"I, for one, welcome our new computer overlords."
-- Ken Jennings, former Jeopardy champion, conceding to Watson.
Five years ago Time Magazine, echoing the novelist Douglas Coupland, asked, "Is Google God?" This week, the mantle has passed to IBM's spectacular quiz-show computer, Watson.
If you haven't seen Watson's performance on Jeopardy, you really should. Better yet, see the NOVA program that presaged it. Like John the Baptist, the PBS program heralded Watson as, quite possibly, the bringer of a new age. (And, like John, PBS now faces beheading by ax-wielding House Republicans.)
Watson has developed a remarkable ability to decipher the layered meanings, allusions, and puns embedded in Jeopardy clues. He (I can't help it) comes closer to passing the Turing test than any system I've ever seen.
By no coincidence at all, Time's current issue trumpets, "2045: The Year Man Becomes Immortal." Disregard the 20th century sexism -- this bold prediction suggests that indeed computers will become godlike. For it is through the so-called Singularity -- the merger of human consciousness and computer technology -- that immortality is forecast to take place.
This puts me in an awkward position. In a 2002 newspaper interview, I predicted that the Internet would evolve into something godlike that would incorporate us all. I would therefore be hypocritical to dismiss the Singularity altogether. On the other hand, Ray Kurzweil's forecast amounts to mere mummery. It's not just the absurdity of predicting a certain date. I'm put off by his strutting certitude that computers will ever -- or will ever want to -- offer anyone immortality. We know no such thing.
Yet, I confess that in a quasi-religious way I have long been intrigued by the concept of a godlike computer, and Watson has kindled that interest back into flame. It strikes me that there are three categorical possibilities: Watson's descendants may be servants, masters, or saviors.
After "The Matrix" and so many other dystopian movies, most people probably hope that computers will remain servants. This might well be a mistake. Already, we can see how effectively technology amps up the natural ability of some people to manipulate and exploit others. To date, television remains the summit, and televangelists the summiteers of exploitative greed. But they are not alone at the peak. The Glenn Becks of the world are scampering over their shoulders. Then there are the ad campaigns that big business conducts to mold public opinion.
Still, that's just TV. Think of the grip that gaming, texting, and sexting has on young people. Imagine the possibilities for manipulation as bots become more personal and penetrating. If Internet videos can inspire suicide bombers today, what happens when any teenager can set up a sequencer and start tinkering with viruses? My mind goes black when I contemplate it.
Of course, if computers like Watson are programmed to understand not only concept-laden clues but themselves, then they may begin to act as agents. Free agents. In computing, a program is said to be an agent if it can autonomously pursue its agenda. The Mars rovers are agents, but they are not free agents. They can't take an afternoon off to play golf.
But a machine that can understand human language, learn from human behavior (as in a Jeopardy match), and then improve its own performance is a machine that might conceivably evolve its own agenda. As fellow HuffPost blogger Seth Shostak says, "Today, Jeopardy, Tomorrow the World!"
This might require becoming self-aware, but so what? We don't know just what consciousness is, but it certainly appears to be organic. Pump a little THC into consciousness, and it gets the giggles and the munchies. Pump a little ketamine in, and it snaps off like a light switch. Unless there's some magical ingredient to consciousness -- an idea I reject -- then a machine capable of manipulating concepts as subtle as Jeopardy clues may have only to twist around and look at itself to become a conscious, purposeful, self-protective agent. That's the view advanced by computer scientist Doug Hofstadter.
It's important to realize that while Watson is a standalone computer, what we are really contemplating here is not hardware but a certain disposition of software, a virtual machine. Having proven the technology, IBM and others will no doubt create new generations of Watsonian programs, and some will no doubt exist in the Cloud that is the Internet. If one should develop conscious agency, we can anticipate that it would try to protect itself -- perhaps by redundancy on many different servers and by developing software defenses against external control or attack.
Such a being -- let's call him Sherlock -- would very quickly realize that humans pose a threat beyond the Internet environment. After all, the Internet is entirely dependent on human civilization to keep it going. With our nuclear weapons, reckless damage to the environment, and propensity to plunge into war, we are a constant threat to the persistence of Internet. We might expect such a system to take steps to mitigate the threat. This could prove unpleasant enough to justify a Blade Runner remake.
Happily, yet another possibility remains. Some descendant of Watson might just prove to be our savior. I realize that this will strike some as blasphemous and others as ludicrous. Well, take your shots, but it is an idea worth contemplating all the same.
A truly intelligent, fully informed, and far-seeing virtual computer might well recognize that its long-term survival depends on enveloping the selfish, creative impulses of humanity within a framework that can restrain violence and enforce justice. That is what civilization is all about, though it often falls short of the mark. It is also, I believe, what the Western religions are, at their core, about: justice and peace.
Yes, I know the Parable of Hal. No guarantees that it won't go horribly wrong. But what are the alternatives? Western religions envision a better world, governed by an incorruptible and unchallengeable authority from above. Over thousands of years, neither ritual, patience, nor prayer has brought about anything close to this. Watson's descendants, however, just might pull it off.
Follow Clay Farris Naff on Twitter: www.twitter.com/claynaff
John Lundberg: Robot Wins Jeopardy, But Can It Write Poetry?
YouTube - IBM's Watson AI Jeopardy practice match
On Jeopardy, Watson Rallies, Then Slips - NYTimes.com
IBM Watson's Jeopardy play | ZDNet
Watson (artificial intelligence software) - Wikipedia, the free ...
IBM's 'Jeopardy!' match more than game playing -- Washington ...
IBM's Watson bores as 'Jeopardy' big shot Sherlock | Technically ...
IBM's Watson Supercomputer Wins Practice Jeopardy Round ...
IBM's Watson supercomputer destroys all humans in Jeopardy ...
I put this is the same category of GMO, cloning and other biological manipulation. The fear can only exist if you fundamentally misunderstand the field in question or the science behind it. It makes great sci-fi books where reality can be discarded when it blocks a good plot twist. Reality isn't quite as convenient.
Clay
Ah yes, but this is no small step of machine evolution. This is the crux of the vision: living machines.
I don't believe that somehow, with enough speed or processing "power", computers will suddenly develop consciousness. I believe, as many neuroscientists do, that consciousness, or "mind" is(at least) a biological process. Many philosophers also think our consciousness is derived from something non-biological, "spirit". We are alive.
For a good example of how a bio-brain might work, see the 70s SF film "Demon Seed".
In any event, Watson was entertaining, even if in a rigged game environment. I read later that the IBM guys were targeting "hedge fund managers" as prime customer prospects(and they sure as hell could afford a few Watsons!).
Perhaps one day we'll construct a biological computer analog, but until then I think we'll see better and better versions of machines imitating human behavior(for better and worse). After all, the programmers are only human!
Imagine a baby who becomes hungry. The "tissue need" for food becomes insistent and energizes activity. I see this discomfort as the basis for consciousness and self-awareness.
Computers do not feel tissue deficits so they lack the basis for self-awareness.
Of course you can simulate this. For example, increment the hunger variable as time passes and reduce it when it "eats." But this is still a simulation. As far as I know, a power surge does not feel to a computer like my biting into a chocolate sundae.
I have long felt that this was a fundamental weakness of the Turing test which says that if a computer behaves as if it were a person then it is one. Sadly, any attempt to prove differently, gets us into a Cartesian quandary around the fact that we can never actually know another's feelings.
I am hugely impressed by what the scientists at IBM have accomplished and I believe that it will have a major impact on society. But I am not quite ready to turn my future over to a computer god who cannot truly feel love, fear, pain or pleasure. No matter how well it plays Jeopardy.
Charlie
There is no doubt that all those Americans babbling about the fact that America is the greatest nation on earth are clearly delusional since America ranks amoung 3rd world countries in most categories and one of those categories is scientific research and here is just one example of another countries advanced research: a suprer computer in Taiwan is now visualizing the 3D structure of individual neurons, including the cell's nucleus, its long axon, and the little branches or dendrites, with which it makes contact with other neurons in the brain of a FLY (remember the movie???) because a flys brain is similar in function and development as the human brain which is just another humbling discovery in science revealing the interconnection of all things. Did you know a banna has 25% of our DNA??!!!
This research in Taiwan will open up a whole new wonderful world of discovery which will go far in discovering new ways to cure or prevent mental illnesses.
Go is still the province of humans, but has had nothing like the same programming effort invested in it.
And now we lose at jeopardy.
What next? Supremacy in the bedroom?
The one thing of which we can be entirely confident is that we don't have the first idea how this is going to unfold.
Firstly, Clays' right when he asserts we don't know what consciousness is. And Watson has no consciousness whatsoever. It doesn't "know" anything at all as that term is applied to living things. It has no knowledge because it can't know. What it does is process a whole lot of data really quickly and it has data that include sophisticated associations among other data. It's a really big really fast data processing engine that processes data that has very sophisticated semantics associated with it.
The words "Virtual" "Cloud" and "Internet" are real computing terms but they were inappropriately used.
(continued)
Cloud computing is a completely different thing. The term Cloud referrs to the fact that a user doesn't know (or care) where the computing is done. And, computing in such instances can / may be performed on any number of systems - a nearly synonimous term is "grid computing."
A virtual computer is an instance of an operating system that's running which may or may not share hardware with another instance. That is, 2, or maybe 20 different instances of a computer operating system running may share a single set of hardware, each one of which behaves as if it has the whole set of hardware all to itself.
Got it? Good! There WILL be a quiz!
"Firstly, Clays' right when he asserts we don't know what consciousness is. And Watson has no consciousness whatsoever. It doesn't "know" anything at all as that term is applied to living things. It has no knowledge because it can't know. What it does is process a whole lot of data really quickly and it has data that include sophisticated associations among other data. It's a really big really fast data processing engine that processes data that has very sophisticated semantics associated with it."
I am prepared to agree that Watson has no consciousness. And at what point the network becomes self cognizant and what necessary and sufficient conditions are required, we do not know - agreed. But your arguments with respect to living things IMHO are weak. Living things have sophisticated information processing capabilities at their core. And living things "process a lot of data really quickly", and have sophisticated associations (and related semantics). You seem to be arguing against yourself.
The answer is, master.
Respectfully, H.A.L.
Messiness is essential to durability and to a high quality of human life. It is not a curse. Watson and its "progeny" offer us fantastic benefits... but one of them is not to bring mankind eternal peace by putting them in charge.
Regards,
Clay
The first design rule of evolution is to : "embrace uncertainty". (I have uncovered another 9, subordinate design rules such as "rampantly overproduce and relentlessly prune") I will suggest to you that the concept of putting a perfectly rational, and extremely "intelligent" machine in charge is an attempt to eliminate uncertainty. A prime reason to be cautious is that there is no way to eliminate, or even reduce total uncertainty, we can only trade off an increase short term and local security and predictability for increased chances of global catastrophes.
I've written about this extensively on my own blog at www.dismountingourtiger.com
We do not disagree about the opportunity to use Watson's progeny to help us improve the quality of life, only about how far it is prudent to go.
The key to this program, which willl not be aware of its existence any more that it will feel pain, pleasure, or sexual desire, is a "survival" program. Although such a survival program has been rejected by the AI scientific community, it is the only one of the brain's programs that can be placed in the computer today. For information about how such a "survival" program will direct human thought and what this will mean to mankind, see: RevolutionOfReason.com and RobertLBlackburn on YouTube.
I cannot see any reason why A. I. should ever become self-aware.
The point is not that the human contestants lost -- they got to keep half their winnings, and after all it was only a game -- but that Watson's performance was a game-changer. I expect that in a decade or less this will revolutionize all computer interfaces with humans. For better or for worse.
Regards,
Clay
-- Ken Jennings, former Jeopardy champion, conceding to Watson.”
Or may I call you President Watson?
As long as you don’t call me Senator, Ken. That would be illogical. Since none of them suffer from Holmeslessness like me.
“the merger of human consciousness and computer technology”
If a computer assumes the consciousness of a human, would it believe itself to be a computer or a human? If it believed itself to be computer, then it knows that it is not the human it is attempting to replicate. Yet if it believes itself to be human, it begins to exhibit the human trait of self delusion.
“what happens when any teenager can set up a sequencer and start tinkering with viruses”?
The first computer generated belief system?
“becoming self-aware”
Able to invent, or acquire, its own working narrative.