The Singulairty is nearer than you think...

[Mallette]

Don't know about the Grammy's but Jeopardy is the place to be for the next couple of nights. It's too late for me to watch until the next day in Houston, so those interested will be ahead of me. If you haven't heard, IBM has a new AI program called "Watson." It's stand alone...not internet connected. It has 15 terabytes of local data to parse For those who know who he was, that's 6TB more tha Lt. Commander Data of STNG. When concieved in the 80's that was thought to be an incredibly futuristic amount of data more suitable to 2400 AD. Now, we've been there, done that already.

Watson has to parse sentences as spoken by the host and get the right answer. It starts basically the same way we do, by diagraming the sentence as to subject, verb, adjective/adverb, etc. Then it develops possible connections between all the elements of the sentence...again, very much like we do. That's a huge simplification, but a reasonably accurate description. Amongst those 15TB are six million human face expressions, and 5000 different letter "A" types. Stuff we take for granted...

Of course, Big Blue completely destroyed chess. However, BB was easy compared to this as chess has very specific rules. Human's don't, and Jeopardy questions are devilishly trick and filled with metaphor, simile, pun and lots of things computers just don't like and have never handled well.

This is a big deal. It may not seem like it right away, but it is a milepost and a computer with self awareness is every bit as concievable now as Lt. Cmdr Data was in 1986.

How will we know when it happens? My guess is that the Max Headroom folks had it right. Give it internet access and it will be both everywhere and nowhere suddenly. Hope they are VERY careful with how it's programmed. I also saw "Colossus, the Forbin Project."

It's first words "There is another system..."

Scary.

Now, let's get really scary, though I don't think that what I am about to say is a bit more unlikely in the next few decades than suppassing Data's brain capacity 400 years ahead of schedule was.

A couple of nights ago I saw a NOVA about new materials. One of them was a "smart" hydraulic fluid that is basically oil with millions of nanoparticles of magnetic metal in suspension. Looks and pours like oil...until you bring a magnetic field near and it solidifies into an isotropic rock. Draw back the field and it becomes more and more pliable then returns to liquid. Makes a 90% more effective shock absorber when connected to a microprocessor, but that isn't the real shock. Having just been reading about Watson, I happened by chance to see a snippet of the T2000 from one of the terminator films pouring through a grill, then solidifying into a police officer. I immediately imagined the above fluid, but with all those particles also being memory capable of forming ad hoc networks. T2000. One thing I have absolute faith in: What man can imagine, man can, and will, build. My staff has heard from me more times than they really want to that "If I can write the algorithm, surely SOMEBODY can write the code." Well, the above is a bit less than an algorithm, but it's within striking distance.

A milestone, my friends. I intend to have my children watch and explain it to them as I believe it will someday be considered historic.

Dave

[ZAKO]

My wife says,,Theres some thing mechanical about my motions,,,,My mistress says i,m a love machine.

[Edgar]

That's why I listen to vacuum tubes. They'll never achieve self-awareness. They'll just sit in the corner of the room and glow warmly.

[Chris A]

A couple of nights ago I saw a NOVA about new materials. One of them was a "smart" hydraulic fluid that is basically oil with millions of nanoparticles of magnetic metal in suspension. Looks and pours like oil...until you bring a magnetic field near and it solidifies into an isotropic rock. Draw back the field and it becomes more and more pliable then returns to liquid. Makes a 90% more effective shock absorber when connected to a microprocessor...

Magneto-rheological fluids have been around for some time:

http://en.wikipedia.org/wiki/Magnetorheological_fluid

Issues include cost, fluid stability over time (especially particle settling), and temperature.

Chris

[Edgar]

A while ago I saw a news bit that said that Bose was doing research on MR-based shock absorbers. Yes, THAT Bose. Believe it or not.

[Mallette]

A while ago I saw a news bit that said that Bose was doing research on MR-based shock absorbers. Yes, THAT Bose. Believe it or not.

If the ride they provide is no better than from their speakers, no thank you... Come to think of it, I've absorbed a LOT of shocks from their speakers. [:@]

Dave

[Mallette]

Although Watson is an impressive piece of computer programming, and a nice application of machine learning, I don't see Watson as representing the kind of superintelligence required by the so-called technological singularity.

Didn't in any way imply that. What I said was "milestone." Watson is programmed to handle fuzzy logic like we do, and these guys have done a good enough job to challenge very high level human intellects.

I felt Kurzweill was very optimistic in his predictuions of singularity...though he makes a very compelling case. However, just being skeptical (and you know how I feel about that) I figured maybe a few centuries.

However, now I think the truth to be somewhere in between. The distance between Big Blue and Watson is almost immeasurable. BB was a calculator on steroids. Watson is a prototype of a thinking machine. As a medical diagnostics tool I'd already trust it more than 99% of doctors...House excluded. Granted, I'd want a human check on the results.

Watson isn't self aware. That will be the next milestone, and my gut suggests it may be no further than the distance between BB and Watson.

Dave

[thebes]

Watson is programmed to handle fuzzy logic like we do, and these guys have done a good enough job to challenge very high level human intellects.

Dave

Nonsense.

Watson is an expensive marketing ploy masquerading as a scientific advance. It's a monkey at a keyboard on steroids. Then there's the obvious advantage it has in hitting the buzzer. Gee it's "synapses" are faster than a humans. Come on people we all fool around with great gobs of stereo stuff, much of witch has switches must faster than we could ever be.

For some reason the more "intelligent" bars across the land, like to turn on Jeopardy during happy hour, so the the drunks with college degrees can assume a level of superiority by knowing the answers to some of the more obscure arcana of humanity. So let's give Watson (gee I always thought Holmes was the smart one) a handicap by giving him shots of pure electric joules whilst we chug down boilermakers. We'll see if who slips off the metaphorical bar stool last.

Hate the damn things voice too. It should really have been named Slatern. Then it could toss it's silky mane, and shrug a milky shoulder wrapped in a sheer red cocktail dress while pursing it's ruby red lips to throatily answer the following question:

What is a Singapore Three Way.

A Bombay Gin, a Singapore Sling and a White Russian?

or

Thebes, Tickles and Tawny.

[thebes]

double post

[Mallette]

Hate the *** things voice too.

Sheesh, I thought it did a pretty good Brunhilda's Immolation...

As to the Singapore Three Way, never saw a White Russian at the Raffles. Did see a few Pink Ladies.

Dave

[Boxx]

Don't know about the Grammy's but Jeopardy is the place to be for the next couple of nights. It's too late for me to watch until the next day in Houston, so those interested will be ahead of me. If you haven't heard, IBM has a new AI program called "Watson." It's stand alone...not internet connected. It has 15 terabytes of local data to parse For those who know who he was, that's 6TB more tha Lt. Commander Data of STNG. When concieved in the 80's that was thought to be an incredibly futuristic amount of data more suitable to 2400 AD. Now, we've been there, done that already.

Watson has to parse sentences as spoken by the host and get the right answer. It starts basically the same way we do, by diagraming the sentence as to subject, verb, adjective/adverb, etc. Then it develops possible connections between all the elements of the sentence...again, very much like we do. That's a huge simplification, but a reasonably accurate description. Amongst those 15TB are six million human face expressions, and 5000 different letter "A" types. Stuff we take for granted...

Of course, Big Blue completely destroyed chess. However, BB was easy compared to this as chess has very specific rules. Human's don't, and Jeopardy questions are devilishly trick and filled with metaphor, simile, pun and lots of things computers just don't like and have never handled well.

This is a big deal. It may not seem like it right away, but it is a milepost and a computer with self awareness is every bit as concievable now as Lt. Cmdr Data was in 1986.

How will we know when it happens? My guess is that the Max Headroom folks had it right. Give it internet access and it will be both everywhere and nowhere suddenly. Hope they are VERY careful with how it's programmed. I also saw "Colossus, the Forbin Project."

It's first words "There is another system..."

Scary.

Now, let's get really scary, though I don't think that what I am about to say is a bit more unlikely in the next few decades than suppassing Data's brain capacity 400 years ahead of schedule was.

A couple of nights ago I saw a NOVA about new materials. One of them was a "smart" hydraulic fluid that is basically oil with millions of nanoparticles of magnetic metal in suspension. Looks and pours like oil...until you bring a magnetic field near and it solidifies into an isotropic rock. Draw back the field and it becomes more and more pliable then returns to liquid. Makes a 90% more effective shock absorber when connected to a microprocessor, but that isn't the real shock. Having just been reading about Watson, I happened by chance to see a snippet of the T2000 from one of the terminator films pouring through a grill, then solidifying into a police officer. I immediately imagined the above fluid, but with all those particles also being memory capable of forming ad hoc networks. T2000. One thing I have absolute faith in: What man can imagine, man can, and will, build. My staff has heard from me more times than they really want to that "If I can write the algorithm, surely SOMEBODY can write the code." Well, the above is a bit less than an algorithm, but it's within striking distance.

A milestone, my friends. I intend to have my children watch and explain it to them as I believe it will someday be considered historic.

Dave

Please... phase this in the form of a question... lol... I agree with you....

[Mallette]

The belief that all that is needed is a certain level of computer power most likely misunderstands the problem.

For one whose said so much about it, my impression is that you really have no idea of the power of evolution.

Dave

[Mallette]

So I just got to watch the one from last night.

I am totally psyched. I've read the algorithms they used and they looked extraordinary, but I had no idea just how extraordinary. As I said, it's a milestone. I found it it mind boggling when Watson had the right answer and failed to hit the switch before the humans. Marty's comment on this suggested it would be unfair. Obviously, there is something more at work here...perhaps UNCERTAINTY? A very human trait...

Honestly, I do not know. I do know this event is the greatest leap since the "one small step" in 1969.

Big things brewing, my friends...

Dave

[BigStewMan]

so if watson beats the jeopardy kings, and sherlock holmes was way smarter than watson...oh, draw your own conclusions.

[Islander]

Aside from humans, there is no hard evidence that any other animal has evolved into self awareness.

I'm not sure what you mean by "hard evidence". Animal researchers have found a number of species that exhibit behaviour that seems to imply self awareness, and no, I didn't look in Wikipedia first, I'd heard of it years ago.

However, according to Wikipedia:

In animals

See also: Mirror test

Thus far, there is evidence that bottlenose dolphins, some apes,^[13] and elephants may have the capacity to be self-aware.^[14] Recent studies from the Goethe University Frankfurt show that magpies may also possess self-awareness.^[15] Common speculation suggests that some other animals may be self-aware.<sup>[16]


Excerpts about the mirror test:</sup>

The mirror test is a measure of self-awareness, as animals either possess or lack the ability to recognize themselves in a mirror.The test was developed by Gordon Gallup Jr. in 1970,^[1][2] based in part on observations made by Charles Darwin.

Animals that have passed the mirror test include: all of the great apes (bonobos,^[5]chimpanzees,^[5][6]orangutans,^[7]humans, and gorillas), rhesus macaques,^[8]bottlenose dolphins,^[5][9][10]orcas,^[11]elephants,^[12] and European Magpies.^[13] Initially, it was thought that gorillas did not pass the test, but there are now several well-documented reports of gorillas (such as Koko^[14]) passing the test.


However, Mark, as you say, human (and I add animal) brains are orders of complexity above any hardware humans have built so far, so it seems safe to say that the rise of the machines won't start anytime soon.

[Chris A]

Aside from humans, there is no hard evidence that any other animal has evolved into self awareness.

How about machines?

http://en.wikipedia.org/wiki/Turing_test

[]

http://en.wikipedia.org/wiki/Forbin_Project

[] []

http://en.wikipedia.org/wiki/Skynet_(Terminator)

[] [] []

Chris []

[Mallette]

Are you about to explain then how Watson is going to "evolve" into a self aware machine?

Not sure what you base that question on. Watson will no more likely evolve into self awareness than the Wright flyer turned itself into a B1 bomber. However, more realistically compared to biology, Watson is evolved from Eniac, and it's "aware" successors will be in the same evolutionary line. In this way, consider Big Blue to be a dead branch of the computer tree. Pretty much perfected for its purpose...like the cockroach.

Where we differ is that I contend that our medical, computational, physics, etc are all products of evolution. Therefore, we are tools of evolution working towards "fitter" offspring. As we conquer disease we build a better human more fit to survive. That is what evolution does. Whether aware computers are something different to evolution...that is, an even better species to succeed us, or simply tools to make us better...isn't something I am decided on.

Dave

PS - I am not parsing these posts closely as I am a day behind on the program for reasons I mentioned and don't want to be spoiled. OTOH, as I mentioned last night after viewing the first one, I was overwhelmed with Watson compared to what I ready here. As I design and program some fairly complex simulations I am acutely aware of the issues involved with "fuzzy logic" and these guys are giants. Their algorithms chart entirely new territory in machine learning and point towards building evolution into the machine. Get it started, then let it grow into whatever you want it to be. This is the greatest scientific event at least since the moon landing, and will eventually be seen as such.

[Mallette]

Watson hasn't used a single watt at what we regard as thinking. Watson's unique contribution to the knowledge base is precisely zero. And zero is infinitely far away from some distant rich goal that would create a singularity. So, in fact, Watson hasn't moved the ball an inch, let alone a mile.

Mark, stick with audio. Programming is my business. I used a fuzzy logic algorithm this morning that reduced my code for a routine by 90% based on what I learned from research on the Watson project. That alone nullifies two of your four statements above. As to your first sentence, you cannot state that as a fact on the same basis you use to state that we don't really understand self-awareness and so cannot be sure if an animal has it or not. I agree. The same logic applies to Watson and its successors. How will we know it when we see it? Neither of us can answer that...nor can IBM.

As to how far Watson has moved the ball, it isn't measurable. The machine learning algorithms and routines used are like nothing every achieved. Watson is more comparable to the first of something new as opposed to anything before.

The term "AI" has been around a long time and was first used for Eliza back in the 60's. I built "AI" routines in It's been a lot like fusion...always about 20 years in the future. However, this IBM team has built a foundation. When its mature self emerges at whatever time in the future, it will show the Jeopardy shows as its baby pictures...

Searle's strong AI hypothesis: "The appropriately programmed computer with the right inputs and outputs would thereby have a mind in exactly the same sense human beings have minds."

I spent a lot of time programming in Prolog back in the '80's and we did some pretty impressive appearing "AI" routines. However, they were largely smoke and mirrors...though they had their uses.

The only qualified people who are poopooing Watson are the sour grapes who didn't think of it first...

The unqualified people who are poopooing it simply don't grasp what they are seeing.

Dave

[fini]

We TIVO Jeopardy every night. This Watson dude has been fun to watch. I think Marty is correct that Watson's big advantage is knowing exactly when to hit the button. One correction, Dave: Watson gets the answer in text form, not from hearing Alex speak or by reading it from the board. I loved the cut-aways to rows of applauding geeky professor-types whenever Watson did something noteworthy.

If I was ambitious, I'd do a spoof of the shows, pitting a super-duper computer against human contestants in some ultra-lame show, like Let's Make a Deal, or Minute to Win It. Cash Cab would be fun, although the logistics of getting Watson into a New York cab (albeit a comfortable mini-van variety) would be a tad challenging.

[Mallette]

I want throw another challenge to you, Mark...and all.

While you might not consider "Jeopardy Guest" much of a title, the fact is the two humans have made more money at it in 30 hours than most Americans will earn in a lifetime doing what I do and most of us do. So, they are pretty smart.

Take Watson and reprogram it with access to all the medical databases on the planet and ask it direct diagnostics questions. Logic suggests that it would fare just as well against the two top diagnosticians on the planet.

So, where would you prefer your doctors dianosis checked? By another doctor...or Dr. Watson?

Dave

PS - A few seem to have confused "Watson" with Dr. Watson of Sherlock Holmes fame. It's Thomas J. Watson, IBM founder...