What do you think?
Rate this book
Artificial Whiteness: Politics and Ideology in Artificial Intelligence
Dramatic statements about the promise and peril of artificial intelligence for humanity abound, as an industry of experts claims that AI is poised to reshape nearly every sphere of life. Who profits from the idea that the age of AI has arrived? Why do ideas of AI’s transformative potential keep reappearing in social and political discourse, and how are they linked to broader political agendas?
Yarden Katz reveals the ideology embedded in the concept of artificial intelligence, contending that it both serves and mimics the logic of white supremacy. He demonstrates that understandings of AI, as a field and a technology, have shifted dramatically over time based on the needs of its funders and the professional class that formed around it. From its origins in the Cold War military-industrial complex through its present-day Silicon Valley proselytizers and eager policy analysts, AI has never been simply a technical project enabled by larger data and better computing. Drawing on intimate familiarity with the field and its practices, Katz instead asks us to see how AI reinforces models of knowledge that assume white male superiority and an imperialist worldview. Only by seeing the connection between artificial intelligence and whiteness can we prioritize alternatives to the conception of AI as an all-encompassing technological force.
Bringing together theories of whiteness and race in the humanities and social sciences with a deep understanding of the history and practice of science and computing, Artificial Whiteness is an incisive, urgent critique of the uses of AI as a political tool to uphold social hierarchies.
Yarden Katz reveals the ideology embedded in the concept of artificial intelligence, contending that it both serves and mimics the logic of white supremacy. He demonstrates that understandings of AI, as a field and a technology, have shifted dramatically over time based on the needs of its funders and the professional class that formed around it. From its origins in the Cold War military-industrial complex through its present-day Silicon Valley proselytizers and eager policy analysts, AI has never been simply a technical project enabled by larger data and better computing. Drawing on intimate familiarity with the field and its practices, Katz instead asks us to see how AI reinforces models of knowledge that assume white male superiority and an imperialist worldview. Only by seeing the connection between artificial intelligence and whiteness can we prioritize alternatives to the conception of AI as an all-encompassing technological force.
Bringing together theories of whiteness and race in the humanities and social sciences with a deep understanding of the history and practice of science and computing, Artificial Whiteness is an incisive, urgent critique of the uses of AI as a political tool to uphold social hierarchies.
352 pages, Paperback
Published November 17, 2020
5 people are currently reading
212 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 6 of 6 reviews
August 5, 2020
You see a title that says Artificial Whiteness and you think here is a whole new angle to artificial intelligence (AI) I know nothing about. It will be a great read and change my outlook. But you, and I, would be wrong. With all the reasons to hate and fear AI, racism remains the very least of it. Politics is a yawn, and ideology comes with the customer.
Yarden Katz is a systems biologist at Harvard. His book tries very hard to be about racism, but it fails. The closest he comes is a bookcover criticizing the Japanese for stealing AI intellectual property. It portrays the Statue of Liberty with a female Japanese face, and a robot arm holding the torch.
He tries very hard to link AI labs at universities with the universities’ nefarious investments overseas, particularly agricultural ones. American universities like to buy up land hand it over to agribusiness, cutting out locals. But by this measure, absolutely everything done by or at these universities could be slammed, including Katz himself for taking their money. But AI? Just more roadkill.
The book gets a little more into reality when Katz criticizes the major uses of AI so far – surveillance and policing. These are traditionally racist functions, and predictive policing systems pretty much automatically target black neighborhoods. They prejudice police work even more than it already is.
Staying with the justice system, there is also algorithmic sentencing that tells judges how long to send people away, as well as risk assessment. They are all part of the same disease, and AI is boosting them. He calls AI’s contribution carcerial-positive logic.
But that is really the only concrete example he has. Does that make AI white?
AI is nebulous and amorphous. It was born without a definition and has been bouncing off the bumpers ever since. The military cottoned on to it because of the word intelligence, pouring millions upon millions of dollars into it annually, hoping for some brilliant new advantage for itself. But again, this is nothing new. The military has always poured untold millions into the academic research world as well as private labs. It has given America an unfair advantage it refuses to permit any other country to employ. But white?
He says the precepts and framework of AI research are epistemic forgeries. But forgeries implies there is a different, true way that AI is mimicking illegally. But there’s no evidence of that in the book. AI is stumbling from discovery to discovery just like every new technology before it. So far, not much works. But phone systems are getting “better”, driverless vehicles are almost ready for prime time, and autonomous drones are in the air. Data mining and facial recognition are odious and imperfect, but “improving”. None of those are great for mankind or the planet. But white?
His view is that “AI has drawn on epistemic forgeries to pass as universal what in fact constitutes a white, elite and masculinized perspective. And AI is often presented with overtly racial and gendered imagery, alongside colonialist narratives.” But no words of proof elaborate on any of this.
He also really doesn’t like the marketing of AI. He is enormously critical of proponents talking about “colonizing” space and other planets, as well as having “slaves” in the form of robots that will ease everyone’s burden. This is evidence of whiteness to Katz.
It turns out that Katz has a different interpretation of white. When he can’t show true racism beyond the Japanese Liberty lady, he adds imperialist and capitalist to his cause. Ever the extreme leftist, he says everything AI does is to further imperial and capital aggressions by the USA. Yes. True. But then, who is the customer paying for all this? How could anyone be shocked that imperialism and capitalism are beefing up on AI innovations? Google did not buy Deep Mind as a playground for frisky engineers. It is making a further fortune off the military. Welcome to the real world.
Halfway through, I had given up on any real proof of concept in the book. But Katz kept insisting he had proven his claims earlier in the book, and so was entitled to move on from there. Which was infuriating. Artificial Whiteness gave me nothing to cite, nothing to change my mind, and nothing to add to my views on race.
He goes off on tangents trying to present rationales (epistemic forgeries?) of his own, using papers from other scientists in other disciplines. The goal is to develop an AI framework that isn’t white. Nothing comes of it.
One last try: “Whiteness gets its significance, and its changing shape, only from the need to maintain the relations of power. AI reproduces this quality of whiteness.”
Nope.
David Wineberg
Yarden Katz is a systems biologist at Harvard. His book tries very hard to be about racism, but it fails. The closest he comes is a bookcover criticizing the Japanese for stealing AI intellectual property. It portrays the Statue of Liberty with a female Japanese face, and a robot arm holding the torch.
He tries very hard to link AI labs at universities with the universities’ nefarious investments overseas, particularly agricultural ones. American universities like to buy up land hand it over to agribusiness, cutting out locals. But by this measure, absolutely everything done by or at these universities could be slammed, including Katz himself for taking their money. But AI? Just more roadkill.
The book gets a little more into reality when Katz criticizes the major uses of AI so far – surveillance and policing. These are traditionally racist functions, and predictive policing systems pretty much automatically target black neighborhoods. They prejudice police work even more than it already is.
Staying with the justice system, there is also algorithmic sentencing that tells judges how long to send people away, as well as risk assessment. They are all part of the same disease, and AI is boosting them. He calls AI’s contribution carcerial-positive logic.
But that is really the only concrete example he has. Does that make AI white?
AI is nebulous and amorphous. It was born without a definition and has been bouncing off the bumpers ever since. The military cottoned on to it because of the word intelligence, pouring millions upon millions of dollars into it annually, hoping for some brilliant new advantage for itself. But again, this is nothing new. The military has always poured untold millions into the academic research world as well as private labs. It has given America an unfair advantage it refuses to permit any other country to employ. But white?
He says the precepts and framework of AI research are epistemic forgeries. But forgeries implies there is a different, true way that AI is mimicking illegally. But there’s no evidence of that in the book. AI is stumbling from discovery to discovery just like every new technology before it. So far, not much works. But phone systems are getting “better”, driverless vehicles are almost ready for prime time, and autonomous drones are in the air. Data mining and facial recognition are odious and imperfect, but “improving”. None of those are great for mankind or the planet. But white?
His view is that “AI has drawn on epistemic forgeries to pass as universal what in fact constitutes a white, elite and masculinized perspective. And AI is often presented with overtly racial and gendered imagery, alongside colonialist narratives.” But no words of proof elaborate on any of this.
He also really doesn’t like the marketing of AI. He is enormously critical of proponents talking about “colonizing” space and other planets, as well as having “slaves” in the form of robots that will ease everyone’s burden. This is evidence of whiteness to Katz.
It turns out that Katz has a different interpretation of white. When he can’t show true racism beyond the Japanese Liberty lady, he adds imperialist and capitalist to his cause. Ever the extreme leftist, he says everything AI does is to further imperial and capital aggressions by the USA. Yes. True. But then, who is the customer paying for all this? How could anyone be shocked that imperialism and capitalism are beefing up on AI innovations? Google did not buy Deep Mind as a playground for frisky engineers. It is making a further fortune off the military. Welcome to the real world.
Halfway through, I had given up on any real proof of concept in the book. But Katz kept insisting he had proven his claims earlier in the book, and so was entitled to move on from there. Which was infuriating. Artificial Whiteness gave me nothing to cite, nothing to change my mind, and nothing to add to my views on race.
He goes off on tangents trying to present rationales (epistemic forgeries?) of his own, using papers from other scientists in other disciplines. The goal is to develop an AI framework that isn’t white. Nothing comes of it.
One last try: “Whiteness gets its significance, and its changing shape, only from the need to maintain the relations of power. AI reproduces this quality of whiteness.”
Nope.
David Wineberg
July 22, 2020
The book, "Artificial Whiteness," by Yarden Katz has been all over the internet with the current racial tensions in the United States. Written by someone who identifies as having grown up white, Katz helps challenge one's thoughts and beliefs about race and the definition of racism. With institutional racism garnering more notoriety, this book truly helped me see other lives in a different light. I found myself constantly processing the information found within to adjust my own beliefs.
I would never expect a person to completely change their belief system or core values based on one book, but Katz helps the reader fill in the area of our lives that most have never had to poke and prod. One concern I would have is that there are certainly those who may read this book and immediately go on the defensive. If that is you, then try to give it a chance. If you can't, that's fine, Katz presents a very heavy topic in a very challenging manner.
I would never expect a person to completely change their belief system or core values based on one book, but Katz helps the reader fill in the area of our lives that most have never had to poke and prod. One concern I would have is that there are certainly those who may read this book and immediately go on the defensive. If that is you, then try to give it a chance. If you can't, that's fine, Katz presents a very heavy topic in a very challenging manner.
February 10, 2022
Each branch of United States government has a long history of reflecting, reinforcing, and perpetuating white supremacist norms. Innovations in technology continue to abet punitive, carceral outcomes, via, for instance, facial recognition systems programmed to interpret data points related to such features as skin tone in terms of “likelihood,” which is programmed into the software by programmers making assumptions (“The likelihood of what?” one might ask), most of which the programmers never examine.
In Artificial Whiteness: Politics and Ideology in Artificial Intelligence, author Yarden Katz’s examples of racism form merely one of several ominous strands in the historical makeup and development of things called “AI.” So, the title is a bit off on stating the book’s actual focus, which is how ethnic (sub)groups with governmental power monopolies use things called “AI” to reinforce the status quo. This problem is as old as the formation of clans and tribes: keep my enemies (real and imagined) at bay and let my power and riches increase.
Although Katz uses “whiteness” as the locus of a particular evil, the evidence he presents shows that problems in AI created by social biases inadvertently and purposely written into computer applications are the social norms of the dominant ethnic makeup of the dominant force in power. For instance, some Chinese AI firms claim to sell facial recognition systems specifically design to identify Uighurs—who are then presumably rounded up by the police or military and sent away to be “re-educated.” (The majority of Chinese, who also have most of the nation’s money and power, are Han and Buddhist rather than Uighur and Islamic.)
Ethical problems related to autonomous vehicles also illustrate the limitations to software programs that cannot avoid embedding predominate cultural norms. Attitudes toward jaywalking, for instance, seem to vary around the world. Something the Singaporeans and Norwegians might agree on, for instance, is that jaywalkers get what they deserve: Walk where the law has determined you should not and pay the price. The same attitude would not work in the U.S. Will separate programs be written for vehicles based on the country in which they’re sold? Or will a uniform, international “norm” be imposed upon unappreciative peoples?
Katz argues that the current resurgence in AI funding from governments and corporations has added the adjective “ethical” to gain university credibility and broaden the base of buy-in from knowledge workers. No surprise, significant amounts of funding come from corporations who routinely engage in acts of civil and ecological shenanigans. But that funding for AI programs comes from racist, ecological evildoers does not mean that the persons writing code for different overt purposes are of similar moral disfigurement. (There are programmers who would be thrilled to think that their skills were being used to help the environment.)
Artificial Whiteness is predicated on two concepts, each, as Katz notes, with critically unstable meanings: intelligence and race, terms that have historically always lacked agreed-upon definitions. Katz shows that definitions of “white” are as elusive as those for “Black,” and yet his title misses the broader point that any culture funded and ruled by one ethnic group will tend to project its assumptions about others onto its notions of normality, which are also assumed universal. Katz notes that in every iteration of AI over the past 80 years, “we find the practice of situating AI within capitalist and imperialist agendas”: that clause represents the full scope of his project. No dominant entho-political bloc wants to be economically and culturally beholden to an xeno-overlord or the “whims” of the “enemy(ies) within.”
In addition to questionable practices that produce adverse economic and legal outcomes for Blacks in the U.S. are those implied by the nascent AI wars among the most powerful nations. These wars are battles both for intellectual-property primacy (and thus national and international economic prowess) and political dominance (national and international), often fought out in the marketplace. The U.S. government’s argument with the Chinese tech firm Huawei (which manufactures and sells a wide range of consumer electronics and telecommunications products around the world) is that Huawei was collecting demographic data from U.S. and European users on behalf of the Chinese government. Few in the U.S. complain, however, when Microsoft, Facebook, and Google do the same for our government.
The ability to rid programs of biases and assumptions has not yet been realized. The sciences routinely operate with incomplete information, acting “as if” their models are complete, adjusting their models until predictions and outcomes match, or until the problem is better defined. (For a philosophically technical argument on the role of “as if” assumptions in and out of the sciences, see Kwame Appiah’s slim book, As If.) Prominent AI critics, such as Hubert Dreyfuss, have not had much more success in tackling the problem than have AI advocates (who tend to see biases in programming as mere technical matters to overcome.
That “intelligence” is as definable as “pornography,” means that AI advocates will continue chasing not a moving target but a darting chimera. In the meantime, the conditions needed for improving the quality of life of consumers, citizens, and nations will be largely ignored.
In Artificial Whiteness: Politics and Ideology in Artificial Intelligence, author Yarden Katz’s examples of racism form merely one of several ominous strands in the historical makeup and development of things called “AI.” So, the title is a bit off on stating the book’s actual focus, which is how ethnic (sub)groups with governmental power monopolies use things called “AI” to reinforce the status quo. This problem is as old as the formation of clans and tribes: keep my enemies (real and imagined) at bay and let my power and riches increase.
Although Katz uses “whiteness” as the locus of a particular evil, the evidence he presents shows that problems in AI created by social biases inadvertently and purposely written into computer applications are the social norms of the dominant ethnic makeup of the dominant force in power. For instance, some Chinese AI firms claim to sell facial recognition systems specifically design to identify Uighurs—who are then presumably rounded up by the police or military and sent away to be “re-educated.” (The majority of Chinese, who also have most of the nation’s money and power, are Han and Buddhist rather than Uighur and Islamic.)
Ethical problems related to autonomous vehicles also illustrate the limitations to software programs that cannot avoid embedding predominate cultural norms. Attitudes toward jaywalking, for instance, seem to vary around the world. Something the Singaporeans and Norwegians might agree on, for instance, is that jaywalkers get what they deserve: Walk where the law has determined you should not and pay the price. The same attitude would not work in the U.S. Will separate programs be written for vehicles based on the country in which they’re sold? Or will a uniform, international “norm” be imposed upon unappreciative peoples?
Katz argues that the current resurgence in AI funding from governments and corporations has added the adjective “ethical” to gain university credibility and broaden the base of buy-in from knowledge workers. No surprise, significant amounts of funding come from corporations who routinely engage in acts of civil and ecological shenanigans. But that funding for AI programs comes from racist, ecological evildoers does not mean that the persons writing code for different overt purposes are of similar moral disfigurement. (There are programmers who would be thrilled to think that their skills were being used to help the environment.)
Artificial Whiteness is predicated on two concepts, each, as Katz notes, with critically unstable meanings: intelligence and race, terms that have historically always lacked agreed-upon definitions. Katz shows that definitions of “white” are as elusive as those for “Black,” and yet his title misses the broader point that any culture funded and ruled by one ethnic group will tend to project its assumptions about others onto its notions of normality, which are also assumed universal. Katz notes that in every iteration of AI over the past 80 years, “we find the practice of situating AI within capitalist and imperialist agendas”: that clause represents the full scope of his project. No dominant entho-political bloc wants to be economically and culturally beholden to an xeno-overlord or the “whims” of the “enemy(ies) within.”
In addition to questionable practices that produce adverse economic and legal outcomes for Blacks in the U.S. are those implied by the nascent AI wars among the most powerful nations. These wars are battles both for intellectual-property primacy (and thus national and international economic prowess) and political dominance (national and international), often fought out in the marketplace. The U.S. government’s argument with the Chinese tech firm Huawei (which manufactures and sells a wide range of consumer electronics and telecommunications products around the world) is that Huawei was collecting demographic data from U.S. and European users on behalf of the Chinese government. Few in the U.S. complain, however, when Microsoft, Facebook, and Google do the same for our government.
The ability to rid programs of biases and assumptions has not yet been realized. The sciences routinely operate with incomplete information, acting “as if” their models are complete, adjusting their models until predictions and outcomes match, or until the problem is better defined. (For a philosophically technical argument on the role of “as if” assumptions in and out of the sciences, see Kwame Appiah’s slim book, As If.) Prominent AI critics, such as Hubert Dreyfuss, have not had much more success in tackling the problem than have AI advocates (who tend to see biases in programming as mere technical matters to overcome.
That “intelligence” is as definable as “pornography,” means that AI advocates will continue chasing not a moving target but a darting chimera. In the meantime, the conditions needed for improving the quality of life of consumers, citizens, and nations will be largely ignored.
December 1, 2020
Don’t let the title scare you off. I had no background prior to reading.
Excellent discussion of historical development of AI by the American/western defense community for specific tasks and the groundbreaking people who did it. The caution is then that if you have a group of almost exclusively white, male, university educated, middle class, military/defense focused members the resulting product reflect those perspectives. Not good or bad, but beware of porting the application(s). Unclear to this reader, however, how much of more modern “AI” work borrows from this earlier work, or how much wider/“modern” those tools’ perspectives are, or international non-western work is influencing.
Excellent discussion of historical development of AI by the American/western defense community for specific tasks and the groundbreaking people who did it. The caution is then that if you have a group of almost exclusively white, male, university educated, middle class, military/defense focused members the resulting product reflect those perspectives. Not good or bad, but beware of porting the application(s). Unclear to this reader, however, how much of more modern “AI” work borrows from this earlier work, or how much wider/“modern” those tools’ perspectives are, or international non-western work is influencing.
April 21, 2021
This is an interesting take on AI and how it relates to contemporary politics. It traces the origins of AI from military contexts and argues that AI has always served regressive goals, and in particular, could be seen as a neoliberal technology aligned with racism (whiteness, as the title says). The author draws from several books and thoughts in the area, and provides a much worthy political perspective on AI. While I found most of the arguments thoroughly agreeable, some aspects such as the cynical perspectives on the state and structures such as the police force seems to be heavily drawn based on some particular US contexts, and may not be applicable everywhere. The central thesis of the work is that AI has been quite nebulous in its make up, with people using AI to refer to several things, but one consistent thread is that it has remained loyal to imperialist, racist and capitalist political projects.
Want to read
June 22, 2020Note: I accessed a digital review copy of this book through Edelweiss.
Displaying 1 - 6 of 6 reviews