About the SA Blog Network



Opinion, arguments & analyses from the editors of Scientific American
Observations HomeAboutContact

Will Computers Ever Know Everything?

Email   PrintPrint

The Alan Turing Statue on display at Bletchley ParkWhat was Alan Turing’s greatest contribution? Here was a man who invented the idea of the modern computer, a man upon whose insights the information technology revolution firmly rests. He was the first to understand that instructions are themselves data, making algorithms capable of the recursive thinking that makes humans unique. (I think that I think, therefore I am.) He realized that machines would become so smart that they would eventually be mistaken for humans—this in an era when the chairman of IBM was claiming that there would be a worldwide market for perhaps five computers—and devised a test to tell man from automaton. Oh, and he also led the successful Allied effort to break Germany’s secret Enigma codes during World War II, as he realized that “no one else was doing anything about it and I could have it to myself.”

An essay in this week’s issue of Science magazine by Andrew Hodges, the dean of Wadham College at Oxford University and the author of Alan Turing: The Enigma argues that, great as these achievements are, Turing’s greatest contribution was defining the limits of what computers can “know”—that is, what is computable. By formalizing the computability question in 1936, Turing illuminated the deeper issue of what humans could know: Is our knowledge limited in the same way as computers? Or do we have some sort of mental “intuition” (Turing’s word) that supersedes the power of mere machinery?

Turing wasn’t sure (though he suspected that the strange rules of quantum mechanics may give our brains some non-deterministic wiggle room). Three quarters of a century later, we’re not much closer to an answer. Even in this age of “big data,” where computers churn through gobs of information to come up with cannily human-like responses (consider IBM’s Jeopardy-beating Watson computer, named, incidentally, after the CEO with poor computer-demand forecasting skills), humans are far better at everyday tasks like making sense of a scene. Artificial intelligence remains a dream.

Turing’s work on computability led to an even deeper question, according to Hodges: “Does computation with discrete symbols give a complete account of the physical world?” In other words, is the world computable? Can a machine, in principle, rise not just to the intellectual capabilities of human beings, but supersede those capabilities? Can a computer know everything?

This past December we asked the Harvard researcher David Weinberger to profile the Living Earth Simulator, an project that would take in lots of data and compute the future of human civilization. It is perhaps the most ambitious effort ever launched to test the computability of the world. (Weinberger came away unconvinced.) Perhaps the Living Earth Simulator will succeed in due time, and computers will be able to predict the future.

More likely it will not, and another 75 years hence, we will still be struggling with the rich philosophical loam that Turing left us with. You don’t need a computer to predict that we’re going to require a mind as exceptional as Turing’s to make sense of it.

Image courtesy Chris Brown (zoonabar) on Flickr

About the Author: Michael Moyer is the editor in charge of space and physics coverage at Scientific American. Follow on Twitter @mmoyr.

The views expressed are those of the author and are not necessarily those of Scientific American.

Rights & Permissions

Comments 16 Comments

Add Comment
  1. 1. agwisreal 7:37 pm 04/12/2012

    There are results in mathematics about billiards and how fast the pattern of which ball hits which fans out depending on the initial uncertainty in the break.

    Any tiny uncertainty, e.g. 1 part in 10^100, balloons into near total randomness after a few dozen collisions.

    Good luck with predicting the exact details of the future. The best that could be done would be to come up with a bunch of reasonably possible futures.

    Link to this
  2. 2. Dr. Strangelove 12:42 am 04/13/2012

    Living Earth Simulator to compute the future of human civilization? LOL That’s like the gigantic computer in the Hitchhiker’s Guide to the Galaxy built by the hyperintelligent pandimensional beings to compute the meaning of the universe, life and everything. After millions of years of computing, it gave the answer 42. They had to build an even bigger computer to answer what the question was. That computer was Earth, which was mistaken for a planet by its dumb ape-like inhabitants.

    Will computers ever know everything? One page of bond paper contains about 2,000 letters. The number of possible permutations of the 26 letters of the English alphabet in one page is 26^2,000. That’s more than the total number of subatomic particles in the observable universe.

    Therefore, a computer as big as the universe cannot contain all the possible letter permutations in one bond paper. A computer can be intelligent but it cannot know everything.

    Link to this
  3. 3. BuckSkinMan 2:43 am 04/13/2012

    Computers that can “know everything” and predict the future? Not even close to being possible at this point.

    Reason #1: Mankind does not “know everything” – articles in  SciAm TYPICALLY pose still unanswered questions relating to the Universe and even our own “simple, insignificant” organism. So how could OUR organism build such an omniscient computer?
    Reason#2: examine any current news which sparks debate & controversy and especially observe the “comments” (aka, outcries & demands) about it: you will quickly see that IRRATIONALITY AND IGNORANCE are the hallmarks of modern mankind.
    Reason#3: Note that a substantial part of humanity BELIEVES in super natural mystical powers “beyond knowing” and INSISTS that Science predicting climate change and supporting evolution of the species is a mere hoax to promote “godless liberal ideology.” Hah!

    So, someone totally wise, totally knowledgeable and totally in control of themselves would “somehow” load “infinite information” into a computer data base and install a perfect all-knowing computer system and come up with the ACTUAL answer to: “How did our universe begin?” Not darned likely.

    Something’s fishy about this premise, even as a question, it ignores a feature of reality. That feature is: that we know we don’t know enough, despite centuries of research and investigation. At this point all we know is that whatever additional knowledge we gain only brings up more questions. Some answers we DO have are useless because we have no way (yet) to put them to use. (Like: our sun produces vast quantities of energy but we still are running a big energy deficit.)

    Link to this
  4. 4. TTLG 1:06 pm 04/13/2012

    Nicely put, Dr. Strangelove. But even if a machine cannot know everything, is it not still possible that it could know far more than any human? For instance, could it not be far better at arranging those 2000 characters into meaningful and worthwhile combinations than could a person?

    Link to this
  5. 5. driwatson 6:49 pm 04/13/2012

    Some computer scientists believe that a breakthrough could be about to happen. They think that thanks to the Internet, the cloud, and massive distributed processing power it will be possible for software to catalog, analyze, correlate, and cross-link everything in the digital realm. “These data and the capacity to analyze them appropriately could allow a machine to answer heretofore computer-unanswerable questions and even pass a Turing Test.” You can read more here

    Link to this
  6. 6. Chris Gerrard 6:54 pm 04/13/2012

    Isn’t the Living Earth Simulator replicating work already accomplished by Hari Seldon?

    Link to this
  7. 7. Ilpalazzo 10:18 pm 04/13/2012

    Computers still need orders. They won’t ever be free-thinking.

    Link to this
  8. 8. Ilpalazzo 10:20 pm 04/13/2012

    And further, they don’t ‘know’ anything – they have access to data stored. Humans on the other hand, don’t have easy to access data slots in their minds, so they actually have to remember and form knowledge. They’d ‘know’ as much as someone who googles information, only much faster.

    Link to this
  9. 9. seanacoy 12:01 am 04/14/2012

    Can a computer be programmed to definitively define “know,” “think,” and “intuit”, and can a computer come to such definitions on its own? Who is competent to evaluate the answers?

    The concept of “non-deterministic wiggle room” is an important insight, and “irrationality” and “ignorance” feed into it. Today (and in most science fiction) a computer basically comes up with an answer based on information fed to it (and therefore given an initial truth value) or reports back that it has “insufficient information” or computing space. The human mind (in part) has developed to deal with “insuffiicent information” (ignorance), unreliability of information, insufficient time or means to analyze, by making guesses, rough assumptions, pattern recognitions within pattern recognitions, cognitive and logical leaps of inconsistent breadth and utility, emotional and other bias, and intuition and inspiration. For example, when a human makes a mistake, the human may learn not to make the mistake again, or may treat it as a possibility for “new” ideas and approaches. Thus the artist meaning to put a line there puts it here – perhaps because of a shaky hand or distraction, an earthquake or fly bussing around, considers the errant line, and builds something slightly different. In this respect, the human mind acts like genetic evolution, where seemingly non-deterministic errors within the mind or in the minds perceptions of reality or in accurately perceived reality can be viewed as analogous to mutations within genetic material. Most such instances make no significant differences in outcome, at most helping to keep actions on course or avoid a collision. Few are disastrous; and a few are Aha! moments.

    At this point I don’t see how a computer could handle the comparable enormous number of external and internal seemingly random and unreliable or extraneous inputs and, like humans, make something of them.

    Link to this
  10. 10. SriArthur 3:55 am 04/14/2012

    I am not comfortable with the concept of algorithms being data. If I create several proofs that 1 plus 1 equals 2 and the difference lies within the method of determination should each be considered a datum? What does it take to distinguish between an algorithm and a heuristic? Are they really equals?

    Turing asked interesting questions, but his answers were already determined by his questions. Cleverness lies within being able to look at a question, consider alternatives, and then derive a solution that is both unexpected and eminently reasonable.

    Computers do not think. I doubt we can ever make that happen. At least for now, we need to be creative in making software more useful and there is this problem. Who decides what is useful and how much are we willing to pay to have it developed? The answer seems to fall prey to power, politics, and persuasion.

    Candidly, I suspect we can htis question of other humans, “Can you prove to me that you can think?” and not be satisfied with the answer.

    Link to this
  11. 11. Dr. Strangelove 10:10 pm 04/15/2012

    “But even if a machine cannot know everything, is it not still possible that it could know far more than any human?”

    Sure. Has Watson computer ever been defeated by any human in Jeopardy?

    “Cleverness lies within being able to look at a question, consider alternatives, and then derive a solution that is both unexpected and eminently reasonable.”

    Didn’t Deep Blue computer do all that when it beat Kasparov in a chess tournament? It answered the question “what’s the best move?” Considered the different alternatives; then derived a solution that is unexpected (to Kasparov) and reasonable (to chess analysts)

    Link to this
  12. 12. Dr. Pete 3:25 pm 04/18/2012

    “This past December we asked the Harvard researcher David Weinberger to profile the Living Earth Simulator, an (sic) project that would take in lots of data and compute the future of human civilization.”

    A computer would never have made that error.

    Link to this
  13. 13. jackedup 5:54 am 04/19/2012

    i think no… computer only knows that things which we told it to remember… MAN created computers, not computers created man..

    Link to this
  14. 14. Dr. Strangelove 9:42 pm 04/19/2012

    If computers can only remember but not think, how come the computer engineers who built Deep Blue can’t beat the computer in chess? The Creation can easily beat the Creators.

    Shannon, one of the founders of computer science, was asked if computers will be able to think in the future. He said sure I can think, don’t I?

    Link to this
  15. 15. Quinn the Eskimo 9:57 pm 04/22/2012

    I was duly impressed as I peered into the cavernous space that beheld the IBM 360-65 System at DLSC in Battle Creek. It was massive!

    Yet for all of its computational capabilities and speed, I’d not trade my current iMac for one 360-65 today. In fact, I wouldn’t want to turn it on.

    Link to this
  16. 16. elisabethshue 4:51 pm 09/26/2012

    Turing asked interesting questions, but his answers were already determined by his questions. Cleverness lies within being able to look at a question, consider alternatives, and then derive a solution that is both unexpected and eminently reasonable.

    Link to this

Add a Comment
You must sign in or register as a member to submit a comment.

More from Scientific American

Scientific American MIND iPad

Give a Gift & Get a Gift - Free!

Give a 1 year subscription as low as $14.99

Subscribe Now >>


Email this Article