AI世代：從政治哲學反思人工智慧的衝擊

by Mark Coeckelbergh

人臉辨識、數位威權、同溫層效應……
科技是中立的嗎？
科技真的能帶來更好的未來嗎？
人工智慧將走向何方，只取決於我們如何利用嗎？
自由、平等、正義、民主、種族議題與氣候變遷──這些對人類生活至關重要的諸多議題與概念，與人工智慧有何關聯？

卡夫卡的《審判》、歐威爾的《一九八四》裡頭所描繪的社會狀態，或許將不再只是文學寓言。AI的發展及其帶來的技術革命，甚至已經超越了二十年前的網際網路，它能夠與人交談、能夠學習甚至創作，它看似能夠大幅改善人類的生活，但也將顛覆我們所熟知的各種價值……
AI不只是一種科技，更是一種政治。它能夠存取龐大的資料，從中進行應用與學習，卻也能從這些數據中進行預測、操控與管制。AI具有「權力」，僅以倫理學角度看待AI已經不合時宜，我們必須正視它為人類社會帶來的挑戰，無論是技術層面的運用，例如執法手段的變革對公民自由形成的威脅，或是潛在的影響，例如大數據對人類行為（勞動、消費、金融）的形塑與規訓。它有可能危害民主、加速社會的透明化、加劇當前的分配不平等、形塑人們的自我認同；甚至，我們不得不超前思考AI在未來形成一種政治社群的可能。
本書大膽嘗試理論跨界，以古典理論回應當代議題，同時提出政治哲學所關懷的重要議題及概念，必須將科技發展納入視野，且必然隨著科技的發展而改變。無論是人文學術界、技術工程界、產業界或是政策制定者，都不可不知與AI相關的規範性理論，及本書所提供的理論框架。本書適合所有關注AI技術的讀者，及所有關心總體趨勢的讀者。

「人工智慧在根本上是一個政治的領域，這本書闡明了其中的原因。本書涵蓋了關於不平等、民主、權力和後人類主義的爭論，並顯示了社會和政治理論對於理解人工智慧的重要性。」
凱特．克勞馥，《人工智慧最後的秘密》（Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence）作者
「與人工智慧相關的倫理學與政治哲學這兩個領域，聚焦在許多相同的問題上，但很難看到後者的豐富歷史被用來理解人工智慧的政治。科克爾柏格在這本重要的著作，給出了一個值得注目的例外。」
Henrik Skaug Sætra，奧斯陸大學資訊學系助理教授
「科克爾柏格的真正重點在於，他展示了數千年的哲學思想不會因為數字工程的成就而變得過時。」
Scott McLemee，《高等教育》（Inside Higher Ed）

Genres: Philosophy, Artificial Intelligence, Nonfiction, Politics, Technology

320 pages, Paperback

First published January 31, 2022

About the author

Mark Coeckelbergh is Professor of Philosophy of Media and Technology at the University of Vienna. He is the author of New Romantic Cyborgs: Romanticism, Information Technology, and the End of the Machine, AI Ethics (both published by the MIT Press), Introduction to Philosophy of Technology, and other books.

Reviews

[Ainhoa · Rating 4 out of 5]

Hace un tiempo escribí una reseña más crítica y sólo le puse 3 estrellas. Me equivoqué. Las ideas que presenta Mark se han quedado conmigo y se cuelan en muchas de mis reflexiones.

Lo único que mantengo de la reseña anterior es que fue un gran libro para leer como lectura conjunta. Particularmente, una de las conversaciones que surgió a raíz de ella ha sido fundamental para mi forma de pensar la autoridad.

[María · Rating 4 out of 5]

"El sentido «positivo» de la palabra «libertad» deriva del deseo, por parte del individuo, de ser su propio dueño. Deseo que mi vida y mis decisiones dependan de mí mismo, no de fuerzas externas de cualquier tipo. [...] Deseo, sobre todo, ser consciente de mí mismo como ser pensante, deseante y activo, que carga con la responsabilidad de sus elecciones y es capaz de explicarlas haciendo referencia a sus propias ideas y propósitos." - M. Coeckelbergh

[SWB · Rating 4 out of 5]

Dense, academic, some cloaked issues which can be teased out:

Equality, fairness, democracy and power.

Equality - ai odds not politically neutral. Biases in data, algorithms and knowledge processes used. Whether that is language (eg “man”kind) or socioeconomic (eg capitalist). Equality of opportunity vs equality of outcome vs prioritarian (giving priority to the more disadvantaged, justified by Rawls’ veil of ignorance) vs egalitarian (everyone gets the same) vs sufficientarian (everyone has a minimum chance of getting some output). Ai capitalism says that banks have no incentive to reduce bias (even if bandaids are used to correct for bias) as these capitalist systems exist to make capitalists richer

Democracy - “democracy as majority rule” is not necessarily a good thing. Representative democracy. Voters seldom act, they react. Ordinary people are not competent to understand the issues that lie behind political decisions. Self education is not sufficient, and general education is needed. Even for selecting representatives. Fake news is dangerous. Social media + ai (personalisation) creates filter bubbles which limit our horizons. Social media requires some editorial self control to avoid polarisation and hateful language. Populist politicians use ai to analyse data about voters preferences. But while it is good in a democracy that politicians know what citizens need, this use of AI "could turn into demagogic mass appeals, rather than the reasoned deliberation process envisioned, for example, by the Founding Fathers. Individualistic western societies are built on the idea that no one knows our thoughts, desires or joys better than we do. Artificial intelligence (AI) will change this. It will know us better than we know ourselves. A government armed with AI could claim to know what its people truly want.

Power - nudges and hypernudges, affective contagion, postphenomenology of tech, personalisation, seduce coerce manipulate, persuasive tech.

[Pianobikes · Rating 4 out of 5]

“Las redes sociales pueden conformar tu identidad, incluso si no eres consciente de ello” ~ Filosofía política de la inteligencia artificial de Marck Coeckelbergh.

Ya hemos pasado del mundo Google al ChatGPT. Cada día que entramos en Internet, encontramos una herramienta nueva de Inteligencia Artificial (IA) que nos asombra con la capacidad que tiene para crear lo que le pidamos, sea texto, imagen, dibujos y cada día con mayor facilidad que para eso la vamos entrenando con nuestras peticiones.

Más allá de este mero entretenimiento, la
IA nos rodea y el algoritmo marca nuestro día a día e incluso nuestras elecciones (¿elecciones?). Y ahí nace el debate. ¿Hay justicia, equidad, libertad y democracia tras la IA? ¿Realmente elegimos libremente, con criterio, o permanecemos en nuestra parcela, polarizados en nuestros gustos, creencias y pensamientos precisamente porque el algoritmo nos enseña aquello con lo que interactuamos, dejando al margen un mundo contrapuesto y, por tanto, dejándonos aún más encerrados en nuestra parcela y negándonos una mayor amplitud de miras? ¿Cómo podemos romper con los sesgos del algoritmo? ¿Y mejorar su justicia, entre otros aspectos? Creo que este libro me ha despertado más curiosidad de la que ya tenía por el impacto social de la IA pero no desde el punto ético y moral de su aplicación –que ya se le presupone– sino desde un enfoque político que evite los sesgos y, por tanto, las discriminaciones positivas y negativas.

Es denso, condensado y requiere una lectura tranquila y profunda. Vamos, que si estás buscando un pasa páginas aquí no es. Pero es un libro con un contenido muy interesante para debatir, reflexionar y profundizar. Reconozco que me he hecho plantearme muchos aspectos de mi día a día; entre otras cosas no veré de igual manera el móvil que tengo en la mano y con el que estoy escribiendo esta reseña.

Leído gracias a la iniciativa #masacritica de Babelio.

[Fred Rose · Rating 4 out of 5]

This is a philosophy textbook which is pretty far field from my experience. Much of it reads like academic literature. But having said all that, I found it to be an extremely interesting book and much of the writing was clear and concise. While I wasn't familiar with some of the terms I'm certainly familiar with some of the writing from people like Zuboff about surveillance capitalism and looking at technology from a capitalist and equity viewpoint. I'm read this book because I'm teaching a course in the fall on AI and environmental policy. While it's easy to find examples of AI doing both good and bad things related to environment and policy, and AI issues with inequity; the more interesting things are some of the underlying issues which is what this book really gets at well. How it can perpetuate existing systems, some of the potential issues to personal freedom, power and democracy. This is what I want policy students to know and think about as they go forward. The actual mechanics of how AI applies to things changes almost every day so there's not much point in spending a lot of time on that but I do want them to think about some of these issues which are constant. Thinking about the ideology of technology and AI is important for policy makers (and makers of the tech).

[Marea sdp · Rating 4 out of 5]

Es una muy buena introducción para abordar el problema de la IA desde una óptica de filosofía política. Incluye autores y obras muy útiles para mapear la cuestión y su lectura es accesible, además de contar con una extensión muy asumible. Lo recomiendo para quienes estén interesados en la filosofía política, en la IA o en las dos.

[Marisa Duarte · Rating 5 out of 5]

This is one of my favorite books on the contemporary political crises emerging since the business and tech world began investing in AI as a panacea for all manner of social challenges. There are many flaws in contemporary writing on AI: 1) lack of understanding and explanation of the inner workings of machine learning, neural nets, models, and other fundamental components, 2) assumption that intelligent agents are accurate and that humans who use the interfaces are rational and sensible, 3) a downright harmful lack of social theoretical analysis or a shallow interpretation of social theory from 4 generations past when human-tech relations were entirely different, 4) avoidance of the problem of technocracy and capitalism, 5) avoidance of environmental and ecological analyses, 6) American exceptionalist arguments that hinge on securitization arguments (i.e., 'next space race.') Coeckelbergh is keenly aware of these flaws in the existing literature. Coeckelbergh has therefore written extensively on philosophical approaches to various human-technology relationships, and true to the rigorous methods of philosophical argument, takes one question at a time, considers challenges from the vantage point of diverse philosophical approaches, and from there identifies a thru-line that satisfies the conditions of the original inquiry. Coeckelbergh asks over 5 chapters, how do AIs signify and imply changes to freedom, equality and justice, democracy, power, and environmental concerns? Thick with citations setting artful summation of key philosophies of justice, freedom, rationality, environmental ethics, this volume is excellent for serious scholars of technological advancement and political power. In other words, if you must read Kurzweil or listen to Altman, treat the head-ache and irritation with some Coeckelbergh.

[Lachlan Sleight · Rating 4 out of 5]

Reads like an undergraduate-level introduction to the political philosophy of technology, in a good way!
The title is a bit misleading as I'd say only perhaps 50% of the book is actually about AI specifically, but that's only because Coeckelbergh seems to want to give a holistic background understanding of the issues, and so talking about technology more broadly is necessary to contextualize anything AI-specific. Is absolutely packed with references (the list of references is almost 20 pages long!) from a variety of perspectives and disciplines, so it's given me many useful springboards to jump into particular areas that I'd like to read more about!

[cellus · Rating 4 out of 5]

excellent introduction on the matter. in-depth enough to be engaging and open up new avenues of thinking and research.

[Sven · Rating 4 out of 5]

4/5

Interessante toepassing van een breed scala aan theoretische stromingen op AI

[Nanako]

Although I read this book for my Capstone summer assignment, I loved it because it got me thinking about many things. I think it was able to

[Geert Hofman · Rating 4 out of 5]

This book is currently one of only a few books on AI I know off that tackles the subjects usually associated with political philosophy. The book is rather introductory but gives many clues on where to go deeper into the matter at hand.

All topics are of high relevance for the future of AI and even mankind in general (freedom, equality and justice, democracy, power) but the last normal chapter (before the conclusion) in particular is intriguing and super-relevant these days: how do we take non-humans into account when thinking about the political consequences of developments related to artificial intelligence. This not only has to do with companion species like animals and plants, but even with entire ecosystems or, not to forget in this context, artificial lifeforms.

The worth of the book is in its breadth. It really is a guide to further literature. I personally would love to start perusing the bibliography to get a stronger grip on the important topics presented in this rather concise book.

[Andy Liu · Rating 3 out of 5]

I don't really have any political philosophy background, but this did a really good job of distilling relevant parts of political philosophy and identifying clear connections to AI. I thought the links between AI and Arendt, Foucault, and libertarian paternalism were especially good.