Marvin Minsky, fully Marvin Lee Minsky

Marvin
Minsky, fully Marvin Lee Minsky
1927

American Cognitive Scientist in the field of Artificial Intelligence, Co-Founder of the Massachusetts Institute of Technology's AI Laboratory, Author

Author Quotes

A computer is like a violin. You can imagine a novice trying ?rst a phonograph and then a violin. The latter, he says, sounds terrible. That is the argument we have heard from our humanists and most of our computer scientists. Computer programs are good, they say, for particular purposes, but they aren?t ?exible. Neither is a violin, or a typewriter, until you learn how to use it.

For avoiding nonsense in general, we might accumulate millions of censors. For all we know, this "negative meta-knowledge" -- about patterns of thought and inference that have been found defective or harmful -- may be a large portion of all we know.

If explaining minds seems harder than explaining songs, we should remember that sometimes enlarging problems makes them simpler! The theory of the roots of equations seemed hard for centuries within its little world of real numbers, but it suddenly seemed simple once Gauss exposed the larger world of so-called complex numbers. Similarly, music should make more sense once seen through listeners' minds.

Once when I was standing at the base, they started rotating the set and a big, heavy wrench fell down from the 12 o'clock position of the set, and got buried in the ground a few feet from me. I could have been killed!

The basic idea in case-based, or CBR, is that the program has stored problems and solutions. Then, when a new problem comes up, the program tries to find a similar problem in its database by finding analogous aspects between the problems.

We wanted to solve robot problems and needed some vision, action, reasoning, planning, and so forth. We even used some structural learning, such as was being explored by Patrick Winston.

A memory should induce a state through which we see current reality as an instance of the remembered event?or equivalently, see the past as an instance of the present. ...the system can perform a computation analogous to one from the memorable past, but sensitive to present goals and circumstances.

For generations, scientists and philosophers have tried to explain ordinary reasoning in terms of logical principles?with virtually no success. I suspect this enterprise failed because it was looking in the wrong direction: common sense works so well not because it is an approximation of logic; logic is only a small part of our great accumulation of different, useful ways to chain things together.

If you just have a single problem to solve, then fine, go ahead and use a neural network. But if you want to do science and understand how to choose architectures, or how to go to a new problem, you have to understand what different architectures can and cannot do.

One reason why the mind-brain problem has always seemed mysterious is that minds seem to us so separate from their physical embodiments. Why do we find it so easy to imagine the same mind being moved to a different body or brain - or even existing by itself? One reason could be that concerns about minds are mainly concerns about changes in states - and these do not often have much to do with the natures of those states themselves. From a functional or procedural viewpoint, we often care only about how each agent changes state in response to the actions upon it of other agents. This is why we so often can discuss the organization of a community without much concern for the physical constitution of its members. It is the same inside a computer; it is only signals representing changes that matter, whereas we have no reason to be concerned with properties that do not change. Consider that it is just those properties of physical objects that change the least - such as their colors, sizes, weights, or shapes - that, naturally, are the easiest to sense. Yet these, precisely because they don't change, are the ones that matter least of all, in computational processes. So naturally minds seem detached from the physical. In regard to mental processes, it matters not what the parts of brains are; it only matters what they do--and what they are connected to.

The way the mathematics game is played, most variations lie outside the rules, while music can insist on perfect canon or tolerate a casual accompaniment.

We'll show you that you can build a mind from many little parts, each mindless by itself.

A related reason why the mind-brain problem seems hard is that we all believe in having a Self - some sort of compact, point-like entity that somehow knows what's happening throughout a vast and complex mind. It seems to us that this entity persists through our lives in spite of change. This feeling manifests itself when we say "I think" rather than "thinking is happening", or when we agree that "I think therefore I am," instead of "I think, therefore I change". Even when we recognize that memories must change our minds, we feel that something else stays fixed - the thing that has those memories. In chapter 4 of The Society of Mind[l] I argue that this sense of having a Self is an elaborately constructed illusion - albeit one of great and practical value. Our brains are endowed with machinery destined to develop persistent self-images and to maintain their coherence in the face of continuous change. But those changes are substantial, too; your adult mind is not very like the one mind you had in infancy. To be sure, you may have changed much since childhood - but if one succeeds, in later life, to manage to avoid much growth, that poses no great mystery.

General fiction is pretty much about ways that people get into problems and screw their lives up. Science fiction is about everything else.

Imagine what it would be like if TV actually were good. It would be the end of everything we know.

One's present personality cannot share all the thoughts of one's older personalities?and yet it has some sense that they exist. This is one reason why we feel that we possess an inner Self?a sort of ever-present person-friend, inside the mind, whom we can always ask for help.

Theorems often tell us complex truths about the simple things, but only rarely tell us simple truths about the complex ones. To believe otherwise is wishful thinking or "mathematics envy."

What connects the mind to the world? This problem has always caused conflicts between physics, psychology, and religion. In the world of Newton's mechanical laws, every event was entirely caused by what had happened earlier. There was simply no room for anything else. Yet common sense psychology said that events in the world were affected by minds: people could decide what occurred by using their freedom of will. Most religions concurred in this, although some preferred to believe in schemes involving divine predestination. Most theories in psychology were designed to support deterministic schemes, but those theories were usually too weak to explain enough of what happens in brains. In any case, neither physical nor psychological determinism left a place for the freedom of will. The situation appeared to change when, early in this century, some physicists began to speculate that the uncertainty principle of quantum mechanics left room for the freedom of will. What attracted those physicists to such views? As I see it, they still believed in freedom of will as well as in quantum uncertainty--and these subjects had one thing in common: they both confounded those scientists' conceptions of causality. But I see no merit in that idea because probabilistic uncertainty offers no genuine freedom, but merely adds a capricious master to one that is based on lawful rules. Nonetheless, quantum uncertainty does indeed play a critical role in the function of brain. However, this role is neither concerned with trans-world connections nor with freedom of will. Instead, and paradoxically, it is just those quantized atomic states that enable us to have certainty! This may surprise those who have heard that Newton's laws were replaced by ones in which such fundamental quantities as location, speed, and even time, are separately indeterminate. But although those statements are basically right, their implications are not what they seem - but almost exactly the opposite. For it was the planetary orbits of classical mechanics that were truly undependable - whereas the atomic orbits of quantum mechanics are much more predictably reliable. To explain this, let us compare a system of planets orbiting a star, in accord with the laws of classical mechanics, with a system of electrons orbiting an atomic nucleus, in accord with quantum mechanical laws. Each consists of a central mass with a number of orbiting satellites. However, there are fundamental differences. In a solar system, each planet could be initially placed at any point, and with any speed; then those orbits would proceed to change. Each planet would continually interact with all the others by exchanging momentum. Eventually, a large planet like Jupiter might even transfer enough energy to hurl the Earth into outer space. The situation is even less stable when two such systems interact; then all the orbits will so be disturbed that even the largest of planets may leave. It is a great irony that so much chaos was inherent in the old, deterministic laws. No stable structures could have evolved from a universe in which everything was constantly perturbed by everything else. If the particles of our universe were constrained only by Newton's laws, there could exist no well defined molecules, but only drifting, featureless clouds. Our parents would pass on no precious genes; our bodies would have no separate cells; there would not be any animals at all, with nerves, synapses, and memories. In contrast, chemical atoms are actually extremely stable because their electrons are constrained by quantum laws to occupy only certain separate levels of energy and momentum. Consequently, except when the temperature is very high, an atomic system can retain the same state for decillions of years, with no change whatever. Furthermore, combinations of atoms can combine to form configurations, called molecules, that are also confined to have definite states. Although those systems can change suddenly and unpredictably, those events may not happen for billions of years during which there is absolutely no change at all. Our stability comes from those quantum fields, by which everything is locked into place, except during moments of clean, sudden change. It is only because of quantum laws that what we call things exist at all, or that we have genes to specify brains in which memories can be maintained - so that we can have our illusions of will.

All intelligent persons also possess some larger-scale frame-systems whose members seemed at first impossibly different -- like water with electricity, or poetry with music. Yet many such analogies -- along with the knowledge of how to apply them -- are among our most powerful tools of thought. They explain our ability sometimes to see one thing -- or idea -- as though it were another, and thus to apply knowledge and experience gathered in one domain to solve problems in another. It is thus that we transfer knowledge via the paradigms of Science. We learn to see gases and fluids as particles, particles as waves, and waves as envelopes of growing spheres.

Get the mind into the (partial) state that solved the old problem; then it might handle the new problem in the "same way."

In general we are least aware of what our minds do best.

Only the surface of reason is rational. I don't mean that understanding emotion is easy, only that understanding reason is probably harder.

There are three basic approaches to AI: Case-based, rule-based, and connectionist reasoning.

What do brains do? Doing means changing. Whenever we learn or 'change our minds', our brains are engaged in changing their states. To comprehend the relationship between mind and brain, we must understand the relationship between what things do and what things are; what something does is simply an aspect of that thing considered over some span of time. When we see a ball roll down a hill, we appreciate that the rolling is neither the ball itself, nor something apart in some other world - but merely an aspect of the ball's extension in space-time; it is a description of the ball, over time, seen from the viewpoint of physical laws. Why is it so much harder to appreciate that thinking is an aspect of the brain, that also could be described, in principle, in terms of the self-same physical laws? The answer is that minds do not seem physical to us because we know so little of the processes inside brains.

An ethicist is someone who sees something wrong with whatever you have in mind.

Author Picture
First Name
Marvin
Last Name
Minsky, fully Marvin Lee Minsky
Birth Date
1927
Bio

American Cognitive Scientist in the field of Artificial Intelligence, Co-Founder of the Massachusetts Institute of Technology's AI Laboratory, Author