What do you think?
Rate this book
Mind over machine: The power of human intuition and expertise in the era of the computer by Hubert L Dreyfus, Stuart E. Dreyfus (1986) Hardcover
Defining the limits of computer technology, the authors make a compelling case that binary logic will always be inferior to human intuitive ability. A stunning reaffirmation of human intelligence.
Hardcover
First published January 1, 1987
18 people are currently reading
791 people want to read
About the author
Hubert L. Dreyfus
48 books192 followersHubert Lederer Dreyfus was professor of philosophy at the University of California, Berkeley, where his interests include phenomenology, existentialism, the philosophy of psychology and literature, and the philosophical implications of artificial intelligence.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 11 of 11 reviews
July 12, 2019
It was very interesting to read this book so long after its publication in 1987. It was enlightening to see that much of the 'hype' surrounding artificial intelligence was already in place at the time, and many of the predictions we see today regarding the great strides that AI will make in the next two decades were made as far back as the 1960s, by serious AI researchers.
Given that this book is more than 30 years old, many of the factual aspects are out of date - there is a focus on contemporaneous technologies, many of which are now obsolete. These parts can be read almost as a primary historical source - what was the state of AI research at the time, where have we come in the subsequent years, and what issues, apparent 30 years ago, continue to plague the field? An interesting aspect to note is that the 'neural net/machine learning/deep learning' approach, so ubiquitous today that is seems almost terminologically interchangeable with AI - at least to an industry outsider such as myself - is one of many approaches describes in the book. Clearly, at the time it had not emerged as the champion AI technique it seems to be today.
The parts of the book which focus on the state of 1980s AI research are then clearly at risk of redundancy, and many of the assertions made have not aged particularly well. The author appears to have an interest in chess, and many paragraphs cover the application of AI to beating human chess players. At the time, no AI program had convincingly demonstrated superiority to the world's best chess players, and the author suggests that one never might. A decade after the book's publication, IBM's Deep Blue beat the reigning world champion Garry Kasparov in a six-game match - not without controversy, it must be noted - and subsequent research efforts have developed AI programs capable of beating the world's best players at many games - most recently, Go and six-man Hold 'Em poker. The author is also pessimistic on the viability of computer vision applications, most conspicuously on the topic of facial recognition. We have seen great strides in facial recognition technology in recent years - for better of worse - and as such this is another prediction which has not aged well.
Despite this, I do believe that the book contains a great deal of timeless wisdom and insight. For one, it illustrates how overlapping are the areas of neuroscience, philosophy, and artificial intelligence research. For one such as myself, whose exposure to AI research is limited to that covered in the media and the odd machine learning-focused practical workshop, it can seem that AI is a purely technical issue - provide the model with better data, train it long and well enough, and eventually we can train computers to perform any number of tasks. But this book explains very well the more fundamental issues at play. I feel that the author made a convincing case in these parts.
I can recommend this book but suggest that it be supplemented with more up-to-date accounts of the state of AI research.
Given that this book is more than 30 years old, many of the factual aspects are out of date - there is a focus on contemporaneous technologies, many of which are now obsolete. These parts can be read almost as a primary historical source - what was the state of AI research at the time, where have we come in the subsequent years, and what issues, apparent 30 years ago, continue to plague the field? An interesting aspect to note is that the 'neural net/machine learning/deep learning' approach, so ubiquitous today that is seems almost terminologically interchangeable with AI - at least to an industry outsider such as myself - is one of many approaches describes in the book. Clearly, at the time it had not emerged as the champion AI technique it seems to be today.
The parts of the book which focus on the state of 1980s AI research are then clearly at risk of redundancy, and many of the assertions made have not aged particularly well. The author appears to have an interest in chess, and many paragraphs cover the application of AI to beating human chess players. At the time, no AI program had convincingly demonstrated superiority to the world's best chess players, and the author suggests that one never might. A decade after the book's publication, IBM's Deep Blue beat the reigning world champion Garry Kasparov in a six-game match - not without controversy, it must be noted - and subsequent research efforts have developed AI programs capable of beating the world's best players at many games - most recently, Go and six-man Hold 'Em poker. The author is also pessimistic on the viability of computer vision applications, most conspicuously on the topic of facial recognition. We have seen great strides in facial recognition technology in recent years - for better of worse - and as such this is another prediction which has not aged well.
Despite this, I do believe that the book contains a great deal of timeless wisdom and insight. For one, it illustrates how overlapping are the areas of neuroscience, philosophy, and artificial intelligence research. For one such as myself, whose exposure to AI research is limited to that covered in the media and the odd machine learning-focused practical workshop, it can seem that AI is a purely technical issue - provide the model with better data, train it long and well enough, and eventually we can train computers to perform any number of tasks. But this book explains very well the more fundamental issues at play. I feel that the author made a convincing case in these parts.
I can recommend this book but suggest that it be supplemented with more up-to-date accounts of the state of AI research.
August 9, 2020
3.5/5
Written in 1987, the things talked in the book are still very relevant today. His model for skill acquisition and expertise is extremely well written. They also deal with some very interesting topics surrounding technology like it's use in education and management. But these topics were fairly dry in the book. Some of the sections in the book were extremely boring to read (and very out of date), but it's still fun to understand the history of the subject and read about it's evolution.
Takeaways from the book are the five stage model of learning to expert. The novice, expert beginner, Competent, Proficient and Expert. In each step of the journey, the decision making shifts from analytical towards intuitive and the commitment shown increases as well. The novice uses rule based reasoning to get to answers and the expert makes almost does no rule based analysis. But this means that when an expert says that they feel a certain way and they can't explain why, it's because they really don't know why. They've seen so many cases over the years and often the causality and the rules for choosing a particular path might not be a straight forward to explain. But this does put an expert into the dealing with the problem of cognitive tunnelling.
Computers and technology should be used where human expertise and intuition is at a minimum (ideally zero) and bots by far out perform humans here. If you have an operation which can be performed using a set of rules, then bots will excel here and it works for the human as well.
Lastly, intuition is that ability of yours to explain what you understand without knowing "why" or "how" you arrived at it. A good example trying to transfer the knowledge of riding a bicycle to another soul. Although you know how, you can never express it in words.
Written in 1987, the things talked in the book are still very relevant today. His model for skill acquisition and expertise is extremely well written. They also deal with some very interesting topics surrounding technology like it's use in education and management. But these topics were fairly dry in the book. Some of the sections in the book were extremely boring to read (and very out of date), but it's still fun to understand the history of the subject and read about it's evolution.
Takeaways from the book are the five stage model of learning to expert. The novice, expert beginner, Competent, Proficient and Expert. In each step of the journey, the decision making shifts from analytical towards intuitive and the commitment shown increases as well. The novice uses rule based reasoning to get to answers and the expert makes almost does no rule based analysis. But this means that when an expert says that they feel a certain way and they can't explain why, it's because they really don't know why. They've seen so many cases over the years and often the causality and the rules for choosing a particular path might not be a straight forward to explain. But this does put an expert into the dealing with the problem of cognitive tunnelling.
Computers and technology should be used where human expertise and intuition is at a minimum (ideally zero) and bots by far out perform humans here. If you have an operation which can be performed using a set of rules, then bots will excel here and it works for the human as well.
Lastly, intuition is that ability of yours to explain what you understand without knowing "why" or "how" you arrived at it. A good example trying to transfer the knowledge of riding a bicycle to another soul. Although you know how, you can never express it in words.
July 4, 2020
I've read this book primarily for understanding of Dreyfus model of skill acquisition, of how learners acquire skills through formal instruction and practicing, but I was truly impressed on the authors point of view on the role of Artificial intelligence in our society, it's potentialities and barriers.
More than approaching AI from a merely technical point of view, they use the theories from Plato, Socrates, Aristotles, Descartes and many other philosophers in order to analyze the society's way of thinking and reacting, which is something more than valuable nowadays, with people discussing just the bits and bytes.
More than approaching AI from a merely technical point of view, they use the theories from Plato, Socrates, Aristotles, Descartes and many other philosophers in order to analyze the society's way of thinking and reacting, which is something more than valuable nowadays, with people discussing just the bits and bytes.
October 10, 2020
The text is dated, which of course makes sense! It was published in 1986, so of course the advancements in artificial intelligence has grown.
However, as the book points out, the problems of "common-sense" or general "know-how", are still very much relevant for the 21st century. I however do have a word of caution, if you've read Hubert Dreyfus's other book on AI "What Computers Can't Do" it'll be more or less the same fair in this book (which he co-wrote with his brother Stuart).
However, as the book points out, the problems of "common-sense" or general "know-how", are still very much relevant for the 21st century. I however do have a word of caution, if you've read Hubert Dreyfus's other book on AI "What Computers Can't Do" it'll be more or less the same fair in this book (which he co-wrote with his brother Stuart).
February 23, 2023
For decades now we’ve been hearing about the “rise of the machines,” that Skynet was going to be online any day now, and we’d be slaves of robots.
In the decades since these wild prognostications first got prognosticated, AI has improved somewhat. Computer actuators integrated with damaged nerve tissue have shown that AI can help badly injured people during rehab. Facial recognition software has also improved, though it can still be foiled by those who stay one step ahead of the cutting edge.
Aside from these exceptions and a few others, though, it’s clear that the machine “singularity” is being oversold. Some of those doing the overselling have an agenda, like scaring up capital, or fearmongering people into buying their latest hardback jeremiad. Others are well-meaning victims of what author and philosopher Hubert Dreyfus calls “the first step fallacy.” Early technological leaps in a field give those in that field the false sense that this exponential rate of progress is just going to continue forever. But, as the old proverb has it, “Trees do not grow to the sky.”
Mind Over Machine by the aforementioned Dreyfus (and another Dreyfus, Stuart E.) does a good job of telling us where AI stands now. My copy’s from 1988, which would make it seem dated in a field as rapidly advancing as computer science, if, that is, the Dreyfus premise weren’t right. And I believe it is.
In brief, the authors hold that skills acquisition is a five step process, from the novice level to expertise. This spectrum can be applied broadly to everything from riding a bicycle to piloting an airplane, to assessing potential loan applicants for their worthiness. The first few stages of proficiency involve rule-learning and a “knowledge about” various subjects. The final stage, however, involves a much more plastic and hard-to-define “knowledge of.” This involves a tricky combination of intuition and experience that can’t quite be articulated in words, or taught to a classroom.
Ask a master of their craft how they did something, and they’re likely to be vague, evasive, or to explain things in terms so simple as to be useless. The master boxer just “knew” his opponent was going to throw that punch; the chicken sexor (yes, these actually exist) just knew this brood was going to produce X amount of pullets and Y amount of cockerels.
The authors argue that this fifth stage of expertise—containing expertise (and perhaps other ineffable traits like consciousness)—is thus far unattained by machines. This is not to say that there are not areas where machines haven’t proven themselves superior to humans, or that they aren’t welcome aids in certain diagnostic activities. Machines have proven themselves superior at all kinds of quantitative operations, from searching for oil to fine positioning cargo into the bays of shipping vessels. Qualitatively, though, man appears to have it all over machine, and is likely to continue to do so for a long time.
The argument put forth in the book, while convincing, is likely to be greeted as a kind of letdown by those fed a steady diet of SF fear porn. Yes, we all dread the rise of the robots, but we’ve somehow all been trained—cross-culturally—to accept our eventually displacement by them. It’s become a strange form of high-tech solipsism, that because we are godlike we are capable of imbuing our tools with the consciousness to eventually destroy us. Nietzsche pronounced God dead and now it’s HAL’s turn to read our obituary to us while we’re sealed inside the space pod.
Mind Over Machine suggests we are likely to be safe from such a threat as long as this fifth stage of intuitive expertise is never attained by machines. The real threat, the authors suggest, comes from those overselling the robot revolution, so intent on fulfilling the prophesy that they may automate decisions best left in human hands. This is a fear that isn’t without basis, as the Stanislav Petrov incident of 1983 proved. Petrov is famous for ignoring a launch order while on duty at a Russian missile silo, despite the machine’s insistence that American warheads were incoming. Petrov, however, overruled the machine, suspecting an error somewhere along the line.
Recommended, for the openminded computer scientist and the general lay reader who wants a non-polemical, non-panicked evaluation of where AI was, is, and will likely remain.
In the decades since these wild prognostications first got prognosticated, AI has improved somewhat. Computer actuators integrated with damaged nerve tissue have shown that AI can help badly injured people during rehab. Facial recognition software has also improved, though it can still be foiled by those who stay one step ahead of the cutting edge.
Aside from these exceptions and a few others, though, it’s clear that the machine “singularity” is being oversold. Some of those doing the overselling have an agenda, like scaring up capital, or fearmongering people into buying their latest hardback jeremiad. Others are well-meaning victims of what author and philosopher Hubert Dreyfus calls “the first step fallacy.” Early technological leaps in a field give those in that field the false sense that this exponential rate of progress is just going to continue forever. But, as the old proverb has it, “Trees do not grow to the sky.”
Mind Over Machine by the aforementioned Dreyfus (and another Dreyfus, Stuart E.) does a good job of telling us where AI stands now. My copy’s from 1988, which would make it seem dated in a field as rapidly advancing as computer science, if, that is, the Dreyfus premise weren’t right. And I believe it is.
In brief, the authors hold that skills acquisition is a five step process, from the novice level to expertise. This spectrum can be applied broadly to everything from riding a bicycle to piloting an airplane, to assessing potential loan applicants for their worthiness. The first few stages of proficiency involve rule-learning and a “knowledge about” various subjects. The final stage, however, involves a much more plastic and hard-to-define “knowledge of.” This involves a tricky combination of intuition and experience that can’t quite be articulated in words, or taught to a classroom.
Ask a master of their craft how they did something, and they’re likely to be vague, evasive, or to explain things in terms so simple as to be useless. The master boxer just “knew” his opponent was going to throw that punch; the chicken sexor (yes, these actually exist) just knew this brood was going to produce X amount of pullets and Y amount of cockerels.
The authors argue that this fifth stage of expertise—containing expertise (and perhaps other ineffable traits like consciousness)—is thus far unattained by machines. This is not to say that there are not areas where machines haven’t proven themselves superior to humans, or that they aren’t welcome aids in certain diagnostic activities. Machines have proven themselves superior at all kinds of quantitative operations, from searching for oil to fine positioning cargo into the bays of shipping vessels. Qualitatively, though, man appears to have it all over machine, and is likely to continue to do so for a long time.
The argument put forth in the book, while convincing, is likely to be greeted as a kind of letdown by those fed a steady diet of SF fear porn. Yes, we all dread the rise of the robots, but we’ve somehow all been trained—cross-culturally—to accept our eventually displacement by them. It’s become a strange form of high-tech solipsism, that because we are godlike we are capable of imbuing our tools with the consciousness to eventually destroy us. Nietzsche pronounced God dead and now it’s HAL’s turn to read our obituary to us while we’re sealed inside the space pod.
Mind Over Machine suggests we are likely to be safe from such a threat as long as this fifth stage of intuitive expertise is never attained by machines. The real threat, the authors suggest, comes from those overselling the robot revolution, so intent on fulfilling the prophesy that they may automate decisions best left in human hands. This is a fear that isn’t without basis, as the Stanislav Petrov incident of 1983 proved. Petrov is famous for ignoring a launch order while on duty at a Russian missile silo, despite the machine’s insistence that American warheads were incoming. Petrov, however, overruled the machine, suspecting an error somewhere along the line.
Recommended, for the openminded computer scientist and the general lay reader who wants a non-polemical, non-panicked evaluation of where AI was, is, and will likely remain.
May 5, 2025
«آرزواندیشی به احتمال همواره روابط ما با فناوری را پیچیده کرده است؛ اما میتوان با اطمینان گفت که پیش از رایانه و پیش از بمب، پیچیدگیها هیچ گاه به اندازه امروز خطرناک نبودند و آرزواندیشی نیز هرگز این چنین خیال پردازانه نبود.» (ص ۳۱)
هیوبرت دریفوس، فیلسوف معاصر آمریکایی و شاگرد هایدگر و گادامر، تو دهه ۷۰ به پروژه مطالعاتیای دعوت میشه که دو نفر از پیشگامان صنعت هوش مصنوعی، یعنی الن نوئل و هربرت سایمون، تو اون پروژه حضور داشتند.
دریفوس که به عنوان کارشناس سیستم شناختی انسان تو اون پروژه مشارکت میکرده، بعد از مطالعه مقالههای تولید شده توسط پژوهشگران پروژه، متوجه نکاتی میشه که عاشقان هوش مصنوعی چشماشون رو به روی اونا بستن. دریفوس انتقادات خودش رو ذیل چهار پیشفرض هوش مصنوعی دستهبندی میکنه:
1. پیشفرض روانشناختی:
این دیدگاه فکر میکنه ذهن ما مثل یه ماشین حساب یا برنامه کامپیوتریه که با یهسری قواعد کار میکنه. مثلاً اگه فلان چیزو بدونی و فلان قانونو بهکار ببری، به فلان نتیجه میرسی. دریفوس میگه اما ما همیشه اینطوری فکر نمیکنیم؛ بیشتر وقتا ناخودآگاه، شهودی و بدون دنبالکردن قانون مشخص عمل میکنیم.
2. پیشفرض زیستشناختی:
این فرضیه مغز آدم رو با کامپیوتر مقایسه میکنه؛ یعنی فکر میکنن اگه ساختار مغز رو بفهمیم، میتونیم با یه ماشین همون کارو انجام بدیم. ولی دریفوس میگه مغز آدم فقط یه پردازنده نیست، بلکه با بدن، احساس، تجربه و محیطش گره خورده.
3. پیشفرض معرفتشناختی:
طبق این فرض، دانش یعنی یهسری اطلاعات مشخص که میشه اونا رو نوشت، ذخیره کرد یا پردازش کرد. مثلا�� مثل کتاب قانون یا دستور آشپزی. ولی دریفوس میگه خیلی از چیزایی که میدونیم، تو ذهنمون به این شکل نیست؛ ما خیلی چیزها رو «بلدیم» بدون اینکه بتونیم دقیقاً توضیحش بدیم.
4. پیشفرض وجودشناختی:
این یعنی میخوان کل تجربهی انسانبودن رو با مدلهای اطلاعاتی یا محاسباتی نشون بدن؛ مثلا انگار بودن توی دنیا، مثل یه بازی کامپیوتریه با قوانین مشخص. ولی دریفوس با هایدگر میگه انسانبودن یه چیز خیلی پیچیدهتره، چون همیشه تو یه زمینهی خاص، با تاریخچه، احساس و بدن زندگی میکنیم، نه صرفاً با اطلاعات.
اون چیزی که تو طول کتاب توجه من رو جلب کرد و علت عمده خوندن این کتاب اون مسئله بود، توجه دریفوس به عقل عملی بود. هر چند ایشون تقریبا هیچ جا اسمی از فرونسیس ارسطویی نمیاره -و واقعا نمیدونم چرا؟!- اما تقریبا میشه گفت که نقد اصلیش به هوش مصنوعی ذیل همین عنوان ساماندهی میشه. توضیح این مطلب مفصله اگه کسی طالب بود عرض میکنم خدمتش.
.
پینوشت اول: من مطالب و انتقادات دریفوس رو به چت GPT دادم و همونطور که تو عکس مشاهده میکنید انتقادات رو قبول داشت و پذیرفت با وجود گذشتن تقریبا پنجاه سال از نوشتن این کتاب هنوز انتقادات دریفوس به هوش مصنوعی وارده.
پینوشت دوم: مترجم کتاب، جناب آقای خوشنویس، مطالب کتاب رو به صورت خلاصه و بسیار شیواتر از بنده تو پیشگفتار مترجم توضیح دادن. فایل پیدیاف این بخش رو تو کانال تلگرامی خودم @mangooon بارگزاری کردم اگر از دوستان کسی طالب مطالعه بیشتر بود میتونه به اون رجوع کنه.
هیوبرت دریفوس، فیلسوف معاصر آمریکایی و شاگرد هایدگر و گادامر، تو دهه ۷۰ به پروژه مطالعاتیای دعوت میشه که دو نفر از پیشگامان صنعت هوش مصنوعی، یعنی الن نوئل و هربرت سایمون، تو اون پروژه حضور داشتند.
دریفوس که به عنوان کارشناس سیستم شناختی انسان تو اون پروژه مشارکت میکرده، بعد از مطالعه مقالههای تولید شده توسط پژوهشگران پروژه، متوجه نکاتی میشه که عاشقان هوش مصنوعی چشماشون رو به روی اونا بستن. دریفوس انتقادات خودش رو ذیل چهار پیشفرض هوش مصنوعی دستهبندی میکنه:
1. پیشفرض روانشناختی:
این دیدگاه فکر میکنه ذهن ما مثل یه ماشین حساب یا برنامه کامپیوتریه که با یهسری قواعد کار میکنه. مثلاً اگه فلان چیزو بدونی و فلان قانونو بهکار ببری، به فلان نتیجه میرسی. دریفوس میگه اما ما همیشه اینطوری فکر نمیکنیم؛ بیشتر وقتا ناخودآگاه، شهودی و بدون دنبالکردن قانون مشخص عمل میکنیم.
2. پیشفرض زیستشناختی:
این فرضیه مغز آدم رو با کامپیوتر مقایسه میکنه؛ یعنی فکر میکنن اگه ساختار مغز رو بفهمیم، میتونیم با یه ماشین همون کارو انجام بدیم. ولی دریفوس میگه مغز آدم فقط یه پردازنده نیست، بلکه با بدن، احساس، تجربه و محیطش گره خورده.
3. پیشفرض معرفتشناختی:
طبق این فرض، دانش یعنی یهسری اطلاعات مشخص که میشه اونا رو نوشت، ذخیره کرد یا پردازش کرد. مثلا�� مثل کتاب قانون یا دستور آشپزی. ولی دریفوس میگه خیلی از چیزایی که میدونیم، تو ذهنمون به این شکل نیست؛ ما خیلی چیزها رو «بلدیم» بدون اینکه بتونیم دقیقاً توضیحش بدیم.
4. پیشفرض وجودشناختی:
این یعنی میخوان کل تجربهی انسانبودن رو با مدلهای اطلاعاتی یا محاسباتی نشون بدن؛ مثلا انگار بودن توی دنیا، مثل یه بازی کامپیوتریه با قوانین مشخص. ولی دریفوس با هایدگر میگه انسانبودن یه چیز خیلی پیچیدهتره، چون همیشه تو یه زمینهی خاص، با تاریخچه، احساس و بدن زندگی میکنیم، نه صرفاً با اطلاعات.
اون چیزی که تو طول کتاب توجه من رو جلب کرد و علت عمده خوندن این کتاب اون مسئله بود، توجه دریفوس به عقل عملی بود. هر چند ایشون تقریبا هیچ جا اسمی از فرونسیس ارسطویی نمیاره -و واقعا نمیدونم چرا؟!- اما تقریبا میشه گفت که نقد اصلیش به هوش مصنوعی ذیل همین عنوان ساماندهی میشه. توضیح این مطلب مفصله اگه کسی طالب بود عرض میکنم خدمتش.
.
پینوشت اول: من مطالب و انتقادات دریفوس رو به چت GPT دادم و همونطور که تو عکس مشاهده میکنید انتقادات رو قبول داشت و پذیرفت با وجود گذشتن تقریبا پنجاه سال از نوشتن این کتاب هنوز انتقادات دریفوس به هوش مصنوعی وارده.
پینوشت دوم: مترجم کتاب، جناب آقای خوشنویس، مطالب کتاب رو به صورت خلاصه و بسیار شیواتر از بنده تو پیشگفتار مترجم توضیح دادن. فایل پیدیاف این بخش رو تو کانال تلگرامی خودم @mangooon بارگزاری کردم اگر از دوستان کسی طالب مطالعه بیشتر بود میتونه به اون رجوع کنه.
January 15, 2020
Read this while working on the concept of knowing back in the day.
May 29, 2020
A seminal book for anyone interested in the field of AI. Dated but still highly relevant to current debates about the potential and limitations of artificial intelligence.
February 4, 2023
第二章写了一个从新手到专家的阶段理论，成为了当代专业发展和专业学习的 最初文本。把规则和能力做了区分，同时认为能力需要「情境认识」，以及某种规范性的「责任」，才能上升到know-how的层次。而后者是不可言说，只能［身体把握］的能力。这种乐观遍布全书。例子，尤其是对当时人工智能缺陷的讨论，也很有趣。
April 13, 2024
Vigtig bog om de tidlige erfaringer med AI selv om den idag fremstår lidt outdated
November 26, 2020
It reads today as it did 40 years ago. AI hype and hubris has not been corrected: then again, there's a lot of money in maintaining its mistakes alive. Human knowledge acquisition is not one problem among many: it is THE problem at the center of all knowledge. Some of the smartest people in History have grappled with it without reaching any definitive conclusions, but here comes a bunch of MIT nerds and Silicon Valley billionaires thinking they can solve it with a handful of mathematical party tricks and billions of dollars. And no matter how many times they get it wrong, they don't concede or take up reading Aristotle.
The AI community complained that Dreyfus was derisive and that constructive criticism instead of sarcasm may have had more of an impact. I doubt it. What drives Dreyfus' sarcasm is his outrage at seeing these philosophical amateurs delving into a well known field of study with complete disregard for anything that came before and making obvious and predictable mistakes as a result. Apparently, Edward Feigenbaum said of Dreyfus: "What does he offer us? Phenomenology! That ball of fluff. That cotton candy!" Whatever tone Dreyfus may have used, these people would have refused to engage because they wanted to see themselves as engineers (as serious and precise people of science) and not as philosophers, whom they viewed with disdain as an inferior species.
And that has not changed either.
The AI community complained that Dreyfus was derisive and that constructive criticism instead of sarcasm may have had more of an impact. I doubt it. What drives Dreyfus' sarcasm is his outrage at seeing these philosophical amateurs delving into a well known field of study with complete disregard for anything that came before and making obvious and predictable mistakes as a result. Apparently, Edward Feigenbaum said of Dreyfus: "What does he offer us? Phenomenology! That ball of fluff. That cotton candy!" Whatever tone Dreyfus may have used, these people would have refused to engage because they wanted to see themselves as engineers (as serious and precise people of science) and not as philosophers, whom they viewed with disdain as an inferior species.
And that has not changed either.
Displaying 1 - 11 of 11 reviews