What do you think?
Rate this book
血汗AI：為人工智慧提供動力的隱性人類勞工
你所不知道的AI
歐洲殖民主義遺毒＋全球化結構的不平等
【全面剖析：新興產業背後不為人知的勞動真相】
如果人工智慧被理解為一台擷取機器，那麼我們就是原料。
◆內容簡介
對娜歐米・克萊恩（Naomi Klein）和妮可・柏勒斯（Nicole Perlroth）的讀者來說，這本書揭露了人工智慧如何剝削人類勞動力的神話，也為更公平的數位未來提出有力論據。
矽谷給我們灌輸了這樣的錯覺，即人工智慧是一種無摩擦科技，將為人類帶來財富與繁榮。但在這個光鮮亮麗的外表下，隱藏著一個殘酷的現實，全球數百萬勞動人口往往在惡劣條件下辛勤工作，人工智慧才得以實現。這個剝削性系統是由一個錯綜複雜的網絡維繫著，這本書對這個網絡進行了急需且精彩的調查，揭示了人工智慧不為人知的真相。
以十多年來數百次的訪談與數千小時的實地調查為基礎，《血汗AI》描繪了刻意被隱藏的工人生活，以及決定其未來的權力結構。這本書讓人工智慧剝削的對象表達他們的心聲，這些被剝削者包括出色的作家與藝術家，以及許許多多資料標註員、內容審核員與倉庫工人等，書中呈現出他們危險、低薪的勞動，如何與性別、種族與殖民剝削的悠久歷史產生關聯。
歐洲殖民主義遺毒＋全球化結構的不平等
【全面剖析：新興產業背後不為人知的勞動真相】
如果人工智慧被理解為一台擷取機器，那麼我們就是原料。
◆內容簡介
對娜歐米・克萊恩（Naomi Klein）和妮可・柏勒斯（Nicole Perlroth）的讀者來說，這本書揭露了人工智慧如何剝削人類勞動力的神話，也為更公平的數位未來提出有力論據。
矽谷給我們灌輸了這樣的錯覺，即人工智慧是一種無摩擦科技，將為人類帶來財富與繁榮。但在這個光鮮亮麗的外表下，隱藏著一個殘酷的現實，全球數百萬勞動人口往往在惡劣條件下辛勤工作，人工智慧才得以實現。這個剝削性系統是由一個錯綜複雜的網絡維繫著，這本書對這個網絡進行了急需且精彩的調查，揭示了人工智慧不為人知的真相。
以十多年來數百次的訪談與數千小時的實地調查為基礎，《血汗AI》描繪了刻意被隱藏的工人生活，以及決定其未來的權力結構。這本書讓人工智慧剝削的對象表達他們的心聲，這些被剝削者包括出色的作家與藝術家，以及許許多多資料標註員、內容審核員與倉庫工人等，書中呈現出他們危險、低薪的勞動，如何與性別、種族與殖民剝削的悠久歷史產生關聯。
266 pages, Kindle Edition
First published August 6, 2024
91 people are currently reading
3933 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 - 30 of 69 reviews
December 7, 2024
Great overview of the different forms of labor implicated in and impacted by the AI revolution - and a hopeful note of how we should reconsider this depressing time as one that could, if we choose to organize, be the perfect opportunity to seize power from the capitalist class determined to break things
November 6, 2024
I got what I expected here and I'm happy with it. The authors do a great job taking apart the current AI industry supply chain, starting with the annotators in Kenya and Uganda, going to Great Britain for a short look at what the engineers are experiencing, then to Iceland to have a close look at the huge data and computing centres there, then looking at the artists affected by the AI revolution, the workers with AI managers and the organisers, those whose goal is to fight back against the predatory practices of their employers.
There's a lot to process here. The authors are very good at discerning trends in the AI industry, then integrating them within a global framework. The book may get a bit technical at times, but the authors still do a great job of telling a story which interweaves these separate human destinies, who have a stake in their own economic realities and as parts of a greater whole on the global market.
The authors limit their inquiry to the current state of the economy, keeping the technical aspects of AI development to the sideline, and its possible development out of the picture. As activists, they give examples of bottom-up strategies to make companies offer more rights and better job security as well as more reliable income sources. In addition, they propose different organisational structures for companies, in order to solve the problems they've identified, ways in which employees can have something to say in relation to how the company's capital is used. Interesting stuff, couldn't actually follow along, though. However, somebody more interested in economic theory might get something out of that.
Franky, I'm satisfied with the breath and the depth of information in the book. It's a mirror of the present state of affairs and the flow of capital in the AI industry.
There's a lot to process here. The authors are very good at discerning trends in the AI industry, then integrating them within a global framework. The book may get a bit technical at times, but the authors still do a great job of telling a story which interweaves these separate human destinies, who have a stake in their own economic realities and as parts of a greater whole on the global market.
The authors limit their inquiry to the current state of the economy, keeping the technical aspects of AI development to the sideline, and its possible development out of the picture. As activists, they give examples of bottom-up strategies to make companies offer more rights and better job security as well as more reliable income sources. In addition, they propose different organisational structures for companies, in order to solve the problems they've identified, ways in which employees can have something to say in relation to how the company's capital is used. Interesting stuff, couldn't actually follow along, though. However, somebody more interested in economic theory might get something out of that.
Franky, I'm satisfied with the breath and the depth of information in the book. It's a mirror of the present state of affairs and the flow of capital in the AI industry.
October 30, 2024
Suurepärane kriitiline ja kaine ülevaade AI tumedast poolest, mis ei ole seotud hirmsate tulevikustsenaariumidega, kuidas masinad võtavad kõik üle, vaid vähemalt sama hirmsate kirjeldustega sellest, kuidas on masinavärk juba praegu üles ehitatud ja toimima pandud. Üksikute konkreetsete inimeste kaudu jutustavad autorid, kuidas kogu süsteem tugineb kolonialistlikule alusele: masina toitmiseks teevad nüristavat ja düstoopilist tööd odava tööjõuga riikidest pärit inimesed, kel pole seejuures mingeid garantiisid ega kindlust, et neil on see töö ka homme, kui nad juhtusid täna näiteks valel ajal vetsu minema. Autorid kirjeldavad töökeskkondade düstoopilisi jälgimissüsteeme, mis jõuavad varem või hiljem meie kõigini. Nad kirjeldavad üleilmset tehnoloogilist masinavärki, kus töötajad on ülesannete kaupa niivõrd killustunud, et neil pole süsteemi ülejäänud osadest mingit aimu. Nad on mõne minutiga asendatavad ja neil on väga raske koonduda ja paremate tingimuste nimel võidelda (aga mitte võimatu). Raamat kirjeldab AI-süsteemide erinevaid tegutsemispõhimõtteid ja tagajärgi, nagu autoritelt loomingu varastamine, hoomamatu mõju sõjapidamisele ja keskkonnale, raha, võimu ja otsuste koondumine üksikute jõukate tehnoloogiahiidude kätte jne. Seejuures on raamatu kandev idee, et see ei pea niimoodi olema. Tehnoloogia pole paratamatult selliseks kujunenud - selle taga on poliitilised huvid ja rahastus, mida on võimalik suunata. Autorid pakuvad ka selleks konstruktiivseid sõnumeid ja lahendusi.
January 9, 2026
Feeding the Machine beginnt äußerst vielversprechend. Besonders eindrücklich sind die Einblicke in die Arbeit von Menschen, die Social-Media-Inhalte manuell filtern müssen. Muldoon beschreibt eindringlich, unter welchem massiven psychologischen Druck diese Arbeiter:innen stehen und welchen verstörenden Bildern und Videos sie täglich ausgesetzt sind.Inhalte, die nie für die Öffentlichkeit bestimmt waren. Diese Passagen sind stark, wichtig und bleiben im Gedächtnis.
Im weiteren Verlauf weitet das Buch seinen Fokus. Es geht grundsätzlich um Regulierung von Künstlicher Intelligenz, um Machtverhältnisse in Tech-Unternehmen und um die systematische Ausbeutung menschlicher Arbeitskraft, die zunächst für KI-Systeme benötigt und später durch diese ersetzt wird. Muldoon zeigt überzeugend auf, wie sehr aktuelle KI-Erfolge auf unsichtbarer, schlecht bezahlter und oft traumatisierender Arbeit beruhen.
Warum mir das Buch letztlich jedoch sauer aufstößt, ist sein grundlegender Ton: Es bleibt über weite Strecken bei einer permanenten Anklage. Missstände werden detailliert beschrieben, moralisch eingeordnet und wiederholt kritisiert,jedoch ohne wirklich neue oder tragfähige Lösungsansätze anzubieten. Konkrete Perspektiven, wie Regulierung aussehen könnte oder wie sich diese Strukturen realistisch verändern lassen, bleiben vage.
So hinterlässt Feeding the Machine am Ende ein ambivalentes Gefühl: ein wichtiges, gut recherchiertes Buch mit starken Beobachtungen, das jedoch mehr diagnostiziert als gestaltet. Wer eine fundierte Kritik an der politischen Ökonomie von KI sucht, wird hier fündig. Wer hingegen auf konstruktive Wege aus der beschriebenen Misere hofft, dürfte enttäuscht zurückbleiben.
Im weiteren Verlauf weitet das Buch seinen Fokus. Es geht grundsätzlich um Regulierung von Künstlicher Intelligenz, um Machtverhältnisse in Tech-Unternehmen und um die systematische Ausbeutung menschlicher Arbeitskraft, die zunächst für KI-Systeme benötigt und später durch diese ersetzt wird. Muldoon zeigt überzeugend auf, wie sehr aktuelle KI-Erfolge auf unsichtbarer, schlecht bezahlter und oft traumatisierender Arbeit beruhen.
Warum mir das Buch letztlich jedoch sauer aufstößt, ist sein grundlegender Ton: Es bleibt über weite Strecken bei einer permanenten Anklage. Missstände werden detailliert beschrieben, moralisch eingeordnet und wiederholt kritisiert,jedoch ohne wirklich neue oder tragfähige Lösungsansätze anzubieten. Konkrete Perspektiven, wie Regulierung aussehen könnte oder wie sich diese Strukturen realistisch verändern lassen, bleiben vage.
So hinterlässt Feeding the Machine am Ende ein ambivalentes Gefühl: ein wichtiges, gut recherchiertes Buch mit starken Beobachtungen, das jedoch mehr diagnostiziert als gestaltet. Wer eine fundierte Kritik an der politischen Ökonomie von KI sucht, wird hier fündig. Wer hingegen auf konstruktive Wege aus der beschriebenen Misere hofft, dürfte enttäuscht zurückbleiben.
December 3, 2024
This book is criminally underrated and it is now my goal to make everyone I know read it.
December 12, 2024
This was so much more radical, anti-capitalist, and pro-worker than I imagined it would be. That to say it was amazing. Do not feed the extraction machine.
September 8, 2024
Picked up this book because I saw a passing mention about the electrical cost of AI and realized how little I understood how, well, basically the internet/AI/computers work.
This book moves through the more hidden pieces of human involvement and not so much about Silicon Valley (must track down another book). When we use AI, it looks like an instanteous computer program, but it didn't come out of thin air. This multibillion dollar industry has high stakes.
The most upsetting part is the slave labor type conditions overseas. Content moderators have to sort through upsetting material and then tag it for what that upsetting material is so that the AI can learn what is inappropriate. So guy being brutally murdered? Definitely not something you want to come across on Facebook, but these workers have to see it all to tag it.
Their work could be so important, but the conditions suck. On the global scale, companies are competing, well, globally to be most efficient but cheapest option - conditions then are low wages, insane hours, and metrics analyzing for accuracy. So not only are you having to watch snuff videos, but also a LOT of them, as fast as you can and with insufficient breaks. Additionally, there's not many other places to work and this specific skillset you learn doesn't translate into other work.
Because it's happening "far from home" it's easy to ignore it. Plus the big corporations funding this work are not particularly interested in the worker's conditions.
That was the ORGANIZER chapter. For the ENGINEER, their job is to figure out how to take information from humans and make it objective. AI itself can't be truly objective. One instance showed AI sorting through job applications and showing preference toward men, based on the data it was fed. Bigger questions arise regarding geopolitics, the delicate process of when something can be labeled "genocide," whether the term "far east" is appropriate. Imagine that someone would think "oh well AI prefers men because men are naturally better, it's objectively correct!" and not the long history of misogyny.
This issue is more known in the art community, but AI is destroying art and artists imho. (The ARTIST chapter) AI can't create art and instead pulls together images from all over the web to fulfill a request. In the book, it's discussed how VAs (voice actors) are seeing clauses in their contracts for rights of their voices to be taken - so you could do the work, and then the company would presumably be able to generate work from the sound bits without paying the VA.
The argument in the book is that, it's not a problem that the technology exists here, but that proper consent wasn't given. What if the VA was paid more substantially and also received royalties? And were on call if additional sound bites were needed? It's the knowledge of what AI can do, how it's being used and how to advocate for your art.
There's also the question of whether AI will follow the path of photography - artists thought photography would kill art, but it became another tool for an artist. Will the same be for AI? I'm inclined to say no, personally. I see a lot of AI generated tarot art now, which particularly bugs me. It seems to lose the spiritual connection, the intent behind the deep meditative art itself. I think AI is more than a medium, but I guess only time will tell.
The OPERATOR chapter follows an Amazon worker in the UK, but also discusses the ways companies monitor their employees to an insane degree. I've actually heard this from a friend but Amazon factories anaylze how fast and accurately you work. Safety precautions would have you bin larger items near the bottom, but the metrics are only considered with how quickly the items are binned. Thus employees are shoving things wherever to keep their jobs. (My friend trying to follow safety protocol ended up being why she was abruptly fired. Make it make sense.)
This chapter also discusses the connection between Ford's car assembly line and how factories are operated. Ford was innovative, but at the cost of employee wellness.
Walmart went under fire for having their employees download a price checker app to their personal phones that also monitored what they were doing on said phones during work hours. WFH employees are put under surveillance by other programs tracking what they are doing, how often their mouse is being moved - insane. What the hell does it matter as long as the work is being completed?
(Side note: Walmart was my first "real" job and I don't know how to describe how dehumanizing the experience was. It was before the pay increase but well after it was considered a good company to work for)
I mentioned the resume thing earlier. This chapter makes the point that AI could very easily replace management - it's already making hiring, firing, and managing decisions. (Wait, does that make you uncomfortable?)
The INVESTOR chapter talks about the pressures of trying to be ethical while, well, making your money back. You want to invest in the tech startups that are ethically run.... but ethics don't make money as easily. This chapter is on why it's complicated, which I'm still trying to understand.
There's also a TECHNICIAN chapter, regarding the physical storage of data. This part I'm still trying to understand but all information on the internet has to be stored somewhere physically and that requires a massive amount of electricity. The chapter focused in on a storage place in Iceland and how the naturally cool temps are great for protecting it, but it drains electricity from the nearby town so there's a limit to how much storage can be placed there.
Also someone please explain to me fiber optic cables because wow, did ya'll know that there are cables buried all over and also along the ocean floor?
Moving on, the last part of the book is how we can improve conditions for these invisible workers
Their proposal:
1. Build collective power of worker organization (unions, but makes them transnational)
2. Build civil societies to hold companies accountable (be informed, shame these companies to do better)
3. Enacting stricter regulations on AI
4. Establishing forms of work governments and ownership of enterprise (in addition to wages, employees own a piece of the company and a say to improve it)
5. Challenging injustices of the broader systems in which the companies operate.
Really valuable information in this book to understand a little more how AI functions and its imperfections. A sobering note in the end talked about "smart bombs" which supposedly don't target civilians. They still do, unfortunately, and further remove responsibilities of war. We just gotta know how this stuff works, what its limits are and how to protect ourselves and each other, as corporate sure as hell doesn't care for the human worker.
Support your unions, folks.
This book moves through the more hidden pieces of human involvement and not so much about Silicon Valley (must track down another book). When we use AI, it looks like an instanteous computer program, but it didn't come out of thin air. This multibillion dollar industry has high stakes.
The most upsetting part is the slave labor type conditions overseas. Content moderators have to sort through upsetting material and then tag it for what that upsetting material is so that the AI can learn what is inappropriate. So guy being brutally murdered? Definitely not something you want to come across on Facebook, but these workers have to see it all to tag it.
Their work could be so important, but the conditions suck. On the global scale, companies are competing, well, globally to be most efficient but cheapest option - conditions then are low wages, insane hours, and metrics analyzing for accuracy. So not only are you having to watch snuff videos, but also a LOT of them, as fast as you can and with insufficient breaks. Additionally, there's not many other places to work and this specific skillset you learn doesn't translate into other work.
Because it's happening "far from home" it's easy to ignore it. Plus the big corporations funding this work are not particularly interested in the worker's conditions.
That was the ORGANIZER chapter. For the ENGINEER, their job is to figure out how to take information from humans and make it objective. AI itself can't be truly objective. One instance showed AI sorting through job applications and showing preference toward men, based on the data it was fed. Bigger questions arise regarding geopolitics, the delicate process of when something can be labeled "genocide," whether the term "far east" is appropriate. Imagine that someone would think "oh well AI prefers men because men are naturally better, it's objectively correct!" and not the long history of misogyny.
This issue is more known in the art community, but AI is destroying art and artists imho. (The ARTIST chapter) AI can't create art and instead pulls together images from all over the web to fulfill a request. In the book, it's discussed how VAs (voice actors) are seeing clauses in their contracts for rights of their voices to be taken - so you could do the work, and then the company would presumably be able to generate work from the sound bits without paying the VA.
The argument in the book is that, it's not a problem that the technology exists here, but that proper consent wasn't given. What if the VA was paid more substantially and also received royalties? And were on call if additional sound bites were needed? It's the knowledge of what AI can do, how it's being used and how to advocate for your art.
There's also the question of whether AI will follow the path of photography - artists thought photography would kill art, but it became another tool for an artist. Will the same be for AI? I'm inclined to say no, personally. I see a lot of AI generated tarot art now, which particularly bugs me. It seems to lose the spiritual connection, the intent behind the deep meditative art itself. I think AI is more than a medium, but I guess only time will tell.
The OPERATOR chapter follows an Amazon worker in the UK, but also discusses the ways companies monitor their employees to an insane degree. I've actually heard this from a friend but Amazon factories anaylze how fast and accurately you work. Safety precautions would have you bin larger items near the bottom, but the metrics are only considered with how quickly the items are binned. Thus employees are shoving things wherever to keep their jobs. (My friend trying to follow safety protocol ended up being why she was abruptly fired. Make it make sense.)
This chapter also discusses the connection between Ford's car assembly line and how factories are operated. Ford was innovative, but at the cost of employee wellness.
Walmart went under fire for having their employees download a price checker app to their personal phones that also monitored what they were doing on said phones during work hours. WFH employees are put under surveillance by other programs tracking what they are doing, how often their mouse is being moved - insane. What the hell does it matter as long as the work is being completed?
(Side note: Walmart was my first "real" job and I don't know how to describe how dehumanizing the experience was. It was before the pay increase but well after it was considered a good company to work for)
I mentioned the resume thing earlier. This chapter makes the point that AI could very easily replace management - it's already making hiring, firing, and managing decisions. (Wait, does that make you uncomfortable?)
The INVESTOR chapter talks about the pressures of trying to be ethical while, well, making your money back. You want to invest in the tech startups that are ethically run.... but ethics don't make money as easily. This chapter is on why it's complicated, which I'm still trying to understand.
There's also a TECHNICIAN chapter, regarding the physical storage of data. This part I'm still trying to understand but all information on the internet has to be stored somewhere physically and that requires a massive amount of electricity. The chapter focused in on a storage place in Iceland and how the naturally cool temps are great for protecting it, but it drains electricity from the nearby town so there's a limit to how much storage can be placed there.
Also someone please explain to me fiber optic cables because wow, did ya'll know that there are cables buried all over and also along the ocean floor?
Moving on, the last part of the book is how we can improve conditions for these invisible workers
Their proposal:
1. Build collective power of worker organization (unions, but makes them transnational)
2. Build civil societies to hold companies accountable (be informed, shame these companies to do better)
3. Enacting stricter regulations on AI
4. Establishing forms of work governments and ownership of enterprise (in addition to wages, employees own a piece of the company and a say to improve it)
5. Challenging injustices of the broader systems in which the companies operate.
Really valuable information in this book to understand a little more how AI functions and its imperfections. A sobering note in the end talked about "smart bombs" which supposedly don't target civilians. They still do, unfortunately, and further remove responsibilities of war. We just gotta know how this stuff works, what its limits are and how to protect ourselves and each other, as corporate sure as hell doesn't care for the human worker.
Support your unions, folks.
July 26, 2024
This was a fantastic book. It was written and structured beautifully, and it was eye-opening. I read it immediately after "Power and Progress" by Daron Acemoglu, and together these two books made me think so much about how the current trend in Artificial Intelligence is similar to what the working class experienced during the previous technology boosts, which resulted in more inequality and exploitation. It's a must-read to understand the not-so-visible costs of the AI hype.
September 4, 2025
Solo me dieron más razones para odiar la inteligencia artificial y el capitalismo
January 16, 2025
Even if you just read the introduction and conclusion it would be a powerful book.
This shows so much behind the “easy to use” chatbot or AI interface we use and how people are continually exploited in the name of capitalism.
This shows so much behind the “easy to use” chatbot or AI interface we use and how people are continually exploited in the name of capitalism.
December 24, 2025
Very interesting topic. Too repetitive towards the end, it could have been much shorter.
July 27, 2025
I could not find the comment review from another (bad) book who recommended this book. But thank you goodread user for this recomendación. Is the first of some of the AI books and content I been reading, I find useful and insightful.Highly recommended. Instead of focusing on narrating how the future could be and present dystopian worlds (which are never good ones and if you are looking for that, there are good science fiction authors who will do a better job at it) this book presents a series of characters that play a role TODAY in the development of AI and explain how the AI is changing the world now.
It is a great startup point. But more reading is necessary to fully understand (or at least to try to) how this new revolution will impact the society. I am on that search.
It is a great startup point. But more reading is necessary to fully understand (or at least to try to) how this new revolution will impact the society. I am on that search.
July 27, 2025
Kurzmeinung / Leseerlebnis
Vielen Dank an Netgalley und Harper Audio für das kostenlose Rezensionsexemplar.
Dem Sprecher Stefan Kaminski konnte man - wie immer - sehr gut zuhören und das Thema ist sehr interessant. Die Inhalte sind informativ, wenn auch teilweise sehr ausschweifend. Als jemand, der von sich aus der KI schon besonders kritisch gegenüber steht, sind einige Betrachtungen hier besonders haarsträubend. Bemerkenswert fand ich zum Beispiel die Aufrechnung dass KI mehr Menschenleben gefordert als gerettet hat (z.B. im militärischen Einsatz, selbstfahrende Autos, etc.). Nach einer solchen Lektüre fragt man sich, warum die Menschen ihr Schicksal dermaßen herausfordern und das auf Kosten von niedrigbezahlten Menschen im globalen Süden, die für einen Bruchteil des Ertrags (im Promille-Bereich) den Apparat am laufen halten.
#FeedingtheMachineHinterdenKulissenderKIImperien #NetGalleyDE
Vielen Dank an Netgalley und Harper Audio für das kostenlose Rezensionsexemplar.
Dem Sprecher Stefan Kaminski konnte man - wie immer - sehr gut zuhören und das Thema ist sehr interessant. Die Inhalte sind informativ, wenn auch teilweise sehr ausschweifend. Als jemand, der von sich aus der KI schon besonders kritisch gegenüber steht, sind einige Betrachtungen hier besonders haarsträubend. Bemerkenswert fand ich zum Beispiel die Aufrechnung dass KI mehr Menschenleben gefordert als gerettet hat (z.B. im militärischen Einsatz, selbstfahrende Autos, etc.). Nach einer solchen Lektüre fragt man sich, warum die Menschen ihr Schicksal dermaßen herausfordern und das auf Kosten von niedrigbezahlten Menschen im globalen Süden, die für einen Bruchteil des Ertrags (im Promille-Bereich) den Apparat am laufen halten.
#FeedingtheMachineHinterdenKulissenderKIImperien #NetGalleyDE
Read
January 3, 2026Rly good book, clearly outlines the exploitative global system of labor that keeps AI chugging along & how these systems use the same labor management frameworks set up during the Atlantic slave trade, Industrial Revolution, etc. Recycled systems of exploitation and complete resource depletion. One closing note I appreciated was to keep in mind that sensationalism & fear-mongering around AI becoming an autonomous force and taking over/destroying the world like a Hollywood move shouldn’t be at the forefront of conversation, because AI doesn’t need to be an autonomous force to destroy the world. It’s already happening with human implementation and there are tangible things we can do to resist it. Trying to also come to terms with the fact that machine learning can allow for incredible strides medicine, philosophy, etc. and can be valuable when people interact with it in a thoughtful way, while also seeing the overwhelming number of data centers and AI companies cropping up across the globe and the localities in the Global South that have been forced into dependency on finding data annotation work being barely compensated for the most horrific work days..
Read
February 4, 2025I believe this is a book that will be a major work in the topic of AI within the next few years. Gives a solid understanding of the history of artificial intelligence, how deeply ingrained it is in our lives in ways we don’t even realize, and details how AI is a reflection of our society and culture in too many ways to count. An abnormal experience for me to read something that came out so recent, and given its 2024 release it was able to talk to things that are going on today and will be going on near into the future. I’ve always thought of AI as being this abstract technology that more or less just came out of nowhere, but acknowledging the human labor, a lot of which being oppressive in nature, behind the ongoing development of the AI we see makes it all the more tangible. AI and tech is only expected to grow exponentially through the rest of the decade and it’s hard to see how this growth won’t continue to primarily benefit the rich and those that run these industries. Highly recommend not matter the level of interest one has in AI as these topics and consequences will affect everyone.
March 22, 2025
Un libro clave para entender todas las aristas de una realidad tan compleja como la industria de la IA.
Una mirada sobre los aspectos éticos de la tecnología que está revolucionando el mundo. La pregunta es: ¿a costa de qué? Y, ¿hay ganadores y vencidos?
Un conjunto de historias extendidas por el mundo, de ejemplos y consecuencias reales. Que nos hablan de nuestra historia común. De nuestra responsabilidad y del impacto de nuestras decisiones.
Los autores nos advierten de la urgencia de crear, establecer y garantizar mecanismos para proteger al elemento más débil de la máquina de extracción: el ser humano; su materia prima.
Lectura altamente recomendable.
Una mirada sobre los aspectos éticos de la tecnología que está revolucionando el mundo. La pregunta es: ¿a costa de qué? Y, ¿hay ganadores y vencidos?
Un conjunto de historias extendidas por el mundo, de ejemplos y consecuencias reales. Que nos hablan de nuestra historia común. De nuestra responsabilidad y del impacto de nuestras decisiones.
Los autores nos advierten de la urgencia de crear, establecer y garantizar mecanismos para proteger al elemento más débil de la máquina de extracción: el ser humano; su materia prima.
Lectura altamente recomendable.
November 16, 2024
It all comes back to capitalism…
July 17, 2025
AI is a hungry beast that is eating human labor because the people who control it want to use it to maximize profit.
This is SUCH a good book if you want to learn ANYTHING about AI.
This is SUCH a good book if you want to learn ANYTHING about AI.
September 10, 2024
This is an amazing book, with years of research in the background. It proves the evil sides of the AI which in most times is presented only in an ideal setting, with huge money. Highly recommend it!
August 27, 2024
These authors are evil. The contempt I have for them knows no bounds
October 6, 2024
Thoroughly explores the hidden Labour of AI, but it leans quite academic for the casual reader
October 29, 2024
solid 4 stars; keep reading or listening; the opening chapter seems to limit the scope of the book to the exploitation of workers feeding data into AI algorithms but the author has more in store for you; a grand tour on the actual impact AI and big tech has on our lives, how it manipulates and exploits those who depend on it; on the subject be sure to check out Kevin Kelly's magisterial; What Technology Wants.
December 4, 2024
Absolutely brilliant. Best book I’ve read all year. I already knew that AI was problematic and exploitative but now I feel even more equipped to share with others how this extraction machine really works. Everyone should read this book, and everyone should be very concerned about AI and how it is deployed in our society.
November 24, 2024
If I could give this book ten stars, I would.
July 2, 2025
Feeding the Machine: The Hidden Human Labour Powering AI ist ein engagiertes und aufrüttelndes Buch, das die menschliche Arbeit sichtbar macht, die hinter dem scheinbar autonomen Funktionieren Künstlicher Intelligenz steht. Die Autoren – James Muldoon, Mark Graham und Callum Cant – richten den Fokus auf die globalen Arbeitsrealitäten von Datenannotatoren, insbesondere im Globalen Süden. Diese Menschen müssen häufig unter prekären Bedingungen psychisch belastende Inhalte sichten und sortieren, um KI-Systeme wie Chatbots, Suchmaschinen oder Empfehlungsalgorithmen zu trainieren. Ihre Geschichten stehen sinnbildlich für eine neue Form der digitalen Ausbeutung im Zeitalter der Automatisierung.
Das Buch bietet darüber hinaus eine kritische Analyse der Funktionsweise und Gefahren von Large Language Models (LLMs), wie sie etwa bei ChatGPT zum Einsatz kommen. Es beschreibt, wie diese Modelle Voreingenommenheiten verstärken, Machtasymmetrien verfestigen und potenziell für Desinformation oder Überwachung missbraucht werden können. Auch die fortschreitende Automatisierung von Arbeitsprozessen in Logistikzentren wie denen von Amazon wird eingehend untersucht – inklusive des Einsatzes von KI zur Taktung, Bewertung und Kontrolle menschlicher Arbeit. Die Autoren zeigen, wie unter dem Deckmantel technischer Effizienz soziale Ungleichheit und Kontrolle neu organisiert werden.
Feeding the Machine ist nicht nur eine Einführung in die verborgene Infrastruktur der KI (Datensammlung und -aufbereitung, Content-Moderation, Datenannotation und -labeling, Menschliche Überprüfung von KI-Outputs, Mikroarbeit auf Plattformen), sondern auch eine glänzend geschriebene Kapitalismuskritik. Die gegenwärtige KI-Ökonomie kann als Neuauflage eines alten Handels mit Wissen verstanden werden– dessen erste Auflage einst von den Sophisten betrieben wurde, nun jedoch von Technologiekonzernen dominiert wird. Gegen diese Machtkonzentration plädieren die Autoren für kollektive Gegenstrategien: Durch Gewerkschaften, zivilgesellschaftlichen Druck und politische Regulierung. Das Buch liefert damit nicht nur Einblicke in die Realität der KI-Arbeit, sondern auch in mögliche Wege zu einer gerechteren digitalen Zukunft.
Das Buch bietet darüber hinaus eine kritische Analyse der Funktionsweise und Gefahren von Large Language Models (LLMs), wie sie etwa bei ChatGPT zum Einsatz kommen. Es beschreibt, wie diese Modelle Voreingenommenheiten verstärken, Machtasymmetrien verfestigen und potenziell für Desinformation oder Überwachung missbraucht werden können. Auch die fortschreitende Automatisierung von Arbeitsprozessen in Logistikzentren wie denen von Amazon wird eingehend untersucht – inklusive des Einsatzes von KI zur Taktung, Bewertung und Kontrolle menschlicher Arbeit. Die Autoren zeigen, wie unter dem Deckmantel technischer Effizienz soziale Ungleichheit und Kontrolle neu organisiert werden.
Feeding the Machine ist nicht nur eine Einführung in die verborgene Infrastruktur der KI (Datensammlung und -aufbereitung, Content-Moderation, Datenannotation und -labeling, Menschliche Überprüfung von KI-Outputs, Mikroarbeit auf Plattformen), sondern auch eine glänzend geschriebene Kapitalismuskritik. Die gegenwärtige KI-Ökonomie kann als Neuauflage eines alten Handels mit Wissen verstanden werden– dessen erste Auflage einst von den Sophisten betrieben wurde, nun jedoch von Technologiekonzernen dominiert wird. Gegen diese Machtkonzentration plädieren die Autoren für kollektive Gegenstrategien: Durch Gewerkschaften, zivilgesellschaftlichen Druck und politische Regulierung. Das Buch liefert damit nicht nur Einblicke in die Realität der KI-Arbeit, sondern auch in mögliche Wege zu einer gerechteren digitalen Zukunft.
July 14, 2025
Ich dachte, ich lese mal eben ein bisschen was über KI – und plötzlich sitze ich mit offenem Mund da, als hätte mir HAL 9000 höchstpersönlich die Augen geöffnet. Feeding the Machine ist kein schnarchiger Tech-Wälzer, sondern eher wie ein Wachrüttel-Buch mit dem Charme eines Investigativ-Krimis und der Schlagkraft eines linken Hakens ins ethische Gewissen. Muldoon, Graham und Cant zeigen einem ziemlich eindrucksvoll, dass hinter jeder smarten KI nicht Magie, sondern ziemlich viel menschliches Elend steckt.
Während ich also gemütlich meinen Kaffee schlürfe, sitzt irgendwo jemand und klickt sich für einen Hungerlohn durch Millionen Bilder, damit meine App versteht, was ein Hund ist. Und das ist kein hypothetisches Beispiel – die Autoren holen echte Menschen aus dem Schatten der Serverfarmen auf die Bühne, und das auf eine Art, bei der mir mehrmals das Lachen im Hals stecken blieb. Ich sag nur: Kapitalismus in Trainingsdatenform.
Die Geschichten sind so absurd und tragisch, dass ich zwischen Schock, Empörung und „Das kann doch nicht wahr sein?!“-Momenten permanent auf Achterbahnfahrt war. Das Buch ist wie ein Blick hinter den Vorhang des großen KI-Zirkus – und Spoiler: Der Zauberer ist ein schlecht bezahlter Clickworker in Kampala.
Was ich besonders gefeiert habe: Die Mischung aus trockener Analyse und süffisantem Sarkasmus. Da wird nichts beschönigt, aber auch nichts mit moralischem Zeigefinger totgeredet. Die Autoren wissen, was sie tun – und das merkt man auf jeder Seite. Sie zerlegen den Mythos KI mit der Präzision eines Algorithmus, aber mit der Wucht eines Vorschlaghammers.
Nach dem Lesen hab ich ehrlich gesagt kurz überlegt, ob ich meinen Laptop rituell verbrennen sollte. Hab’s dann doch gelassen – aber dafür mein Konsumverhalten überdacht.
Kurz: Wer glaubt, KI sei einfach nur cool und effizient, wird hier ordentlich geerdet. Wer schon skeptisch war, bekommt das passende Feuerwerk dazu. Und wer gar nichts über KI weiß – na, der sollte dieses Buch sowieso sofort lesen.
5 Sterne, weil es weh tut – aber auf die gute Art.
Während ich also gemütlich meinen Kaffee schlürfe, sitzt irgendwo jemand und klickt sich für einen Hungerlohn durch Millionen Bilder, damit meine App versteht, was ein Hund ist. Und das ist kein hypothetisches Beispiel – die Autoren holen echte Menschen aus dem Schatten der Serverfarmen auf die Bühne, und das auf eine Art, bei der mir mehrmals das Lachen im Hals stecken blieb. Ich sag nur: Kapitalismus in Trainingsdatenform.
Die Geschichten sind so absurd und tragisch, dass ich zwischen Schock, Empörung und „Das kann doch nicht wahr sein?!“-Momenten permanent auf Achterbahnfahrt war. Das Buch ist wie ein Blick hinter den Vorhang des großen KI-Zirkus – und Spoiler: Der Zauberer ist ein schlecht bezahlter Clickworker in Kampala.
Was ich besonders gefeiert habe: Die Mischung aus trockener Analyse und süffisantem Sarkasmus. Da wird nichts beschönigt, aber auch nichts mit moralischem Zeigefinger totgeredet. Die Autoren wissen, was sie tun – und das merkt man auf jeder Seite. Sie zerlegen den Mythos KI mit der Präzision eines Algorithmus, aber mit der Wucht eines Vorschlaghammers.
Nach dem Lesen hab ich ehrlich gesagt kurz überlegt, ob ich meinen Laptop rituell verbrennen sollte. Hab’s dann doch gelassen – aber dafür mein Konsumverhalten überdacht.
Kurz: Wer glaubt, KI sei einfach nur cool und effizient, wird hier ordentlich geerdet. Wer schon skeptisch war, bekommt das passende Feuerwerk dazu. Und wer gar nichts über KI weiß – na, der sollte dieses Buch sowieso sofort lesen.
5 Sterne, weil es weh tut – aber auf die gute Art.
Read
May 31, 2025(DNF @ 50%) From the blurb, I was expecting person-centred journalism of the Patrick Radden Keefe ilk about various people in the AI chain, such as the underpaid 'annotator' in the emerging world, charged with watching car crash footage. In fact, those portraits of "the humans behind AI" are glancing at best.
This is a dry and unfortunately jargon-filled nonfiction read about the various issues around AI. Little about the way that information is presented feels novel or distinctive.
If you have not engaged with the AI debate at all, you might find this book illuminating, but if you've... been a person in the world... who reads the news... most of this will feel old hat.
I ultimately abandoned the book after the chapter about genAI and art, which glosses over the fact that pirated ebooks from every author you've ever heard of were used to train LLMs. Lest we forget, that means that when you generate some text from a chatbot and it comes up with something artful, there is a 99.9% chance that phrase is fully yanked from a living, breathing author who did not get paid at any point in this process.
Feeding the Machine is so woolly on the subject of whether stealing from artists is a bad thing that it's infuriating ("maybe it's not fair that only the talented are able to create things", wah wah wahhhhhh I'm a wittle baby). Indeed, this book ends up reading like -- gulp -- AI, where every subject is given an exhaustive pro/con list.
Ultimately, Feeding the Machine appears to not have a viewpoint at all -- and if you don't have a viewpoint, what's the point?
This is a dry and unfortunately jargon-filled nonfiction read about the various issues around AI. Little about the way that information is presented feels novel or distinctive.
If you have not engaged with the AI debate at all, you might find this book illuminating, but if you've... been a person in the world... who reads the news... most of this will feel old hat.
I ultimately abandoned the book after the chapter about genAI and art, which glosses over the fact that pirated ebooks from every author you've ever heard of were used to train LLMs. Lest we forget, that means that when you generate some text from a chatbot and it comes up with something artful, there is a 99.9% chance that phrase is fully yanked from a living, breathing author who did not get paid at any point in this process.
Feeding the Machine is so woolly on the subject of whether stealing from artists is a bad thing that it's infuriating ("maybe it's not fair that only the talented are able to create things", wah wah wahhhhhh I'm a wittle baby). Indeed, this book ends up reading like -- gulp -- AI, where every subject is given an exhaustive pro/con list.
Ultimately, Feeding the Machine appears to not have a viewpoint at all -- and if you don't have a viewpoint, what's the point?
February 1, 2025
Another day, another book about the evils of capitalism.
While discourse on the evils of AI tends to focus on futuristic "big-brother" scenarios, this book drives home the reality that (certain kinds of) AI technologies are in use right now, and affect the lives of millions around the world. It is just another extraction machine, albeit on a truly global scale, and with data as the raw material. The authors introduce us to a cast of characters that support this industry, from the underpaid and overworked data-annotator to the VC investor who just wants to turn a profit before the bubble bursts. The book also offers some broader commentary about the social and political dynamics that govern the AI industry today, and makes a compelling case about why our existing legal and regulatory framework is not suited for this kind of intercontinental capitalist enterprise.
As a technology enthusiast, I felt that the book could have done a better job distinguishing the underlying tech from the systemic evil that surrounds it. AI tech is not inherently bad, and could yet play a pivotal role in fighting societal dangers such as climate change. Not all people working in AI (on the engineering side, atleast) are morally conflicted or brooding over the potential impacts of their work on society while pocketing fat paychecks.
While discourse on the evils of AI tends to focus on futuristic "big-brother" scenarios, this book drives home the reality that (certain kinds of) AI technologies are in use right now, and affect the lives of millions around the world. It is just another extraction machine, albeit on a truly global scale, and with data as the raw material. The authors introduce us to a cast of characters that support this industry, from the underpaid and overworked data-annotator to the VC investor who just wants to turn a profit before the bubble bursts. The book also offers some broader commentary about the social and political dynamics that govern the AI industry today, and makes a compelling case about why our existing legal and regulatory framework is not suited for this kind of intercontinental capitalist enterprise.
As a technology enthusiast, I felt that the book could have done a better job distinguishing the underlying tech from the systemic evil that surrounds it. AI tech is not inherently bad, and could yet play a pivotal role in fighting societal dangers such as climate change. Not all people working in AI (on the engineering side, atleast) are morally conflicted or brooding over the potential impacts of their work on society while pocketing fat paychecks.
November 13, 2024
Logra sacar a la luz el aspecto menos visible — pero de mayor importancia— de las operaciones de la IA: su base material, constituida por enormes estructuras de infraestructura, trabajo humano y una cantidad cuasi infinita de datos e información acumulados.
El libro describe los avances en la explotación del trabajo de esta nueva industria naciente. Además de sus consecuencias actualmente efectivas para aquellos que, con su trabajo, sostienen la 'Extraction Machine'; es interesante ver los gastos en infraestructura, el consumo absurdo en recursos naturales, la invención de unas herramientas de control de la clase obrera, de su uso belico (el ultimo capitulo dedicado al uso de la IA por Israel es horroroso pero excelente: "AI hasn't saved lives; it has increased the bloodshed"), etc. Pero, sumado a lo anterior, ver como la clase obrera, en todos lados del mundo, afronta tales consecuencias, y las posibilidades actuales para combatir uno de los nuevos modos del capital, es de suma importancia, y el libro realiza un trabajo excelente en ello. Ademas, no se olvida de recordarnos lo siguiente, tan antiguo como verdadero: "It is only through collective organisation and action that their interests can be served. If workers don't stand up for themselves, the nobody else will."
El libro describe los avances en la explotación del trabajo de esta nueva industria naciente. Además de sus consecuencias actualmente efectivas para aquellos que, con su trabajo, sostienen la 'Extraction Machine'; es interesante ver los gastos en infraestructura, el consumo absurdo en recursos naturales, la invención de unas herramientas de control de la clase obrera, de su uso belico (el ultimo capitulo dedicado al uso de la IA por Israel es horroroso pero excelente: "AI hasn't saved lives; it has increased the bloodshed"), etc. Pero, sumado a lo anterior, ver como la clase obrera, en todos lados del mundo, afronta tales consecuencias, y las posibilidades actuales para combatir uno de los nuevos modos del capital, es de suma importancia, y el libro realiza un trabajo excelente en ello. Ademas, no se olvida de recordarnos lo siguiente, tan antiguo como verdadero: "It is only through collective organisation and action that their interests can be served. If workers don't stand up for themselves, the nobody else will."
September 9, 2025
"[W]orkers are forced to work like robots in the hopes that AI will become more like a human."
This was an essential read! It is really a book for this post-COVID era, and it speaks clearly to the larger race-to-the-bottom tendency that cannot be left unchecked. I enjoyed the last two chapters immensely (maybe because pure tech really isn't my thing), when the authors took a sharp look at the global labour rights situation and the consequently "wicked choices" faced by countries that rely heavily on BPOs. The authors also framed the global labour landscape really sensibly with a healthy balance of ambition and realism (which I struggle to find in today's leftist literature). It would be helpful to consider in greater detail how we can more quickly enter "a world in which responsibility stops at national boundaries"!
My biggest takeaway is the reminder that any science/technological development is never pre-determined; it reinforces existing social relationships and inequalities, and therefore echoes the larger socio-economic in which these innovations emerge. It is not a trend we have to just adapt to, but requires will to change, if we want it to.
This was an essential read! It is really a book for this post-COVID era, and it speaks clearly to the larger race-to-the-bottom tendency that cannot be left unchecked. I enjoyed the last two chapters immensely (maybe because pure tech really isn't my thing), when the authors took a sharp look at the global labour rights situation and the consequently "wicked choices" faced by countries that rely heavily on BPOs. The authors also framed the global labour landscape really sensibly with a healthy balance of ambition and realism (which I struggle to find in today's leftist literature). It would be helpful to consider in greater detail how we can more quickly enter "a world in which responsibility stops at national boundaries"!
My biggest takeaway is the reminder that any science/technological development is never pre-determined; it reinforces existing social relationships and inequalities, and therefore echoes the larger socio-economic in which these innovations emerge. It is not a trend we have to just adapt to, but requires will to change, if we want it to.
Displaying 1 - 30 of 69 reviews