to the refuge in the South Pole, was tacked on. The original ending would have been an Icehenge riff with the mysterious temple. But Jane was right about that—the extra scene is important, and I like it. A better ending by far.

HF – It gives people that breadcrumb trail to the next.

KSR – Jane was right. She’s a great editor. The extra scene allowed me to show the reunion between Peter and Ann and Simon, which wouldn’t have been possible without it. It’s the right ending for sure, but I grumbled at first. This was all a long time ago, but I haven’t forgotten it.  It was important for me, a big lesson. But I haven’t thought of it in a long time.

HF – So I see the Mars Trilogy as being all about group dynamics and geology (or areology), but Icehenge is a chain of individual relationships, of mistakes, of misprision, and people reading into others what they want to read.

KSR – That’s true. The change in form changes the emphases, inevitably. Icehenge is three first person stories, and none of the Mars trilogy is in first person, it’s all third person limited, or free indirect, what have you.

I’m so impressed by Joyce Cary’s books. First person narratives are immensely revealing of character. You do characterization at the same time as plot because one character is telling you the story, and the way they tell it reveals them. I loved that when I began writing, then I shifted over into third person.  I needed that for other things I wanted to do.

Now, if and when I get back to fiction, I’d like to try first person again. The eyewitness accounts in The Ministry for the Future were my favorite parts, and now I want to write more things like that.

HF – Usually, when we think about revolutions or revolutionary visions, it’s visions of the future. So this is a vision of the past, and an ambiguity in the past, which perhaps opens up possibilities. Did that emerge organically from these different elements or in some other way?

KSR – I think I was pondering the idea that the winners write the history books.  Or it could just be that I was following the logic of a story. And The Fifth Head of Cerberus has this theme as well, and I loved that book.  So, I’ve got someone hundreds of years old, so his memory problems are crucial. What if you couldn’t remember what was significant in your life? And what if there had been some major event that you had seen that had been suppressed in the historical record, and then it came back to you as an abreaction? It seemed nicely dramatic.

HF – So Wolfe’s Soldier of the Mist was in 1986.

KSR – That was later, but I laughed when I saw what Wolfe had done with this theme of memory.  It’s a very simple structural flip: his Severian can remember everything, but his Latro can’t remember anything.  It’s just a joke in a way, a structural turn in his own evolution of story ideas, as with his Doctor Death series.  For me, I can’t remember precisely, but I think it was more the exigencies of forcing myself to make a story out of elements that were somewhat disparate. I wanted to make it cohere, and then it made me do what it wanted, in order to work.

HF – Icehenge, like many of Wolfe’s stories, is quite ambiguous. It could be that Emma herself created a fake. It could be that this too is another imposed narrative that will lead to later disenchantment.  I’m not going to ask you what the actual resolution, if any, was – but do you have your own personal resolution of the story, or did you set up the story as a maze of ambiguities in which readers can arrive at their own explanation?

KSR – More the latter. I wanted the reader to be forced to come up with their own explanation, but as Jameson told me when he read the book in draft, ambiguity is all very well, but you want it to be a clear ambiguity.  In other words, there should be two or three clearly articulated alternative explanations, and the clues in the text will support all of these explanations, while no clue will deny or contradict any of them. Jameson told me this in a letter or conversation, and I thought, Yes of course that’s right. Because when I was young I read a ton of detective stories, as he did, and I understood the principle involved; but it was Jameson who reminded me.  When young I loved most John Dickson Carr and his locked room mysteries, which were like jigsaw puzzles in prose.  Ellery Queen also. I was surprised recently when Jameson told me he too had been a fan of Carr and Queen. The locked-room mystery—is it dialectical, does it lead to the Greimas rectangle etc.? Maybe, but who knows—it was just a pleasure to learn, given that we were young readers about twenty years apart. When young I was in a bubble world of books, detached from my context in many ways, including time itself. Maybe English majors are always like that.

Anyway, The Fifth Head of Cerberus was a big influence on me in this regard too.  Did John V. Marsch write the middle story of the trio before he was taken over by an abo, or did the abo write it after he took over John V. Marsch? Wolfe sets that problem up for the reader, but you cannot unravel it in a definitive way. I studied that book in great detail and became certain it was a deliberate ambiguity. I loved The Fifth Head of Cerberus for that, and was imitating it.

For me, the most likely solution is that Carolyn Holmes made it all up, and also set all the clues in the real world. She was a hoaxer like Olof Ohman, the Swedish farmer. She wanted to tell a story, and she wanted people to think it was a real story, God knows why. In the very last pages, they find an egg-shaped cave underneath one of the pillars of ice at Icehenge itself.  So the monument has male and female symbols, also embryo symbols, maybe that was about the birth of a story. But then also, I did include the clue that can’t be denied, that skews toward just one explanation, which would be the hoax explanation.  Because the ceramic lining the egg-shaped cave isn’t old enough, as determined by thermal luminescence dating, to be from the supposed builders of the ice circle on the surface. So Icehenge isn’t old enough to have been made by Davydov’s expedition, as dated by Emma’s journal:  so someone else did it.

HF – But you leave some ambiguity. It might be Emma, having decided to commemorate the actual expedition.

KSR – That’s true! That’s a beautiful thought, which I had forgotten. But yes— I recall now that I thought then, Emma’s still got to be in the Solar System somewhere.  Maybe she’s taken the name Caroline Holmes.  And maybe she wants to tell the story without telling the story, of Davydov and his crew. She wants people to know about what happened to these poor people who took off in a starship to nowhere, because that starship was jury-rigged and it was not going to work. No, that’s an even better explanation. I had just forgotten it.

At the time I was solid in my mind, I think. It’s either Emma, or it’s a woman making up Emma. But at a certain point, the meta question is: does it matter? Because it’s stories about stories about stories. Could you care intensely about Emma, knowing that someone else had made her up? How is that different from caring about Elizabeth and Darcy, or any other fictional character that you care about intensely that, you know, is quite obviously made up?  And which in fact you yourself have collaborated in making up, since it all happens in the reader’s mind?  It was a little commentary on the power of fictional characters.  It’s a book out of books; but our lives are out of books too.

It’s the typical bind of the third-culture kid: too much of one thing, not enough of the oth­er, not enough, not enough, not enough …