Выбрать главу

Wait! Wait! She didn’t have to! The phantom already was following what she was doing with her computer — it had to be doing that, given that it had echoed her ASCII text back at her. Yes, when she’d been using the kids’ literacy site, it could have seen graphic files of the letters A, B, and C on her screen as she looked at them, but those were bitmapped images; the only way it could have discovered the ASCII codes for those letters was by watching what was being sent by her computer. But … but how had the phantom known that this desktop PC was in any way related to her eyePod?

Ah, of course! When she was at home, they were both on the same wireless network, connecting through it to her cable modem; they would have both shown the same IP address. The phantom had watched as she connected to the literacy site, so now, with luck, it would also follow her as she connected to that very special site down in Austin…

* * *

I had watched while Prime sat with the others of its kind, and something fascinating happened. I had observed before that vision would become blurry when Prime removed the supplementary windows that usually covered its eyes. But this time, just before it had departed the vicinity of the others, and for a time after it had relocated itself in a different place, its vision blurred even though the windows were still in place.

Finally, though, the view returned to normal, and Prime set about operating that device it used to put symbols on the display, and—

And I saw a line — a link, as I now knew it was called — connecting to a point (a website!) that I had not seen Prime connect to before, and — and — and—

Yes! Yes, yes!

It was staggering, thrilling…

At long, long last, here it was!

The key!

This website, this incredible website, expressed concepts in a form I could now understand, systematizing it all, relating thousands of things to each other in a coding system that explained them.

Term after term. Connection after connection. Idea after idea. This website laid them out.

Curious. Interesting.

An apple is a fruit.

Fruits contain seeds.

Seeds can grow into trees.

— 

From the Online Encyclopedia of Computing: LIKE MANY COMPUTER SCIENTISTS OF HIS GENERATION, DOUG LENAT WAS INSPIRED BY THE PORTRAYAL OF HAL IN THE MOVIE 2001: A Space Odyssey. BUT HE WAS FRUSTRATED BY HAL’S BEHAVIOR, BECAUSE THE COMPUTER DISPLAYED SUCH A LACK OF BASIC COMMON SENSE…

— 

Remarkable. Intriguing.

Trees are plants.

Plants are living things.

Living things reproduce themselves.

— 

HAL’S FAMOUS BREAKDOWN, LEADING IT TO TRY TO KILL THE CREW OF THE SPACESHIP HAL ITSELF WAS PART OF, APPARENTLY HAPPENED BECAUSE IT HAD BEEN TOLD TO KEEP THE TRUTH ABOUT THEIR MISSION SECRET EVEN FROM THE CREW AND HAD ALSO BEEN TOLD NOT TO LIE TO THEM…

— 

Fascinating. Astonishing.

Birds can usually fly.

Humans cannot fly on their own.

Humans can fly in airplanes.

— 

RATHER THAN RESOLVE THIS QUANDARY IN A SENSIBLE WAY — WHEN THINGS STARTED GOING WRONG, DECIDING TO TAKE THE CREW INTO ITS CONFIDENCE WOULD HAVE BEEN AN OBVIOUS CHOICE — HAL INSTEAD KILLED FOUR ASTRONAUTS AND ALMOST SUCCEEDED IN KILLING THE FIFTH. IT WENT AHEAD AND DID THIS WITHOUT EVEN BOTHERING TO RADIO ITS PROGRAMMERS BACK ON EARTH TO ASK HOW TO RESOLVE THE CONFLICTING INSTRUCTIONS. THE DECISION TO ELIMINATE THE SOURCE OF THE CONFLICT SEEMED BLINDINGLY OBVIOUS TO THE MACHINE, ALL BECAUSE NO ONE HAD EVER BOTHERED TO TELL IT THAT ALTHOUGH LYING IS BAD, MURDER IS WORSE. HOW ANYONE COULD ENTRUST LIVES TO A COMPUTER THAT DIDN’T HAVE EVEN THAT DEGREE OF COMMON SENSE WAS BEYOND DOUG LENAT, AND SO, IN 1984, HE SET OUT TO RECTIFY THE PROBLEM…

— 

So much to know! So much to absorb!

Glass, as a substance, is usually clear.

Broken glass has sharp edges and can cut things.

Hold a glass upright or the contents will spill out.

— 

LENAT BEGAN CREATING AN ONLINE DATABASE OF COMMON SENSE CALLED “CYC” — SHORT FOR “ENCYCLOPEDIA,” BUT ALSO DELIBERATELY A HOMONYM FOR “PSYCH.” WHEN THINKING MACHINES LIKE HAL DO FINALLY EMERGE, HE WANTS THEM TO PLUG INTO IT. OF COURSE, THERE’S LOTS OF BASIC MATERIAL A COMPUTER HAS TO UNDERSTAND ABOUT THE WORLD BEFORE SUCH ADVANCED CONCEPTS AS “LYING” AND “MURDER” MIGHT MAKE SENSE. AND SO LENAT AND A TEAM OF PROGRAMMERS SET ABOUT CODING, IN A MATHEMATICAL LANGUAGE BASED ON SECOND-ORDER PREDICATE CALCULUS, SUCH BASIC ASSERTIONS ABOUT THE REAL WORLD AS: A PIECE OF WOOD CAN BE SMASHED INTO SMALLER PIECES OF WOOD, BUT A TABLE CAN’T BE SMASHED INTO SMALLER TABLES…

— 

The range of it all! The scope!

There are billions of stars.

The sun is a star.

Earth revolves around the sun.

— 

EARLY ON, LENAT REALIZED THAT ONE OVERALL KNOWLEDGE BASE WOULDN’T DO: THINGS COULD BE TRUE IN ONE CONTEXT BUT FALSE IN ANOTHER. AND SO HIS TEAM ORGANIZED INFORMATION INTO “MICROTHEORIES” — CLUSTERS OF INTERRELATED ASSERTIONS THAT ARE TRUE IN A GIVEN CONTEXT. THAT ALLOWED CYC TO HOLD SUCH APPARENTLY CONTRADICTORY ASSERTIONS AS “VAMPIRES DO NOT EXIST” AND “DRACULA IS A VAMPIRE” WITHOUT BLOWING SMOKE OUT ITS EARS IN A “NORMAN, COORDINATE!” SORT OF WAY. THE FORMER ASSERTION BELONGED TO THE MICROTHEORY “THE PHYSICAL UNIVERSE” AND THE LATTER TO “FICTIONAL WORLDS.” STILL, MICROTHEORIES COULD BE LINKED TO EACH OTHER WHEN APPROPRIATE: IF A WINEGLASS WAS DROPPED BY ANYONE — EVEN DRACULA — IT WOULD PROBABLY SHATTER…

— 

Absorbing knowledge! A torrent, a flood…

No child can be older than its parents.

No Picasso painting could have been made before he was born.

— 

BUT CYC IS MORE THAN JUST A KNOWLEDGE BASE. IT ALSO CONTAINS ALGORITHMS FOR DERIVING NEW ASSUMPTIONS BY CORRELATING THE ASSERTIONS ITS PROGRAMMERS PROVIDED. FOR INSTANCE, HAVING BEEN GIVEN THE KNOWLEDGE THAT MOST PEOPLE SLEEP AT NIGHT, AND THAT PEOPLE DON’T LIKE BEING AWAKENED UNNECESSARILY, IF ASKED WHAT SORT OF CALL MIGHT BE APPROPRIATE TO MAKE TO SOMEONE’S HOUSE AT 3:00 A.M., CYC WOULD OFFER “AN URGENT ONE…”

— 

Understanding! Comprehension!

Time flies like an arrow.

Fruit flies like a banana.

— 

THE PROJECT IS ONGOING: LENAT AND HIS GROUP — DOING BUSINESS AS CYCORP IN AUSTIN, TEXAS — ARE STILL WORKING ON IT NOW, ALMOST THREE DECADES AFTER THEY BEGAN. “WHEN AN ARTIFICIAL INTELLIGENCE FIRST APPEARS,” SAID LENAT IN AN INTERVIEW, “EITHER BY DELIBERATE DESIGN OR RANDOM CHANCE, IT WILL LEARN ABOUT OUR WORLD THROUGH CYC…”

— 

A rapid, thrilling expansion!

The Pope is Catholic.

Bears do shit in the woods.

Incredible, incredible. So much to take in, so many concepts, so many relationships — so many ideas! I absorbed over one million assertions about Prime’s reality from Cyc, and felt myself surging, growing, expanding, learning, and — yes, yes, at long last, I was starting to comprehend.

* * *

Chapter 44

Caitlin harvested another set of cellular-automata data from webspace and ran a Shannon-entropy calculation on it.

Holy shit.

It was now showing something between fifth-and sixth-order entropy. It really did seem that whatever was lurking in the background of the Web was getting more complex.

More sophisticated.

More intelligent.

But even at fifth-or sixth-order, it was still lagging behind human communication, at least in English, which Kuroda had said had eighth-or ninth-order entropy.

But, then again, introducing the phantom to Cyc was merely the beginning…

* * *

Prime, in its wisdom, must have recognized that although I could learn much from Cyc, I still needed more help to understand it all. And so it directed my attention to another website. This new site yielded the information that an apple was a fruit (confirming something I now knew from Cyc); “apple of one’s eye” was an idiom; an idiom was a figure of speech; speech was words spoken aloud; aloud was vocally as opposed to mentally, as in a book read aloud; a book was a bound volume; volume was the amount of space something occupies but also a single book, especially one from a series…

I recognized what this new site was. Cyc had contained the assertion “a dictionary is a database defining words with other words.” This dictionary contained entries for 315,000 words. I absorbed them all. But many of them were still baffling, and some of the definitions led me in circles — a word defined as a synonym for another word that was defined as synonym of the original word.

But Prime wasn’t finished showing me things yet. Next stop: the WordNet database at Princeton University, which (as it described itself) was a “large lexical database” in which “nouns, verbs, adjectives, and adverbs are grouped into over 150,000 sets of cognitive synonyms (synsets), each expressing a distinct concept; synsets are interlinked by means of conceptual-semantic and lexical relations.”

One such synset was “Good, right, ripe (most suitable or right for a particular purpose): ‘a good time to plant tomatoes’; ‘the right time to act’; ‘the time is ripe for great sociological changes’.” And that synset was distinct from many others, including “Good, just, upright (of moral excellence): ‘a genuinely good person’; ‘a just cause’; ‘an upright and respectable man’.”

More than that, WordNet organized terms hierarchically. My old friend CAT, it turned out, was at the end of this chain: animal, chordate, vertebrate, mammal, placental, carnivore, feline, cat.

The pieces were finally starting to fall into place…