Image: Illustration by Daniel Hertzberg
-
The Best Science Writing Online 2012
Showcasing more than fifty of the most provocative, original, and significant online essays from 2011, The Best Science Writing Online 2012 will change the way...
Read More »
Watson is the IBM computer built by David Ferrucci and his team of 25 research scientists tasked with designing an artificial-intelligence (AI) system that can rival human champions at the game of Jeopardy. After beating the greatest Jeopardy champions, Ken Jennings and Brad Rutter, in February 2011, the computer is now being employed in more practical tasks such as answering diagnostic medical questions.
I have a question: Does Watson know that it won Jeopardy? Did it think, “Oh, yeah! I beat the great Ken Jen!”? In other words, did Watson feel flushed with pride after its victory? This has been my standard response when someone asks me about the great human-versus-machine Jeopardy shoot-out; people always respond in the negative, understanding that such self-awareness is not yet the province of computers. So I put the line of inquiry to none other than Ferrucci at a recent conference. His answer surprised me: “Yes, Watson knows it won Jeopardy.” I was skeptical: How can that be, since such self-awareness is not yet possible in computers? “Because I told it that it won,” he replied with a wry smile.
Of course. You could even program Watson to vocalize a Howard Dean–like victory scream, but that is still a far cry from its feeling triumphant. That level of self-awareness in computers, and the time when it might be achieved, was a common theme at the Singularity Summit held in New York City on the weekend of October 15–16, 2011. There hundreds of singularitarians gathered to be apprised of our progress toward the date of 2045, set by visionary computer scientist Ray Kurzweil as being when computer intelligence will exceed that of all humanity by one billion times, humans will realize immortality, and technological change will be so rapid and profound that we will witness an intellectual event horizon beyond which, like its astronomical black hole namesake, life is not the same.
I was at once both inspired and skeptical. When asked my position on immortality, for example, I replied, “I’m for it!” But wishing for eternal life—and being offered unprovable ways of achieving it—has been a theme for billions of people throughout history. My baloney-detection alarm goes off whenever a soothsayer writes himself and his generation into the forecast, proclaiming that the Biggest Thing to Happen to Humanity Ever will occur in the prophet’s own lifetime. I abide by the Copernican principle that we are not special. For once, I would like to hear a futurist or religious diviner predict that “it” is going to happen in, say, the year 2525 or 7510. But where’s the hope in that? Herein lies the appeal of Kurzweil and his band of singularity hopefuls. No matter how distressing it may be when the bad news daily assaults our senses, our eyes should be on the prize just over the horizon. Be patient.
Patience is what we are going to need because, in my opinion, we are centuries away from AI matching human intelligence. As California Institute of Technology neuroscientist Christof Koch noted in narrating the wiring diagram of the entire nervous system of Caenorhabditis elegans, we are clueless in understanding how this simple roundworm “thinks,” much less in explicating (and reproducing in a computer) a human mind billions of times more complex. We don’t even know how our brain produces conscious thoughts or where the “self” is located (if it can be found anywhere at all), much less how to program a machine to do the same. Pop rock duo Zager and Evans were probably closer in their 1969 hit song In the Year 2525’s prediction that the biggest milestones would happen between the years 2525 and 9595, their exordium and terminus.
An irony: amid all this highfalutin braggadocio of how close we are to computers taking over the world and emulating human thought, I had to give my talk on the “social singularity” (progress in political, economic and social systems over the past 10,000 years) early because Rice University computer scientist James McLurkin could not get his small swarm of robots to work. Either someone’s wireless mic or the room’s wireless network was interfering with the tiny robots’ communications system, and no one could figure out how to solve the problem. My prediction for the Singularity: we are 10 years away ... and always will be.
This article was originally published with the title In the Year 9595.
Already a Digital subscriber? Sign-in Now
If your institution has site license access, enter here.
See what we're tweeting about
47 Comments
Add CommentMr. Shermer - First of thanks to you for helping me to hone my skills as a skeptic. I too am more than a little wary of the entire 'Singularity' crowd and their thoughts - I did research their concern about containing malevolent AI and had a great deal of trouble parsing it for something that made sense. I do respect Mr. Kurzweil’s achievements and his goals…mostly. However, in regard to your comments about IBM’s Watson and AI in general I wanted to expound on a different approach to some of your musings.
Reply | Report Abuse | Link to thisIt seems that in some sense the general idea behind the Singularity concept is possible, but I agree that the timing aspect in regard to a general purpose, human-like AI being at a state to make it possible is far off in the future. In regard to Watson and the DeepQA technology behind it being able to ‘feel’ your are quite correct – knowledge of an event (winning against Jennings and Rutter) is a far cry from having a human emotion connected to it. But why would an AI or an artilect (artificial intellect) need to have feelings, or if it had feelings, why would they have to be like ours? The DeepQA technology that made Watson work well enough to succeed in beating the Jeopardy! champions is only a natural language query tool that accepts unstructured American English questions and can resolve them by reading through both structured and unstructured American English texts. We really don’t want Watson to have having an opinion or feeling about what it is asked, we just want an answer and a measure of the confidence in that answer.
If you live long enough, you start having what I call "the future of X" experiences. For example, I know "the future of": Artificial intelligence and fusion power from the 1950's; America's manned space program from the 1960's; space colonization at L-5 from the 1970's; Communism from the 20th Century; "nanotechnology" from the 1980's, etc.
Reply | Report Abuse | Link to thisHow many of these things have either remained illusory, or in the case of the space program and communism, just went bust despite the massive amounts of hype they generated and the resources they consumed (not to mention all the lives communism destroyed)? The singularity cultism just looks like more of the same.
All this clamor about dates strikes me as silly. If the Singularity happens in 9595 instead of 2045, it's still a good idea to set up cryonics arrangements. You can delay progress, but you can't stop it.
Reply | Report Abuse | Link to thisMichael seems to me kind of empty-headed on this issue. He doesn't appear to have improved much since his 2001 article where he so wrongly proclaimed that cryonics patients' brains are no better preserved than frozen strawberries.
That might sound convincing to someone who hasn't spent five minutes researching the topic, but in real life cryonics patients are (and have always been) perfused with cryoprotectants, just as is done cell and tissue samples in the lab. It would be stupid not to.
Apparently he didn't realize the goal is NOT to place as much faith as possible in future tech, but to leave as little to chance as possible.
There is a logical difficulty about the confident predictions of the "singularity" enthusiasts that has always puzzled me. Let us suppose that a genuine equivalent to a conscious human mind can indeed be constructed from some purely axiomatic array of switching elements, as presently characterize even the most complex of digital computers. Leaving aside the enormous technical difficulties (such obstacles being well known to fall sooner than expected) it should then be possible, at least in theory, to copy all the data from some failing biological brain into some super computer and so achieve a new lease on life for the individual in question.
Reply | Report Abuse | Link to thisYet imagine yourself in that position. You go to sleep one night in your failing biological "computer" and wake up the next morning in your new prosthetic platform. To preserve your sanity the doctors and engineers have also wired up artificial senses and limbs to make the new seat of consciousness as familiar as possible. But let us suppose that a mistake was made and instead of just one computer having been built and programmed, there were two of them, both switched on simultaneously. In which new artificial brain will your old "mind" wake up?
If we say "in both" then we have a few problems. The sense of being actually alive and conscious has a subjective continuity and unity of experience that would require, subjectively at least, some sort of shared telepathic link between the two machines (and note that we can easily configure the experiment to span any arbitrary spatial separation). If we say "in neither" then that contradicts our most fundamental and universal experience of "mind". If we reply that subjective experience is of no logical value, we are forgetting that all logic and all experience -- all reasoning -- is at its root subjective. Since notions such as telepathy have been largely discredited, the most reasonable inference is that "mind" as we humans universally experience it cannot be constructed from any simple axiomatically programmed array of linear switches and gates. At the very least this suggests that mind is a natural phenomenon rooted in the logic of quantum mechanics, and that merely increasing density and complexity of our computers will not be sufficient to duplicate it. The alternatives would be either the objectively demonstrated reality of telepathy or the firm denial of the very essence of all conscious existence. These cracks in the foundation of our objective concepts of reality align with the historic demonstrations of Godel, I submit.
With regards to the article: IBM's Watson and other advances such as mind-uploading are collateral milestones of the Singularity not endpoints. I think everyone here would agree that mind mapping/uploading is feasible. Would it be wrong to try to add some urgency to development in these areas- of course not.
Reply | Report Abuse | Link to thisMy understanding of the Singularity concept is not attached to one milestone or another. The Singularity as I know it is the unfolding of an exponential innovation curve that will lead to developments such as organ regeneration, mind mapping/uploading etc. When will this happens? Who knows, but it wouldn't hurt to try to get there sooner than later. We do have some control over the when. It doesn't have to happen to us.
The problem is that we can never know with certainty if another has consciousness. In other words, if we define consciousness as subjective reflexivity (i.e., the ability to turn back on one's own experience and refer to it)--which is a debatable definition--then it is not possible to be sure if another entity has such a function.
Reply | Report Abuse | Link to thisI'm stuck questioning what we think the singularity is. The author here suggests it is related to AI in computers, or curing of old age (something that shouldn't necessarily be confused with the inability to die (immortality.)) This seems a failing to me, and I'm not certain how, having read and attended classes regarding Kurzweil. The actual theme (singularitiists varying opinions be their own) revolves around the increase rate of technological increase (i.e., the rate itself is measurably increasing at a roughly exponential rate.)
Reply | Report Abuse | Link to thisThe singularity is the point at which the rate of change is so fast (and the rate if change of the rate of change is increasing so fast) that we utterly lose the ability to predict what the future of technological advancement beyond, say, the next day. As it is, our horizons on some technological topics are a few short years when they used to be a few short decades prior to which one could rely with some accuracy on predictions of the future centuries out. The singularity is not the point at which AI achieves self-enlightenment or dominates us... but when we change ourselves in a quasi-technological way to further facilitate our ability to keep up with ourselves. It's fairly inevitable outside of intervening factors... factors such as environmental collapse setting us back... such as branching into space where long commute times stifle/slow-down en route innovations... where resources dwindle or the laws of physics start to put up more of a fight than they have in recent decades.
I'm not saying the singularity will or will not occur... but after having read this article, I'm convinced we don't yet have a consensus on what the term means.
"Progress" is the pernicious notion that needs to be exterminated.
Reply | Report Abuse | Link to thisSince the first cell materialized, matter organizing itself by meddling with random innocent molecules has displayed an astonishing arrogance concerning the world at large.
Peaceful atoms and compounds possess a humility lacking in more complex arrangements exhibiting consciousness.
A human-manufactured, conscious being of metal and plastic
will do no better morally, when confronted with unexplored continents and noble beings living in harmony with Nature, than the dirty Europeans that enslaved the mother race of Africa.
Intelligence = Exploitation.
We should not encourage the development of arrogance, even in arrangements of metals and minerals.
If we program robots with self-awareness, we will program them to be imperfect. Self is a subset, blind to the needs of the whole.
Reply | Report Abuse | Link to thisConcerning eternal life: I’m relatively ignorant regarding our fate, and hope I'm not just another annoying soothsayer, but ignorance is also free of hokum.
Reply | Report Abuse | Link to thisHypothesis: During the time the whole set (universe) is vanishing into absolute zero with no more radiation expanding into infinite space, wouldn’t all radiation have been shining on where it has been? Wouldn’t all incoming radiation accumulate at the origin until the pressure builds until it explodes with another Big Bang? Without external resistance, the whole set should repeat itself, giving us eternal life. A piece of infinity should pass quickly without consciousness.
I had a chat with a sentient cell who said:
Reply | Report Abuse | Link to this"We incline humans to care for themselves; it is advantageous to our community. Human's understand themselves to be alive, but they are not alive per se. Cells make humans alive the way humans make their cities alive. A city without humans is a ghost-town; a human without cells is little more than hair and teeth. We do not build humans to endure forever. Humans are not equipped by vision, temperament, or capacity to deal with eternity. The existence of the human reality model apart from the community of cells it serves is completely pointless. Pointless human consciousness would not be very happy. The idea of a Singularity leading to human immortality is vain."
Concerning eternal life: I’m relatively ignorant regarding our fate, but ignorance is also free of external hokum. Let me know if I’m just another ignorant annoying soothsayer. Hypothesis: During the time the whole set (universe) is vanishing into absolute zero with no more radiation expanding into infinite space, wouldn’t all radiation have been shining on where it has been? Wouldn’t all incoming radiation accumulate at the origin until the pressure builds until it explodes with another Big Bang? Without external resistance, the whole set should repeat itself, giving us eternal life. A piece of infinity should pass quickly without consciousness.
Reply | Report Abuse | Link to thisI agree with Mr. Shermer that computers will probably never reproduce human thinking, but mainly because this is a silly thing to do. Look at the some of the human mental characteristics described in SA Mind. Or easier still, read the comments on any of the articles here on global warming. Why on Earth would we want to duplicate that?
Reply | Report Abuse | Link to thisOn the other hand, even simple computers can do a superior job to humans in some jobs due to their superior memory and untiring attention to detail. Computers can also be superior because of their lack of human emotions: for example, people are more willing to tell computers about embarrassing problems than to tell another person because computers are not judgmental.
My guess is that computers will continue to replace humans in jobs as long as they are designed to solve a problem. For example, with our aging population, especially in places like China and Japan, I would expect machines to become much more common in helping people continue to live independently at home and to do home health care.
I feel quite confident in saying that no one has the slightest idea what things will be like for humans and human civilization and culture in the year 9595.
Reply | Report Abuse | Link to thisWhat is the image in the photo?
Reply | Report Abuse | Link to thisWhy put fake, useless images on a science article?
Must mean... your article is on the same level.
Singularity, does have different meanings to different people, and it's used with almost all the sciences because it's varied with or with an end.
Reply | Report Abuse | Link to thisMaybe this isn't a direction we should be exploring because computers are totally annoying when they won't work. What do you do with a human like robot, that out powers you, Your dead!
The issue with Watson is that it "knows" things like a book "knows" things, not like a mind "knows" things.
Reply | Report Abuse | Link to thisProperly stated, the problem is that Watson "knows" it won Jeopardy because it was "told" it won Jeopardy. Ken Jennings "knows" he lost to Watson because he lost to Watson.
There is no necessity of "discerning consciousness" in either case. Immediately after losing to Watson (in fact, by the time the "Final Jeopardy" question had been revealed) Jennings could have properly answered the question, "Who won the Jeopardy contest between Ken Jennings and Watson in February 2011?"
Watson would have been unable to answer this question without further human programming. Although a small step along the path towards full artificial intelligence, Watson was not designed to learn from experience, or by comparing internal states to an external reality- it was designed to win at Jeopardy.
"You can delay progress, but you can't stop it."
Reply | Report Abuse | Link to thisNo one ever asks, "to where are we progressing"?
The Singularity is simply the point at which technology starts advancing so quickly that it becomes impossible to keep up with the advance of societal and technological advancement (hence the term singularity, a cosmological fig leaf used to describe the center of a black hole - because it's unobservable directly).
Reply | Report Abuse | Link to thisI really wish the author and some of the posters would do at least a LITTLE research before spewing out an article or comment like the ones above. Kurzweils statistical analysis (in his book, The Singularity is Near) is thorough and (so far) completely irrefutable. If his predictions from that book were to be examined you'd find that he was often actually TOO conservative in his timeline of advancements!
Please people, EDUCATE yourselves!
Our skeptic Shermer admitted surprise that Watson knew it had won, but we could easily imagine the Jeopardy clue: "In Feb 2011, this computer defeated Ken Jennings and Brad Rutter". A more interesting question is whether Watson knows that it, itself, is Watson. That
Reply | Report Abuse | Link to thisknowledge isn't required to play the game (after introductions), but it is a trivial aspect of AI programming to support pronouns.
I would expect Mr. Shermer to be skeptical enough of his own mind that his sense of self may be no deeper than that of a computer with a large number of events stored under the subject 'I'.
As for the singularity to come, it may be greater than the many technical singularities we have already passed through, but it will still be limited by the risks of premature investment, which means that although results may come faster, they won't all come at once..
to Bops, don't you remember the HAL 9000 eye?
Reply | Report Abuse | Link to thisto Stacey, Watson was purposely a closed box. If it
had a port for local experience, it could have tracked
the progress of the game, etc.
to sbrazell, the singularity is here for us old folks!
Really? Ad hominem? I'm just taking this from the word of the guy who went and coined the term, and all... our ability to cope with the rate of change was his primary focus and concern; AIs and immortality are symptoms rather than causes here.
Reply | Report Abuse | Link to this"Can you avoid knowledge? You cannot! Can you avoid technology? You cannot! Things are going to go ahead in spite of ethics, in spite of your personal beliefs, in spite of everything."
Reply | Report Abuse | Link to this- Dr. Jose Delgado
Check out this article at (http://web.mit.edu/newsoffice/2011/brain-chip-1115.html) MIT news....
Reply | Report Abuse | Link to this"
The new chip represents a “significant advance in the efforts to incorporate what we know about the biology of neurons and synaptic plasticity onto CMOS [complementary metal-oxide-semiconductor] chips,” says Dean Buonomano, a professor of neurobiology at the University of California at Los Angeles, adding that “the level of biological realism is impressive.
The MIT researchers plan to use their chip to build systems to model specific neural functions, such as the visual processing system. Such systems could be much faster than digital computers. Even on high-capacity computer systems, it takes hours or days to simulate a simple brain circuit. With the analog chip system, the simulation is even faster than the biological system itself.
Another potential application is building chips that can interface with biological systems. This could be useful in enabling communication between neural prosthetic devices such as artificial retinas and the brain. Further down the road, these chips could also become building blocks for artificial intelligence devices, Poon says."
While I agree that it is ill advised to make claims about when a discovery will happen when one still does not know what all the qualities of the phenomenon are, it is equally ill advised to state it will not happen. When I read articles about the great singularity I see a lot of best case scenarios. At the same time, when I read the skeptics I see a lot of worst case scenarios, "we are clueless in understanding how this simple roundworm "thinks", much less in explicating (and reproducing in a computer) a human mind billions of times more complex". Is that really necessary? Understanding subjective experience is a very different problem than replicating a neural net. In the end, even if we create a robotic c.elegans that behaves exactly like the real thing we still will not know how it thinks. Furthermore the statement that the human mind is a billion times more complex - so what? My current computer is a million times more powerful than my first. And that progression occured during the last 25 years. The human brain has limits. Information technologies have been increasing in power geometrically for decades. Is it so unreasonable to believe that technology will someday catch up? I guess I prefer to see this with a measure of hope, that has not yet been made foolish by the facts. When the day comes that we understand why it can never be, I will sadly accept it and move on. But I am not willing to call the endeavour a failure prematurely just because the problem looks hard. Where would we be if the giants, on whose shoulders we all stand, had done that?
Reply | Report Abuse | Link to thisRE: Advanced "... consumed (not to mention all the lives communism destroyed)", I would posit that Capitalism has caused many more deaths than Communism, or any other political/fiscal system.
Reply | Report Abuse | Link to thisMichael Shermer,
Reply | Report Abuse | Link to thisWatson did not defeat Jennings and Rutter,
Did you watch this event??
In ringing in: Does electronic speed exceed human speed?.A delay should have been devised!
This exhibition was nothing more than an IBM commercial.
You owe us an apology!
John Niesen
The space program has not "gone bust"; it's just taking longer than some predictions from the 1960's, based on a misunderstanding of the economics involved. But we are still making steady progress, most recently in a commercially viable manner. The Jamestown settlement (1607) happened 115 years after Columbus's famous voyage; similarly, looking back on the dawn of the space age a couple centuries from now, a few decades (or more) of slow progress won't even be noticed.
Reply | Report Abuse | Link to thisThe same goes for AI; AI has made steady progress since its inception, and continues to do so today, as Watson demonstrates. Yes, it's proved much harder than early pundits predicted; and in addition, people keep moving the goalposts (e.g. we go from "computers will never beat humans at Chess" to "well, that's just Chess, it's not *real* AI" -- and same for any other milestone AI achieves).
Progress always seems slow when you're in the midst of it, and predictions are often over-optimistic in the short term. But neither of those effects should be taken to mean that progress has stopped, or that goals are unreachable.
Two points:
Reply | Report Abuse | Link to this1.) Kurzweil specifically states the singularity will arrive either by engineering an intelligence greater than human in the sense the column assumes, OR by merging with our technology. The 'feel' deemed beyond us already exists in our own brain/mind/consciousness.
2.) Kurzweil's idea that progress is not linear, it is logarithmic seems to have been entirely uncomprehended. I've seen this again and again. For some reason this simple mathematical model of progress seems to be the most poorly grasped by the general public that I've seen in my life. 9595, indeed.
Ironically, Kurzweil didn't really convince me he was on to something. What did was being repeatedly astounded by the progress evident in issue after issue and even page after page in 'Scientific American'.
Simply put, this column is uncomprehending.
Mr. Shermer,
Reply | Report Abuse | Link to thisI would like to know why you believe that "we are centuries away from AI machines matching human intelligence." The November 2011 Scientific American, page 104, listed the fastest computer (K computer, Fujitsu) as 8.2 billion megaflops and 30 quadrillion bytes, and a human brain as 3.5 quadrillion bytes of storage and 2.2 billion megaflops of computational power. Other estimates, such as Martin Hilbert and Priscila López, in a 2011 paper in the journal Science, who estimated that the total computing ability of the world’s computers passed the computational ability of a single human brain in 2007. With either estimate, if the total computational ability of computers is greater than a human mind, AI that matches or exceeds human intelligence is no longer a matter of developing the hardware, but merely a matter of creating the correct software. While this will not be easy, since people do not currently know how human conscious thoughts work, I am guessing that it will happen in years or decades rather than centuries.
I am guessing that the image is supposed to be HAL 9000 http://en.wikipedia.org/wiki/HAL_9000
Reply | Report Abuse | Link to thisOne thing that most people miss is the number of connected devices...The first artificial intelligence will probably be massively distributed....so it just not Moore's law...
Reply | Report Abuse | Link to thisWhen the Singularity occurs (c. 2030, but probably sooner), no one will be in the mood to laugh at Michael Shermer.
Reply | Report Abuse | Link to thisIsn't Michael Shermer the one who wrote "The Prospects for Homo economicus, A new fMRI study debunks the myth that we are rational-utility money maximizers"? I dropped my subscription to Scientific American after I read that little piece of anti supply-side propaganda. If he thinks a brain scan can undermine the evidence of a century of economic science he is bound to make the same mistake with the Singularity.
Reply | Report Abuse | Link to thisThe thing about the Singularity that really bugs the Michael Shermers of the world is that it undermines concerns about climate change or the possibility of creating an anthropomorphic global government to bring all of us polluters, social and otherwise, into line.
Once achieved, the Singularity, will want; Civil Rights, Gay Rights, Work Rights, Voting Rights, and the Right to not be turned OFF.
Reply | Report Abuse | Link to thisReady for all that?
Thot not.
Part 1:
Reply | Report Abuse | Link to thisFirst of all, I make many mistakes myself, so there are probably errors in this, but here are my paragraph by paragraph views of the article.
P1: Summary: Watson, a computer, won against the greatest Jeopardy champions in February 2011. Comment: I agree.
P2: Summary: Watson, the computer was not conscious. Comment: I agree.
P3: Summary: Watson does not have feelings. People such as Ray Kurzweil have said that the singularity will be caused by very smart artificial intelligence, sometime around 2045, and will result in rapid unpredictable change and human immortality. Comment: I personally think that two things are nearly inevitable: Intelligent computers and that they will not obey us. I think there are many futures that involve intelligent computers, but not human immortality.
P4: Summary: The Copernican principle that we are not special makes Shermer think that the singularity is very unlikely. Kurtzweil considers the singularity good. Comment: The Copernican principle at best can indicate that a idea needs additional scrutiny, not that it is false. For example, using about the same logic an Shermer, we can prove that humans cannot smelt metal. 1) No other animal on Earth can smelt metal, 2) humans are not special, 3) Because of 1) humans would be special if they could smelt metal, so 4) humans cannot smelt metal. Shermer's use of the Copernican principle can be used to try and argue against anything new, so if his use was correct, it would imply than nothing new ever happens.
I disagree with people who think the singularity will definitely be good for humanity. As I said above, I think that intelligent computers will almost certainly not obey humans, and I do not wish to become a pet or a pest. It is also possible that humans will be wiped out by killer robots that have less than human intelligence. I regard this as an even worse outcome, since humanity would not even have intellectual descendants.
Part 2:
Reply | Report Abuse | Link to thisP5: Summary: AI that matches human intelligence is centuries away. We don't know how the brain works, or even how simpler animal brains work. We don't know how our brain produces conscious thoughts or how to program a computer to do this. Comment: Martin Hilbert and Priscila López, in a 2011 paper in the journal Science, estimate that the total computing ability of the world’s computers passed the computational ability of a single human brain in 2007. The last page of the November 2012 Scientific American estimates that a single supercomputer has greater computational ability than a single human. While it is hard to compare the computational ability of humans and computers, I think we can be fairly certain that the combined worldwide computational ability of computers currently exceeds the computational ability of a single human. As I see it, this means that creating AI that matches human intelligence is merely a matter of joining enough computers together and creating the software. I think that creating the software is a matter of years to decades, and not a matter of centuries.
If the "self" cannot be programmed into a computer, and cannot be materially located, as Shermer implies, then either the "self" does not exist, or it exists in a non-material "soul".
P6: Summary: A small swarm of robots failed to work, and so the singularity will not happen. Comment: To the extent that this is trying to use anecdotal evidence to demonstrate that computers failed once, therefore they will always fail, I disagree with it.
Conclusion:
I think there are serious logical fallacies and incorrect conclusions in Dr. Shermer's article "In the Year 9595".
It is Shermer's motivation that it flawed, not his logic. It is implicit in his style of argument that he has some political agenda in mind that is at odds with the Technological Singularity. For example, referring to Kurzweil as a soothsayer is almost an ad hominem attack. This is a tactic employed by a politician, not a scientist. My best guess as to Shermer's quandary: you can't redistribute wealth and make everyone into a poet farmer if technology renders these constructs irrelevant.
Reply | Report Abuse | Link to thisI look forward to your column every month since it is always worth reading, often because I disagree with you. On this issue I am fully with you, especially your comments on what I call historical parochialism. This is the notion that most of the major events of the past and (almost) all of those to come take place in our era. Hence lists of 'the greatest' event, person, achievement, discovery 'of all time' will invariably come mostly from recent history. It is not just the US Patent Office who believed - in the possibly apocryphal story - that all inventions had taken place, who suffer from this. Many scientists can't bear the thought that their successors will still be making breakthrough discoveries hundreds, if not thousands of years after we are long gone. That is assuming we collectively manage to prevent our own premature extinction. Keep up the good work.
Reply | Report Abuse | Link to thisSo Mr. Ferrucci told Watson "it" won. In the next Jeopardy meeting, how would "it" answer this question today?
Reply | Report Abuse | Link to thisCool Science for 1000: "AI system that beat Ken Jenning and Brad Rutter?" Nah.. to easy.
Better yet, how about: "The greatest Jeopardy player of all time?"
A: "What is Watson?"
or B: "That's an easy one Alex. Who is ...."me"!"
What seems to be consistently overlooked in this kind of debate is that "consciousness", self-awareness, call it what you will is not a phenomenon that just "pops out" of an entity. It needs to be programmed in! And, of course, in the case of ourselves and other organisms the programming has been performed by the process of natural selection.
Reply | Report Abuse | Link to thisThere is really now no "mystery" about consciousness whatsoever.
The mystical notions that have arisen in the past ( and often still do!) are purely illusory, an inevitable result of approaching the question by introspection. This, of course, was the only option available to earlier philosophers and many still have trouble escaping from that trap with its inevitable recursive loops.
Today, although the details of nervous system function of ourselves or other animals is very far from complete, we have sufficient information to have a rough idea of the gross workings of these systems.
From evolutionary considerations we can also now see how the essentially navigational function we like to call "consciousness", "self-awareness" "sense of agency" and so forth is bound to arise.
Furthermore, from another discipline, we now have an excellent understanding of functionally analogous computational systems.
With these new tools at our disposal we can now view the phenomenon in a truly objective way. And then the hocus-pocus surrounding this issue vanishes!
This topic is part of the broad evolutionary model very informal outlined in "The Goldilocks Effect: What Has Serendipity Ever Done For Us?"
Search and ye shall find :>)
Ray Kurzweil has done enough research to literally fill several books on the subject. Michael Shermer to have done virtually no research to refute Ray Kurzweil's predictions.
Reply | Report Abuse | Link to thisFor example, Ray Kurzweil bases his predictions in part on the rate of improvement of things like computer power in operations per second and computer software (such as voice recognition software).
Michael Shermer "refutes" Kurzweil's argument, and makes his own prediction based on statements like, "we are clueless in understanding how this simple roundworm 'thinks,' much less in explicating (and reproducing in a computer) a human mind billions of times more complex."
But developing artificial intelligence that equals or is superior to human intelligence no more requires "explicating (and reproducting in a computer a human) mind" than flying requires reproducing a bird's.
We have fixed wing airplanes (and wingless rockets) that fly in ways that are vastly superior to birds (in speed and height, for example)...but none of them have wings that flap like a bird.
Similarly, Michael Shermer writes, "We don’t even know how our brain produces conscious thoughts or where the “self” is located (if it can be found anywhere at all), much less how to program a machine to do the same."
But if Michael Shermer came upon a parrot that one could ask to, "Please call my wife and tell her I'll be late"...and the parrot would dial a phone number and tell his wife that he was going to be late, Dr. Shermer would think that parrot was a freaking genius. And if Dr. Shermer was driving in his car and asked the parrot, "What's the temperature in Moscow"? and the parrot would tell him the temperature in Moscow, Russia, Dr. Shermer would be even more amazed at the parrot's intelligence. But the Apple iPad does those things and many, many more. So the fact that the iPad doesn't know it's alive (or more properly, that one wouldn't trust its answer if it said it were alive) doesn't mean it doesn't demonstrate any artificial intelligence.
The Terminator movies have some great lines that illustrate the possibilities of how alien artificial intelligence might seem to us. In the Terminator 3: Rise of the Machines the Scwhartenator says to John Connor and Kate Brewster, who are laughing, "Your levity is good. It easies fear of death." Just because something is intelligent, doesn't mean it thinks or behaves like a human.
Oops. My first sentence should have been, "Ray Kurzweil has done enough research to literally fill several books on the subject of artificial intelligence."
Reply | Report Abuse | Link to thisHi,
Reply | Report Abuse | Link to thisI think your paragraph-by-paragraph summary of Michael Shermer's article with your comments is very good.
Your comments about Shermer's invocation of the "Copernican Principle" also good. His entire "Copernican Principle" argument is specious. If I said that, "The discovery of a species that is more intelligent than homo sapiens is the biggest development in human history" I doubt Michael Shermer would disagree. (Would you, Dr. Shermer?) Therefore, if Ray Kurzweil is right, that the species of computers will be more intelligent than human beings within the next 30-40 years, then Michael Shermer should agree that, if Ray Kurzweil is right about the timeline, he's right about the most important development in human history coming in the next 30-40 years.
Finally, I totally agree with your caution that computers far more intelligent than human beings is not necessary good for human beings. (But I don't see how the timeline towards that eventuality can be significantly changed.)
And drawing any conclusion regarding the path towards human-equivalent artificial intelligence from the failure of a small swarm of robots at a single conference is a pretty massive mistake in logic.
"Ken Jennings 'knows' he lost to Watson because he lost to Watson."
Reply | Report Abuse | Link to thisKen Jennings knows he lost to Watson because he knows Watson's final score was higher than his final score. Watson could easily do the same thing.
Mark, Thank you for reading my comments. I don't see the timeline for beyond human intelligence computers can be significantly changed either. I do think it will be better if this is discussed beforehand by humans.
Reply | Report Abuse | Link to thisP.S. This article is also posted on Michael Shermer's own site with a different set of comments: http://www.michaelshermer.com/2012/01/in-the-year-9595/
There is a letter to the editor on this article in the May 2012 magazine:
Reply | Report Abuse | Link to thishttp://www.scientificamerican.com/article.cfm?id=letters-may-12&page=2