Artificial Intelligence: The Very Idea

by John Haugeland

"Machines who think—how utterly preposterous," huff beleaguered humanists, defending their dwindling turf. "Artificial Intelligence—it's here and about to surpass our own," crow techno-visionaries, proclaiming dominion. It's so simple and obvious, each side maintains, only a fanatic could disagree. Deciding where the truth lies between these two extremes is the main purpose of John Haugeland's marvelously lucid and witty book on what artificial intelligence is all about. Although presented entirely in non-technical terms, it neither oversimplifies the science nor evades the fundamental philosophical issues. Far from ducking the really hard questions, it takes them on, one by one. Artificial intelligence, Haugeland notes, is based on a very good idea, which might well be right, and just as well might not. That idea, the idea that human thinking and machine computing are "radically the same," provides the central theme for his illuminating and provocative book about this exciting new field. After a brief but revealing digression in intellectual history, Haugeland systematically tackles such basic questions What is a computer really? How can a physical object "mean" anything? What are the options for computational organization? and What structures have been proposed and tried as actual scientific models for intelligence? In a concluding chapter he takes up several outstanding problems and puzzles—including intelligence in action, imagery, feelings and personality—and their enigmatic prospects for solution.

Genres: Philosophy, Artificial Intelligence, Science, Nonfiction, Computer Science, Ebooks

299 pages, Paperback

First published January 1, 1985

Reviews

[Brian Powell · Rating 4 out of 5]

This is the book that put symbolic AI (so-called "good old fashioned AI") on the map. It's casual, lighthearted, and non-technical, but don't be fooled: there are deep issues here to ponder and chew on. It's also old, and so in no way gives an accurate portrayal of modern-day AI, but there's nothing particularly perishable here. Haugeland ponders the nature of intelligence and what a "thinking machine" might be capable of; he surveys a variety of formal computer architectures that might be called on to realize such a thinking machine (while implementations of these architectures have changed dramatically since the 80's, the architectures themselves remain much the same); and he enumerates the major stumbling blocks with contemporary (as of the 80's) AI. Indeed, some of these hurdles are technical and so surmountable with the advent of better algorithms, more computing power, and so on. But most of them are *conceptual*, requiring what seem to be much more significant leaps in the development of computational cognitive science.

Much of the emphasis is on creating machines that can act, decide, and reason like humans; conspicuously, there is virtually no discussion of creating machines that can *learn*. In contrast, much of modern AI is taken up with machine learning--teaching machines to complete some task, whether like a human or not. These tasks are typically not about reasoning like humans, but about perception (image recognition) or even imitation (conversational text generators) and so the recent promise of "connectionist" AI might appear to be a rather parallel effort. But, it occurs while reading this book that perhaps the execution of reasoned action (what symbolic AI as presented by Haugeland concerns itself with) and the learning of what's being reasoned about are not so separate, and that treating them as isolated "modules" is the wrong way to do things.

As a researcher in applied machine learning, I picked this book up hoping for a quaint introduction to the "old" ways of thinking about AI, but instead have been greatly inspired by how many of the "old" problems remain unsolved and by just how formidable these problems are.

[Andrew]

Obviously, a lot of chatter is happening around AI right now, and as Haugeland pointed out decades ago, a lot of it is pretty dumb. That being said, few people are advocating for the humbler GOFAI approach that Haugeland liked. And I've got my own skepticism about that (q.v. Hubert Dreyfus)

Of course, it might seem a bit silly to read a book about AI from more than 30 years ago. But keep in mind that I was interested in both the evolution of the idea of AI as well as what has stayed the same, and what differs. Because even if philosophy correlates to technology, it stays relevant for much longer than the technology itself.

It helps that Haugeland is a good storyteller, even if I have reservations about some of his concepts. I tested out his example of a non-AI translatable sentence from back in the day, "the box is in the pen." Let's see how it went in Google Translate English-to-French, about as easy a language pairing as I can imagine, given both their cognate idiom structures and widespread use for the AI to draw from...

"La boîte est dans le stylo."

Whoopsies! Let's give him a hand...

"The pig is in the pen. The box is also in the pen."

"Le cochon est dans l'enclos." (OK so far!) "La boîte est aussi dans le stylo."

So yeah Googs, your neural network can suck my fleshy human dick.

[Chant · Rating 5 out of 5]

More so a book about the philosophical implications of AI and how it failed before with GOFAI. People aware of Haugeland’s philosophical interests, one can tell that it is heavily influenced by Heideggerian philosophy and probably Dreyfus’s interpretation of Heidegger.

Dated? Yes, books on technology tend to go dated but the problem (to my knowledge) is still there (or at least partially there).

[Vinay Sharma · Rating 3 out of 5]

A really good introduction to the history and development of the field itself (up until the 80s of course). The primary thesis of rational symbolic manipulation is a very gentle and well thought out argument. However, it does lack any sort of exploratory or speculative prowess, but that’s not really analytic philosophy’s wheel house (until it is, then it’s great!). That all being said, it’s an enjoyable book; well written and extremely easy to follow.

[Mohir · Rating 5 out of 5]

Somewhat outdated, to say the least
Still, a pleasure to read,
and for sure some principles can be extracted

[Kelly Kemp]

for school :(

[Andrea Nicaretta · Rating 3 out of 5]

prime considerazioni socio-filosofiche su IA (letto in italiano)

[John · Rating 4 out of 5]

I'm not going to rate a textbook.

This was a nice history of the field of Artificial Intelligence. The last two chapters seemed to finally get into the more practical application of the theories introduced, but the book ended with me wanting more.

[jeremiah]

Haugeland's a fantastic writer of philosophy. Great introduction that demonstrates how the discipline of artificial intelligence can lead us to a better understanding of what constitutes thought and how we interpret our experience.

[Nat · Rating 5 out of 5]

My favorite part of this book are the information "boxes" that are scattered throughout, like the meaning of "meaning", and the role of truth in interpretation.

[Jesse · Rating 3 out of 5]

Good summary of AI without too much hype. Has a definite shelf life though.

[Heather]

By this point, it's probably more of a period piece...