21.03.2015 Views

Darwin's Dangerous Idea - Evolution and the Meaning of Life

Darwin's Dangerous Idea - Evolution and the Meaning of Life

Darwin's Dangerous Idea - Evolution and the Meaning of Life

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

444 THE EMPEROR'S NEW MIND, AND OTHER FABLES<br />

3. THE PHANTOM QUANTUM-GRAVITY COMPUTER:<br />

LESSONS FROM LAPLAND<br />

I am a strong believer in <strong>the</strong> power <strong>of</strong> natural selection. But I do not see<br />

how natural selection, in itself, can evolve algorithms which could have<br />

<strong>the</strong> kind <strong>of</strong> conscious judgements <strong>of</strong> <strong>the</strong> validity <strong>of</strong> o<strong>the</strong>r algorithms that<br />

we seem to have.<br />

—ROGER PENROSE 1989, p. 414<br />

/ don't think <strong>the</strong> brain came in <strong>the</strong> Darwinian manner. In fact, it is<br />

disprovable. Simple mechanisms can't yield <strong>the</strong> brain. I think <strong>the</strong> basic<br />

elements <strong>of</strong> <strong>the</strong> universe are simple. <strong>Life</strong> force is a primitive element <strong>of</strong><br />

<strong>the</strong> universe <strong>and</strong> it obeys certain laws <strong>of</strong> action. These laws are not<br />

simple <strong>and</strong> not mechanical<br />

—KURT GODEL 2<br />

When Penrose insists that <strong>the</strong> brain is no Turing machine, it is important to<br />

underst<strong>and</strong> what he is not saying. He is not making <strong>the</strong> obvious (<strong>and</strong><br />

obviously irrelevant) claim that <strong>the</strong> brain is not well modeled by Turing's<br />

original thought-device: a smallish gadget sitting astride a paper tape, examining<br />

one square <strong>of</strong> <strong>the</strong> tape at a time. Nobody ever thought o<strong>the</strong>rwise. He<br />

is also not merely saying that <strong>the</strong> brain is not a serial computer, a "von<br />

Neumann machine,'' but, ra<strong>the</strong>r, a massively parallel computer. And he is not<br />

just saying that <strong>the</strong> brain makes use <strong>of</strong> r<strong>and</strong>omness or pseudo-r<strong>and</strong>omness in<br />

running its algorithms. He sees—though some o<strong>the</strong>rs have not—that<br />

algorithms availing <strong>the</strong>mselves <strong>of</strong> large doses <strong>of</strong> r<strong>and</strong>omness are still<br />

algorithms within <strong>the</strong> purview <strong>of</strong> Artificial Intelligence, <strong>and</strong> still fall under<br />

<strong>the</strong> limitations Godel's Theorem places on all Turing machines, <strong>of</strong> whatever<br />

size <strong>and</strong> shape. 3<br />

2. A remark made in 1971, quoted in Wang 1993, p. 133. See also Wang 1974, p. 326:<br />

"Godel believes that mechanism in biology is a prejudice <strong>of</strong> our time which will be<br />

disproved. In this case, one disproval, in Godel's opinion, will consist in a ma<strong>the</strong>matical<br />

<strong>the</strong>orem to <strong>the</strong> effect that <strong>the</strong> formation within geological times <strong>of</strong> a human body by <strong>the</strong><br />

laws <strong>of</strong> physics ( or any o<strong>the</strong>r laws <strong>of</strong> a similar nature ), starting from a r<strong>and</strong>om distribution<br />

<strong>of</strong> <strong>the</strong> elementary particles <strong>and</strong> <strong>the</strong> field, is as unlikely as <strong>the</strong> separation by chance<br />

<strong>of</strong> <strong>the</strong> atmosphere into its components."<br />

3. Someone who doesn't realize this is Gerald Edelman, whose "neural Darwinism"<br />

simulations are both parallel <strong>and</strong> heavily stochastic (involving r<strong>and</strong>omness), a fact he<br />

<strong>of</strong>ten cites, mistakenly, as evidence that his models are not algorithms, <strong>and</strong> that he himself<br />

is not engaged in "strong AI" ( e.g., Edelman 1992 ). He is; his protestations to <strong>the</strong> contrary<br />

betray an elementary misunderst<strong>and</strong>ing <strong>of</strong> computers, but that just goes to show, as<br />

The Phantom Quantum-Gravity Computer 445<br />

Moreover, in <strong>the</strong> wake <strong>of</strong> <strong>the</strong> commentary his book provoked, Penrose<br />

now grants that heuristic programs are algorithms as well, <strong>and</strong> acknowledges<br />

that, if he is to find an argument against AI, he has to concede <strong>the</strong>ir<br />

tremendous power to track <strong>the</strong> truths <strong>of</strong> arithmetic <strong>and</strong> everything else, if not<br />

perfectly, <strong>the</strong>n at least impressively. He <strong>of</strong>fers a fur<strong>the</strong>r point <strong>of</strong> clarification:<br />

any computer that operates by indulging in interactions with an external<br />

environment is an algorithmic computer provided <strong>the</strong> external environment is<br />

itself entirely algorithmic. (If skyhooks grew like toadstools—or, more to <strong>the</strong><br />

point, like oracles perched on toadstools—<strong>and</strong> a computer was helped along<br />

by its occasional communication with <strong>the</strong>se skyhooks, <strong>the</strong>n what it did would<br />

be no algorithm.)<br />

Now, with all this useful clarification in place, what does Penrose maintain?<br />

In May 1993, I spent a week with Penrose <strong>and</strong> some Swedish physicists<br />

<strong>and</strong> o<strong>the</strong>r scientists discussing our different views about <strong>the</strong>se matters, at a<br />

workshop in Abisko, a tundra-research station well north <strong>of</strong> <strong>the</strong> Arctic Circle<br />

in Sweden. Perhaps <strong>the</strong> midnight sun helped as much as our Swedish hosts to<br />

illuminate <strong>the</strong> path, but, in any event, I think we both came away<br />

enlightened. Penrose proposes a revolution in physics, centered on a new—<br />

<strong>and</strong> still unformulated—<strong>the</strong>ory <strong>of</strong> "quantum gravity," which he hopes will<br />

explain how <strong>the</strong> human brain transcends <strong>the</strong> limitations <strong>of</strong> algorithms. Does<br />

Penrose envisage <strong>the</strong> human brain, with its special quantum-physics powers,<br />

to be a skyhook or a crane? That was <strong>the</strong> question I went to Sweden to<br />

answer, <strong>and</strong> <strong>the</strong> answer I came back with is this: He has definitely been<br />

looking for a skyhook. I think he'd settle for a new crane—but I doubt that<br />

he's found one.<br />

Descartes <strong>and</strong> Locke, <strong>and</strong> more recently Edgar Allan Poe, Kurt Godel, <strong>and</strong><br />

J. R. Lucas, thought that <strong>the</strong> alternative to a "mechanical" mind would be an<br />

immaterial mind, or a soul, to speak with tradition. Hubert Dreyfus <strong>and</strong> John<br />

Searle, more recent skeptics about AI, have shunned such dualism <strong>and</strong> opined<br />

that <strong>the</strong> mind is indeed just <strong>the</strong> brain, but <strong>the</strong> brain is not any ordinary<br />

computer; it has "causal powers" (Searle 1985) that go beyond <strong>the</strong> running <strong>of</strong><br />

any algorithms. Nei<strong>the</strong>r Dreyfus nor Searle has been very forthcoming about<br />

what special powers <strong>the</strong>se might be, or which <strong>of</strong> <strong>the</strong> physical sciences might<br />

be <strong>the</strong> right one to give an account <strong>of</strong> <strong>the</strong>m, but o<strong>the</strong>rs have wondered<br />

whe<strong>the</strong>r physics might hold <strong>the</strong> key. To many <strong>of</strong> <strong>the</strong>m, Penrose appears to be<br />

a knight in shining armor.<br />

Quantum physics to <strong>the</strong> rescue! Several different proposals have been<br />

advanced over <strong>the</strong> years about how quantum effects might be harnessed to<br />

give <strong>the</strong> brain special powers beyond those <strong>of</strong> any ordinary computer. J. R.<br />

everybody in AI knows, that although you may not have "Absolute Ignorance" (as Mac-<br />

Kenzie anonymously put it, back in chapter 3, p 65), you still don't have to underst<strong>and</strong><br />

what you are making in order to make it.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!