Part VI. Common Sense

PART VI. COMMON SENSE........................... 1

§6-1. What do we mean by Common Sense?........ 2

The Telephone Call............................................. 3

The concept of a ‘Panalogy.’.............................. 4

Sub-Realms of the Telephone World.................. 6

§6-2. Commonsense Knowledge and Reasoning. 8

How much does a typical person know?.......... 10

Could we build a Baby-Machine?.................... 11

Remembering................................................... 14

§6-3. Intentions and Goals.................................. 16

Difference-Engines........................................... 17

Goals and Subgoals.......................................... 20

§6-4. A World of Differences............................. 22

Rhythmic and Musical Differences.................. 25

Difference-Networks........................................ 26

§6-5. Making Decisions 27

§6-6. Reasoning by Analogy............................... 29

A Geometric Analogy Program............................ 29

§6-7. Knowledge needs Multiple Representations 32

"The way to make money is to buy stock at a low price, then when the price goes up, sell it. If the price doesn't go up, don't buy it." —Will Rogers.

Soon after the first computers appeared, their actions became the subjects of jokes. The tiniest errors in programming them could wipe out their clients’ bank accounts, credit them with outlandish amounts, or trap the computers in circular loops that kept repeating the same mistakes. This maddening lack of common sense led most observers to suspect that machines could never have genuine minds.

Today many programs do outstanding jobs more efficiently and reliably. Some of them can beat people at chess. Others can diagnose heart attacks. Yet others can recognize pictures of faces, assemble cars in factories, or even pilot planes or ships. But no machine yet can read a book, clean a house, or baby-sit.

Then why cannot our computers yet do so many things that people can do? Do they need more memory, speed, or complexity? Do they use the wrong kinds of instruction-sets? Do their limitations come from the fact that they only use zeros and ones? Or do machines lack some magical attribute that only a human brain can possess? This chapter will try to show, instead, that we don’t need to look for excuses like these, because most deficiencies of today’s machines stem from the limited ways we’ve been programming them.

One of these limitations is that we usually give a present-day program only the knowledge we think it will need to solve each particular problem. In contrast, every normal child learns millions of fragments of knowledge and skills that people regard as ‘obvious.’ For example, if you heard that someone tied a package with ‘string’ you might connect that word with ideas like these:

You can use a string to pull, but not push.

But you cannot push a thing with a string. Loose strings tend to get tangled up. Fill your package before you tie it up. A string will break when pulled too tight.

The first parts of this chapter will discuss the need for very large bodies of commonsense knowledge, as well as the kinds of skills we need for retrieving and applying such knowledge.

The middle parts of this chapter explore another cause for the weakness of present-day programs: they specify what the computer should do—without telling it which goals to achieve, or the intentions of those who programmed it. This means that they have no ways to reflect on whether those goals were achieved at all—or, if they were, at what cost and how well. Furthermore, those computers will still lack resourcefulness, even with access to great stores of knowledge because few fragments of knowledge are of use by themselves, unless they are also connected to reasons or goals for using them.

If you break something, you should replace it. (Because its owner wants it intact.)

People usually go indoors when it rains. (Because they do not like to get wet.)

It is hard to stay awake when you’re bored. (Why would one want to stay awake?)

People don’t like to be interrupted. (Because they want you to hear what they say.)

It is hard to hear in a noisy place. (You might want to hear what others say.)No one else can tell what you’re thinking. (Why might you value that privacy?)

Another deficiency is that a typical program will simply give up when it lacks some knowledge it needs—whereas a person can find other ways to proceed. So the final parts of this chapter discuss some of the tactics that people can use when we don’t already know just what to do— for example, by making useful analogies.