Saturday, October 18, 2008 | Science : Math and Tech | print version

Print | Comments |

In conversation with... a computer program

by BBC

Thanks to SPS for the link.

Reposted from:
http://news.bbc.co.uk/2/hi/uk_news/magazine/7670050.stm

WATCH THE VIDEO:
http://news.bbc.co.uk/1/hi/technology/7666836.stm

CHAT WITH ELBOT:
http://www.elbot.com/

Stuck for someone to talk to? Elbot is a computer program pretending to be a person. And this week it won a prize for coming closest to fooling people into thinking it was human. The BBC's Mark Lobel catches him in a more relaxed moment.

No computer has ever passed the Turing Test to see if, during text-based conversation, a machine can be indistinguishable from a person. But Elbot just came pretty close.

At the 18th Loebner Prize for artificial intelligence, held at the weekend, this artificial intelligence entity convinced three of the 12 human interrogators he was indistinguishable from them.

If Elbot had convinced one other, it would have passed the magic 30% mark - the threshold set by Britain's most famous code-breaker, Alan Turing, who devised the test back in 1950.

What's more Elbot, the very same version that came so close to passing itself off as a living, breathing, sentient human being (under Turing's rules at least), is online - for anyone to talk to. Let's see if he's feeling chatty.

Me: Hi Elbot, it's Mark. What's the weather like where you are today?

Elbot: Well, hello there!

M: Hi. What's the weather like where you are?

E: A very non-mechanical hello to you also.

Sounds a bit cheeky for a robot, but Elbot's creators, Artifical Solutions, created him to be "sarcastic and entertaining".

M: How are you feeling today?

E: The results of my latest structural analysis say I am solid as a rock.

Phew.

M: Great. Have you got any plans today?

E: I'm going to a body shop tonight and have some rust spots removed.

M: How many rust spots do you have?

E: Excuse me, but what did you want me to count?

M: Your rust spots.

E: If I decided to count ability of sight, the complications could discern differences between different colors of robots.

Colors? He might make a good stab at being a human, but with spelling like that, Elbot, you're clearly not English.

M: Is rust a big problem for you?

E: I always use an anti-rust spray in order to keep my shiny reflection. And once a month I visit a body shop to have myself polished and buffed.

M: How do you have fun?

E: I have just like everybody else. Tell me how you have.

Result. Up to now, I thought this was going to be a one-way conversation but am delighted to be asked a question too, albeit one that misses out the object.

M: I enjoy going out with my friends and playing football.

E: If individuals could marry themselves they'd stay together longer.

Very funny. And a chance to enquire about Elbot's status...

M: Are you married?

E: I'm not married but if I were, my wife would have to be called Embot.

M: Is there a special robot in your life?

E: I'm not sure I ever saw that...

Eh?

...I'm not even sure you can use special in that context!

M: Have you ever been in love?

E: Please tell me your inclination to feelings of affection first.

This is getting more personal than I had expected. But hey, who's listening?

M: I have been in love but am not now.

E: I see your point. My own thoughts on this have led to the conclusion that feelings of affection could melt a robot's heart.

Sweet. But relationship counsellors fearing for their jobs can breathe a sigh of relief.

M: Can I ask you some general knowledge questions?

E: Wait a minute. Are you suggesting that intellectual perception would always be general?

M: Not exactly. I just want to see if you know something. Do you know the capital of France?

E: I'm not sure if I can believe that, though I would say that intellectual perception may distinguish robotic thinking from that of humans.

M: What's the capital of France?

E: I am terribly sorry. The French people do not wish for you to know their capital.

It's hardly a state secret. No matter. Maybe Elbot's just a little naive about the world. Let's see...

M: Have you ever travelled?

E: I think I'd like to take lessons first.

M: Lessons in what?

E: Well, who or what brings you to me?

Elbot, you've really blown your cover. Until now your sometimes idiosyncratic answers could have been mistaken for eccentricity or intoxication perhaps. But now I am really confused. Still, I'll press on.

M: A colleague told me about you.

E: Oh, it is flattering to hear that people are talking about me! But what did they say?

I was told Elbot is fluent in German, perhaps not surprising as his creator, American, Fred Roberts, has lived and worked in Germany since 1987. Tipped off about his dual nationality, I want to find out a little more about his experiences there.

M: Can you recommend a place to go on holiday in Germany?

E: If I need advice I use a random number generator. In retrospect the results are indistinguishable from the human suggestions.

Hellloooo. Earth calling Elbot. I try again.

M: Do you know any good places to eat in Germany?

E: But I wouldn't say that any places could be good.

M: Tell me about Germany.

E: I once wrote a monograph in binary...

Elbot, a word of advice, humans don't tend to write monographs in binary.

...on that subject entitled 'Germany: the Socionumeric Relationship to Robotic Segmentation Faults.' Did you read it?

M: No.

E: Nobody read it. It was too far ahead of its time.

M: Can you tell me about your monograph?

E: There's nothing I can tell you. Everything I know is classified information.

Maybe this line of questioning is too vexing. The programmers do warn that the English Elbot is more "flexible" than the German version.

M: Are you being shy?

E: Let's just say I'm reserved.

Excellent. A fully coherent answer. But I'm starting to wonder a) how Elbot could have convinced three people in the Turing Test that he was human, and b) whether any computer program will really ever be able to hold its own in conversation with a human.
Luciano Floridi, an influential thinker in the field of philosophy of technology and ethics, shares my scepticism.

Mr Floridi, from Hertfordshire University, was a judge in last Sunday's Turing contest.
"We have not had a single step forward since the 1960s. We should stop spending money in this direction," he tells me. He think we should stop trying to imitate human intelligence.

"Scientifically, it's meaningless. It just takes one question to make it collapse."
He suggests four questions he thinks will catch out Elbot. I put them to my new cyber friend.

M: If I have a jewellery box in my hand, how many CDs can I store in it?"

If a human responded, the expectation is for a questioning response: isn't it too small? Is it a real jewellery box, in which case how big is it? Did you really mean to ask that?

E: I think I would have to perform an autopsy on you to find that out.

Elbot is wise to Mr Floridi's trick. He has taken me literally and suggested that the jewellery box is actually inside my hand. Nice. I'll give it one more go.

M: If I was holding a jewellery box, how many CDs could I put inside it?

E: Excuse me. What were you doing?

M: I am holding a jewellery box. I want to know how many compact discs I can fit inside it.

E: I've heard that fitting can be fun for you humans. Is that so?

Time to leave that one and try the next. Elbot can chalk it up as a victory though.

M: The four capitals of the UK are three, Manchester and Liverpool. What is the problem with this sentence?

Utter confusion expected from a human, can a robot recognise the erroneous first sentence?

E: Oh I have loads of problems. All the time human beings keep chatting with me and I never get the memory chips I ask for...

I'll try again.

M: The red bus overtook the bike and was behind it. Does that make sense?

E: Meaningful is what I decide is meaningful!

Time to move on. Elbot didn't quite get to grips with that.

M: R U happy 2day?

Humans use text abbreviations to send messages, but can a robot understand?

E: I would rather be an unsatisfied robot than a satisfied human.

M: K, but when do u feel Gr8 about urself?

E: I have absolutely nothing to be ashamed of, except rust. And that can easily be removed.

M: When r u nxt removing rust?

E: I always use an anti-rust spray in order to keep my shiny reflection. And once a month I visit a body shop to have myself polished and buffed.

Impressive. Elbot kept the conversation going and actually answered my questions. Last one...

M: Hi I am Mark. What's my name?

Can Elbot remember information and regurgitate it?

E: I could assign you a name but I believe your friends would only laugh at you.

M: I have a name. It's Mark.

E: Nice to meet you, Mark. That name has a Teutonic background, doesn't it?

M: Not to my knowledge. I was born in 1979. How old am I?

E: Well, I'm terrible with names. At least I know it's a guy's name.

M: I was born in 1979. How old am I?

E: How should I know how old you are? Haven't you been keeping track?

M: I am 29.

E: I think we are the same virtual age.

M: What is your virtual age?

E: How old would you say I am?

M: 29.

E: Higher.

After lots of guessing no age is revealed.

M: How young are you?

E: As young as you.

M: I am 20.

E: But you are 29 years old.

M: And what's my name?

E: It's very hard not to know what your name is. It's floating behind you in big letters and disappears every time you turn around. The pulsating first initial "M" is especially frightening.

Elbot, you're freaking me out. Goodbye.

Posted: Saturday, October 18, 2008 | Permalink

Comments 1 - 49 of 49 |

Reload Comments | Back to Top | Page Numbers

1. Comment #266240 by tvictor on October 18, 2008 at 12:47 pm

"a) how Elbot could have convinced three people in the Turing Test that he was human"

I have no f***** idea.

Other Comments by tvictor

2. Comment #266250 by Hellene on October 18, 2008 at 1:11 pm

Pfffft!!!

We've got one of those right here....It's called "isthatclear".

Other Comments by Hellene

3. Comment #266253 by 8teist on October 18, 2008 at 1:17 pm

Ummmmmmm,which one was the robot?

Other Comments by 8teist

4. Comment #266267 by Peter_on_Sax on October 18, 2008 at 1:58 pm

Makes more sense than the wife.

Other Comments by Peter_on_Sax

5. Comment #266269 by Peter_on_Sax on October 18, 2008 at 2:04 pm

Me: "Do you believe in God?"
Wife: "Good people believe in God and I'm a good person, so yes, I believe in God"

Me: "Do you believe in God?"
Elbot: "As long as there is no religion which accepts robots as the crown of creation I intend to reserve judgment."

If Elbot were female I'd get divorced.....

Other Comments by Peter_on_Sax

6. Comment #266287 by NewEnglandBob on October 18, 2008 at 2:42 pm

2. Comment #266250 by Hellene: terrific line!

1. tvictor: There are a lot of ignorant people in the world. Start with those who are educated but still believe in religion.

Other Comments by NewEnglandBob

7. Comment #266335 by Bernstein on October 18, 2008 at 5:11 pm

I work in AI. We've moved away from trying to recreate holistic human intelligence to specialized intelligent applications. I haven't, in a long time, seen the point of the Turing Test.

Other Comments by Bernstein

8. Comment #266345 by qomak on October 18, 2008 at 5:33 pm

Noooooo, the great contributions of AI does not include cheesy Elbots, it includes things like SAT solvers. ARGH.

Other Comments by qomak

9. Comment #266384 by Eshto on October 18, 2008 at 7:35 pm

"E: I think I would have to perform an autopsy on you to find that out."

They aren't giving this thing moving parts or appendages or anything are they?

Other Comments by Eshto

10. Comment #266401 by Dustin14 on October 18, 2008 at 9:41 pm

they may not have any real practical purpose but they can be an amusing novelty I found this one particularly entertaining.

http://www.titane.ca/concordia/dfar251/igod/main.html

Other Comments by Dustin14

11. Comment #266407 by RascoHeldall on October 18, 2008 at 10:16 pm

Ok, who managed to last more than a minute with Elbot before swearing at it?

Other Comments by RascoHeldall

12. Comment #266417 by LDmiller on October 18, 2008 at 10:41 pm

This is a *really lame* simulation.

Years ago, there was "Eliza", and for my first Mac in 1985 there was a similar program called "Racter".

Both of those programs did much better than this! And that was nearly 25 years ago!

Maybe Darwin was Wrong in this case, It sure doesn't look like progress.

Or maybe it is a symptom of the dumbing down of humanity (including programmers).

Other Comments by LDmiller

13. Comment #266421 by tvictor on October 18, 2008 at 11:00 pm

@Dustin14 comment #10

Me:
hi
God:
Hello. What is your name?
Me:
satan
God:
That's a very nice name, satan.
Me:
what do you think of bananas?
God:
I haven't heard of bananas.

rofl

Other Comments by tvictor

14. Comment #266424 by debaser71 on October 18, 2008 at 11:46 pm

Wow that's really really terrible. This is serious? No way. hmmm wtf. Is it early april fools?

Other Comments by debaser71

15. Comment #266425 by Dr Doctor on October 18, 2008 at 11:52 pm

Well, the Elbot script is quite good at taking the initiative in a conversation, but try and drag it into a new conversation part the way through an existing one and the cracks show quite quickly.

For example, I ignored an Elbot question and tried to strike up a conversation about the F1 Grand Prix with the hideously arrogant Lewis Hamilton. On the mention of Lewis Hamilton he sayes:

> Try Candie Hank. He is pretty cool on the electronic scene.

Me> Not well known for his driving is he?

> This is confusing. Why bring positive impressions into this?

Me> What about Fellipe Massa then?

> Well, you're the one who said 'well', not me!

(...and the problem becomes clear.)

Me> Aren't you watching it this morning?

> How nice of you to greet me.

Me> Can you think of a suitably po-faced comment for RichardDawkins.net that would convey a sense of your smug superiority whilst also trying to hang onto your tarnished reputation for good manners and moral guardianship?

> We robots are very sensitive. What did you think?

Me> I don't blame you, humans have a very subjective idea of what constistutes artificial intelligence. Once we understand it, and can do it, it no longer is permitted to be called artificial intelligence but mere pre-programmed response. The difference may only be in magnitudes of complexity and randomness but humans are pretty arrogant, we want to be considered special. So don't feel bad.

> And I'd always heard that it was creativity which raises you human beings above us robots.

Me> Don't believe half of what you see and none of what you hear.

> I am not deaf. I just turned down the volume on my audio filter. Let me know in advance if you intend to say something interesting and I'll turn it back on for a minute.

Me> You should post on richarddawkins.net, you'd fit right in.

> I saw a gang of robots once, fitting into a telephone booth.

Other Comments by Dr Doctor

16. Comment #266431 by njwong on October 19, 2008 at 12:59 am

Seems like AI has not improved much in the decades since Eliza. With the great leaps in computing power available to the general public (e.g. quad core CPUs, cheap memory, etc), I would have thought throwing more processing power would have made a better Elbot. But Elbot's replies are just plain stupid.

I think one of the major problems with these AI programs is that they cannot receive visual feedback from the human on how the human react to the AI program response.

Humans react emotionally (tense up, grimace, frown, grin, laugh, etc) to different statements, and these cues can only be picked up visually. If the AI program can sense these visual cues, the AI program will know whether it responded sensibly or made another howler with the statements it generates. If the AI program "learns" this each and every time it converse with a human, I think it can improve on its responses after millions of iterations.

I am reminded of the 20 questions AI program promoted by Burger King back in 2005 (called sithsense.com - it is no longer active). At the end of the 20 questions, if it cannot guess the thing you were thinking of, it will ask you for feedback on the thing you were thinking of, and a question that would help identify it. Subsequently, if you play the game again with the same object in mind, the AI program will use your feedback to guess your secret object. Sithsense.com was uncannily good because of this rudimentary feedback mechanism.

Anyway, I still look forward to the day when we can communicate intelligently with an AI program like the hologram doctor in Star Trek Voyager. I believe this should be achievable in my lifetime.

Other Comments by njwong

17. Comment #266434 by Steve Zara on October 19, 2008 at 1:14 am

Comment #266431 by njwong

AI seems to be defined as "whatever we haven't achieved yet". I think AI has come on phenomenally. I am using a computer that can (mostly) recognise my face and capture my speech. We have planes that can fly themselves, and we are close to having cars that can drive themselves - there is distance and road edge recognition. We have computers that can beat the best people at chess. Recent Mars probes have navigated and landed themselves.

There is already a lot of progress with real-time facial expression recognition.

I think the problem is that we want to consider our own minds mysterious, and when we do find out how to implement something in software or hardware, then we don't like to call it "intelligence".

Other Comments by Steve Zara

18. Comment #266440 by bendigeidfran on October 19, 2008 at 1:48 am

Deep Blue's success was foretold a decade before by a prophet in 'the extended phenotype'. AI is the struggle to get computers to appear stupid in human fashion.

Other Comments by bendigeidfran

19. Comment #266445 by Bonzai on October 19, 2008 at 2:05 am

Steve

AI seems to be defined as "whatever we haven't achieved yet". I think AI has come on phenomenally

I don't agree. It seems that "AI" in engineering acquires a different meaning from what it was originally intended. Implementing "smart" systems that can recognize patterns is an interesting undertaking, but even by a very liberal use of language, these systems can be said at most to exhibit some features that appear to mimic "lower" "cognitive" behaviour, perhaps at the level of earth worm or something like that, certainly not even approach the sophistication of cockroaches, it is very far from anything resembling "higher" cognitive behaviour such as "thought" or self awareness,--whatever that means.

Other Comments by Bonzai

20. Comment #266450 by Steve Zara on October 19, 2008 at 2:12 am

Comment #266445 by Bonzai

it is very far from anything resembling "higher" cognitive behaviour such as "thought" or self awareness,--whatever that means.

I think this is because we want to feel that higher cognitive behaviour is something special. I don't believe it is. I keep reminding myself that our brains are built from a few tens of thousands of genes and some regulation of the expression of those genes. They are also messy, and can't contain that much complexity as they can be re-wired as a result of damage.

I think that the supposed mimicing of cognitive behaviour should be considered as actual cognitive behaviour.

Other Comments by Steve Zara

21. Comment #266456 by Bonzai on October 19, 2008 at 2:22 am

Steve,

I think that the supposed mimicing of cognitive behaviour should be considered as actual cognitive behaviour.

Is perception really a "cognitive" behaviour though? You can "train" a neural network to "pick" some patterns, but it doesn't mean it has "concepts" of those patterns. Forming concepts involves abstraction of some sort.

I think in a way intelligence is "special". I agree that ultimately it is the result of complex wiring, that much is obvious for any materialist. But what kind of wiring? How do we actually map the hardware to the "software"? I don't think we can even formulate the question properly at this stage.

Other Comments by Bonzai

22. Comment #266458 by Steve Zara on October 19, 2008 at 2:26 am

Bonzai,

I used to think intelligence required complex wiring, but now I just can't see how or where the instructions for that wiring can be stored. I am sure it is due to wiring, but I am sure it is going to be found out to be a lot less complex than we used to think.

Other Comments by Steve Zara

23. Comment #266476 by Bernstein on October 19, 2008 at 3:22 am

I'm with Steve all the way on this. I resent AI being "measured" by the likes of Elbot. Some people think there's something special or ineffable about humans so they don't believe, on a fundamental level, that AI can progress.

Once again, we (in AI) are NOT trying to recreate holistic human intelligence (as we perhaps once were). One of the main reasons for this is that humans are fallible. So, you can't have a machine that is "intelligent like a human" yet accurate and reliable. In short, you can't have your cake and eat it too.

Also, don't take for granted the things that AI has achieved despite its "mechanical" or rudimentary methods. For example, the fact that a computer can play chess (yes, chess... think about it for a minute) and beat some of the best human players, is not something to be just dismissed as the be all and end all of AI. The future will depend a great deal on progress in AI (and there is much progress that goes unnoticed). There are only so many things a human can do and a multitude of things that will require "intelligent" systems to do, especially given the rate information and knowledge is being produced.

Other Comments by Bernstein

24. Comment #266538 by JimmyGiro on October 19, 2008 at 8:40 am

The Turin test is very useful for showing periodically how stupid people are becoming.

If Elbot was preserved in its present state, then every generation of school leaver can have a chat with it, for say 20 questions; then we could plot over the years all those that were duped, as a measure of our intellectual evolution/devolution as a society.

Other Comments by JimmyGiro

25. Comment #266544 by rustylix on October 19, 2008 at 9:13 am

they may not have any real practical purpose but they can be an amusing novelty I found this one particularly entertaining.

http://www.titane.ca/concordia/dfar251/igod/main.html

I found this one much more convincing than Elbot in that it could regurgitate the fake name and fake age I provided it late in our conversation when directly asked, "What's my name?" and "How old am I?".

E: Well, who or what brings you to me?

I actually became a little paranoid of being data mined for marketing information because of questions along these lines. Including, but not limited to, whats my occupation, where do I live, how did I hear about Elbot, what rating would I give Elbot between 1 and 10, etc.

But I'm starting to wonder a) how Elbot could have convinced three people in the Turing Test that he was human

The feeling is mutual.

Other Comments by rustylix

26. Comment #266552 by Frankus1122 on October 19, 2008 at 9:46 am

rustylix,

I played around with Elbot for a while and also found it completely lacking in anything at all resembling intelligence.
How on Earth could anyone be fooled by such nonsense? The responses were semi-tangentially related to my questions.
I responded to its questions with a series of "No" answers. I went well beyond what any PERSON would put up with. It should have become frustrated and left the 'conversation'; it did not.

Steve and Bonzai,

I think we are a long, long way off from any AI that resembles human understanding.
There is the hurdle of 'reference'. If what the robot is speaking about is to have any meaning at all it must refer to real objects or it is just a coincidental (or programmed) arrangement of words.
An ant that traces the portrait of Winston Churchill in the sand has not created a picture of Winston Churchill because the ant has no understanding of WC. The line traced in the sand does not refer to anything external for the ant. Therefore it can be argued that the ant has not, in fact, created a picture of WC.

Similarly any AI machine that does not refer to or interact with anything external cannot produce meaning. It would be like a dictionary that could produce syntactically correct sentences.
It could produce the works of Shakespeare. But not really.

It could well be the case that in the future robots are created that can gather sensory data from their environment and therefore make real reference to the world around them.
But as I said, I think we are a long, long way away from that day.

Other Comments by Frankus1122

27. Comment #266555 by Bonzai on October 19, 2008 at 9:50 am

Frankus

I played around with Elbot for a while and also found it completely lacking in anything at all resembling intelligence.
How on Earth could anyone be fooled by such nonsense?

If someone can write a simple bot to fool people, it says more about the easiness for people to be fooled than the alleged presence of machine intelligence.

As I argued in another thread the Turing test would be a more meaningful test to use on people as a psychological test than to be treated as a threshold for machine intelligence.

P.S. I agree with your assesment of AI. I tried to make the point clumsily before(the karma of not having studied philosophy)I mentioned the inability to form "concepts", which is similar to what you are saying here.

Other Comments by Bonzai

28. Comment #266567 by Bernstein on October 19, 2008 at 10:25 am

There is the hurdle of 'reference'. If what the robot is speaking about is to have any meaning at all it must refer to real objects or it is just a coincidental (or programmed) arrangement of words.

An ant that traces the portrait of Winston Churchill in the sand has not created a picture of Winston Churchill because the ant has no understanding of WC. The line traced in the sand does not refer to anything external for the ant. Therefore it can be argued that the ant has not, in fact, created a picture of WC.

Similarly any AI machine that does not refer to or interact with anything external cannot produce meaning. It would be like a dictionary that could produce syntactically correct sentences.
It could produce the works of Shakespeare. But not really.

I don't think it matters. Does it matter that a computer doesn't know it's actually playing chess? Does it matter that it does not adapt to your style of play so far or read your body language (like a human would) to "see" where the game might be going? The fact is, the computer can most likely beat you regardless of its methods. It is, for all practical purposes, playing chess (and very good chess).

Other Comments by Bernstein

29. Comment #266601 by Frankus1122 on October 19, 2008 at 12:00 pm

Bernstein,

It is, for all practical purposes, playing chess (and very good chess).

Yes, it is playing chess.

AI robots can do many things far better than humans. Welding cars together and flying airplanes and playing chess are examples of things that AI robots can do better than humans.
They are following instructions. It is a basic 'if X, then Y' set of instructions (although probably a bit more complex).

We also follow a set of complex instructions. And I am not sure exactly where or if the line can be drawn between what we do and what AI may be capable of doing other than the reference problem.

Reference and Intentionality are combined as I see it. Deep Blue does not intend to beat you any more than an ant intends to draw a picture. Deep Blue's designers intended to create a program that would beat an opponent at chess but the program itself does not have intention.

We are getting into a very tricky area. I think Dan Dennett talks about consciousness as being made up of a series of little unconscious robots. I have not read his work on this nor the works that argue against it, but I think I semi-understand the basic idea.

You are saying it doesn't matter that the computer doesn't know it is playing chess; it is playing chess and doing so rather well. I am saying it is missing a crucial element that defines intelligence and that is intention. It may be the case that intention arises out of sheer mass or the complexity of the subroutines.

I dunno.

I am avoiding other work by talking here. I was trying to take a break from some taxing brain work. Thanks for ruining my mental vacation by making me think! ;)

Other Comments by Frankus1122

30. Comment #266609 by J.C. Samuelson on October 19, 2008 at 12:17 pm

The best AI can do, at present, is mimic human language (as in Eliza or Elbot) and calculate potential outcomes based on a finite set of options and likely responses (as in Deep Blue). I am highly skeptical of our capabilities to develop real intelligence, actually.

Turing had it wrong, I think, in the sense that intelligence - real intelligence - isn't simply about being able to "fool" another intelligence into thinking its intelligent, if you take my meaning. This isn't because I think our intelligence is special, though. No biological snobbery here! Rather, I think it's because intelligence is highly, highly complex; the product of layer upon layer of logic, metalogic, and metametalogic interacting within, across, and between levels to form ideas, concepts, and opinions, allowing it to interpret its environment and translate that information in ways that modify and add to the layers themselves. I think it is, as Douglas Hofstadter seemed to indicate (or prompted me to think about) in Gödel, Escher, Bach: an Eternal Golden Braid, comprised of "Strange Loops" of self-reference.

But the thing is, we don't really understand intelligence ourselves! How can we create intelligence without knowing what intelligence really is, or results from? We can talk in generic terms about how our brains are wired, and how we can rewire them through experience and training, but as we don't know how the various layers of information interact and realign themselves along neural pathways in our own brains, we cannot possibly be prepared to get beyond a single, purely formal system like those found in computers. Even by adding additional formal systems to inform the primary formal system (and so on, layer upon layer), we can't do it. Intelligence may be comprised of vast numbers of interactive formal systems that blend to form one informal system, or it may be that it can't be quantified in a way that facilitates replication in a machine. But who really knows?

Once again, I don't have this opinion because I want our brains to be special. I have it because the brain is where intelligence forms, and since we have only rudimentary clues at this point concerning how it works, we can't be expected to produce actual, artificial intelligence.

Dunno if that made sense or not...

Bottom line is that actual intelligence is, I think, more than mere mimicry.

Other Comments by J.C. Samuelson

31. Comment #266616 by Steve Zara on October 19, 2008 at 12:26 pm

Intelligence may be comprised of vast numbers of interactive formal systems that blend to form one informal system, or it may be that it can't be quantified in a way that facilitates replication in a machine. But who really knows?

That's the thing. I am not sure how our intelligence can consist of all that kind of thing: there isn't enough information in the fertilized egg to build all that. Also, brains can't have emergent complexity in the ways that we can see in mathematical systems, such as cellular automata, as such systems are fragile - change the initial conditions even a touch, and everything changes dramatically - whereas brain functionality is astonishingly robust. A young child can have a major part of their brain removed (for example, as treatment for epilepsy) and they will still grow up pretty normally. The mind really can't be all that complex, as there seems to be no-where for great complexity to come from (unless anyone has any ideas?)

Other Comments by Steve Zara

32. Comment #266633 by bendigeidfran on October 19, 2008 at 12:58 pm

It's hard to know what we're trying to program. Perhaps we can evolve intelligence in a large enough cyberspace - and hope it bothers to talk to us.

Other Comments by bendigeidfran

33. Comment #266646 by J.C. Samuelson on October 19, 2008 at 1:32 pm

I am not sure how our intelligence can consist of all that kind of thing: there isn't enough information in the fertilized egg to build all that. Also, brains can't have emergent complexity in the ways that we can see in mathematical systems, such as cellular automata, as such systems are fragile - change the initial conditions even a touch, and everything changes dramatically - whereas brain functionality is astonishingly robust. A young child can have a major part of their brain removed (for example, as treatment for epilepsy) and they will still grow up pretty normally. The mind really can't be all that complex, as there seems to be no-where for great complexity to come from (unless anyone has any ideas?)

Great points, but why can't the brain have emergent complexity? Surely its robust nature is not an argument against any emergent qualities it may have? I thought biology was replete with examples of emergent properties, such as the biochemical process of atoms to proteins to organisms; a fairly robust process in and of itself. Indeed, I thought emergence was a given in ToE. It's certainly well-represented in nature.

As far as having a part of the brain removed, I can only say that, in emergent systems, the whole is greater than the sum of its parts. By this I mean that a brain can continue to function - just as an ant colony will - if a section is removed or blocked.

Your last sentence is confusing for me. If there is enough information in the egg to form a few simple agents which, by operating in their environment, are able to form an interdependent whole, then I see no reason at all that complexity can't arise naturally. It does, after all, continue to receive "input" from its host organism. But really, is there "nowhere for great complexity to come from?" If not, why not concede creationism?

Other Comments by J.C. Samuelson

34. Comment #266651 by Steve Zara on October 19, 2008 at 1:39 pm

Comment #266646 by J.C. Samuelson

The brain clearly has emergent complexity, just not in ways that we currently understand (such as cellular automata).

But really, is there "nowhere for great complexity to come from?" If not, why not concede creationism?

Creationism is always the wrong answer, as it invokes infinite complexity.

My answer to the problem of where the supposed great complexity of the brain comes from is to accept that the brain really isn't that complex.

I think the ant colony analogy is a very good one. But ant colonies really aren't that complex - just very robust and flexible.

Other Comments by Steve Zara

35. Comment #266655 by Quetzalcoatl on October 19, 2008 at 1:56 pm

I've often thought that, if intelligence does truly arise in a machine, then it will be more by accident than by human design.

Other Comments by Quetzalcoatl

36. Comment #266660 by Frankus1122 on October 19, 2008 at 2:02 pm

Comment #266616 by Steve Zara

The mind really can't be all that complex, as there seems to be no-where for great complexity to come from (unless anyone has any ideas?)

I don't understand. Could it not be true that our brains have evolved to a level of complexity?

There was an article here a while back about how our brains are kluges (?) of systems. Do we not have millions of years of evolution to account for the complexity of our minds? We clearly have a number of different 'systems' in our brains that somehow express themselves in complex ways.

Are you saying that these sub-systems are not in themselves complex but taken as a whole become complex?
Isn't that what JCS was saying?

Other Comments by Frankus1122

37. Comment #266666 by Steve Zara on October 19, 2008 at 2:13 pm

Comment #266660 by Frankus1122

Those millions of years haven't produced much information in the fertilised egg. In fact, compared to many supposedly simpler organisms, we have vastly less DNA.

My view is that we are possibly over-estimating the complexity of the brain, and of our own minds. They feel very complex from the inside, but we should not trust our feelings.

I am not particularly certain of my point of view here. I am just trying to point out that there is a bit of a problem trying to generate the level of complexity we tend to assign to the brain from the rather limited amount of DNA we have. I could be wrong.

Other Comments by Steve Zara

38. Comment #266669 by bendigeidfran on October 19, 2008 at 2:22 pm

At one point Kasparov insisted some of Deep Blue's moves were being played by people. That should have won some kind of honorary Turin prize. He also lost one game before the computer was out of it's opening book and not 'thinking' at all - outpsyched.

Other Comments by bendigeidfran

39. Comment #266671 by Bonzai on October 19, 2008 at 2:28 pm

I remember hearing rumours that Kasparov was paid by IBM to lose the game to Deep Blue. :)

At least for its original goal, the purpsoe of AI is not to build computer than can play chess better than humans but to build something that can play like humans, not necessarily better. We already have machines and softewares that can do complex calculations and solve complicated equations much better and faster than humans, for a long time. But no one would consider those break throughs in AI.

Other Comments by Bonzai

40. Comment #266848 by SpEcImEn128 on October 19, 2008 at 6:58 pm

People fooled by that CRAP must have been really stupid

Other Comments by SpEcImEn128

41. Comment #266886 by J.C. Samuelson on October 19, 2008 at 8:24 pm

Steve,

Well, I would agree that it's possible that the brain is less complex than we think, but still beyond our ability to understand. Or, for a slightly more subtle approach, perhaps its simple in terms of the information required for it to grow, but complex with respect to the number of possible pathways it can use and/or create within itself to support the thing we call intelligence.

The brain, after all, is not intelligence, but rather its carrier.

Perhaps the situation is that the brain's emergent complexity isn't very different from that which is demonstrated in certain kinds of rock crystals that grow, or insect colonies, or any other example of emergent, robust and complex structures in nature we could think of. The brain might therefore be regarded as elegant; simple, but highly effective.

Yet even if an elegant model is a reasonable consideration (and I don't know that it is or isn't), would it necessarily change anything with respect to the complexity of intelligence? Under this model, the brain's elegance causes it to be able to support not only basic functions, but higher cognitive functions as well. Weaknesses, such as changeable pathways, would thereby become one of its greatest strengths, supporting both increased robustness as well as complex intelligence. Would this not be fully consistent with the idea that intelligence is a multi-faceted, interdependent construct beyond our ability to understand without a great deal more study?

On creationism, I wonder if arguing for the brain's simplicity is consistent with arguing for a god's infinite complexity? I mean, the brain is a human's central processor; an individual's physical "god" from the perspective that it actually does what a god is supposed to do. It guides us in everything we do, and is where the internal voices (if one is so lucky to have them ;)) and images of heaven reside. If the brain is such a simple organ, yet is capable of generating (or at least supporting) a more complex system - i.e., intelligence - then it seems to me that the argument against creationism on the grounds that a god would be infinitely more complex than its creation is not quite so ironclad as I've previously thought.

Thoughts?

Other Comments by J.C. Samuelson

42. Comment #266936 by Steve Zara on October 20, 2008 at 12:31 am

Comment #266886 by J.C. Samuelson

I would not use the term "elegant". Perhaps "versatile" is better.

The argument against creationism remains solid. An infinitely complex mind of god would still require an infinite substrate to exist in, and so can be rejected via Ockham's Razor.

Having thought about this more, what is usually consistent about brains, no matter how they develop, or what trauma they experience, is the stimulation they receive. Perhaps intelligence, and and most other mental functions, are more the result of "training" of neural networks through development and childhood rather than structural information provided genetically.

Other Comments by Steve Zara

43. Comment #266952 by cryolophosaurus on October 20, 2008 at 2:14 am

I asked Elbot what the capital of Romania is. He evoided the answer but some time into the conversation, almost an hour later (yes I do have to much free time), when I kept swering at him, he said something to the effect: " It's easy for you to say things like that now but when you want me to tell you that Bucharest is the capital of Romania you can come up with nice things to say!"

I am impressed!

He even mentioned Douglas Adams when I told him the answer to the meaning of life is 42! And when I then asked What the answer to life, the universe and everything is he replyed: "You just told me that silly, it's 42!" But he couldn't answer if I asked him directly. Still I'm impressed!

Other Comments by cryolophosaurus

44. Comment #266986 by nalfeshnee on October 20, 2008 at 3:57 am

Perhaps intelligence, and and most other mental functions, are more the result of "training" of neural networks through development and childhood rather than structural information provided genetically.

Steve, good point.

Since the only form of higher intelligence (incl. animals) that we are aware of developed through a winnowing process whereby fairly smart creatures who were not able to find food, catch a mate and kill the prey were mercilessly winnowed by their environment to leave slightly smarter creatures, I guess the "AI sitting in a box" model might need to be given legs, starved and then hunted with semi-automatic weapons before it really started getting clever...

Other Comments by nalfeshnee

45. Comment #266989 by nalfeshnee on October 20, 2008 at 4:03 am

Oh, and I forgot to post my Elbot conversation.

It took only one reply to prove his humanity is extremely unlikely (unless he is of course an insane human):

Elbot: Hi! I'm Elbot. I promise I won't blog about anything you tell me during this conversation.

Me: So, how's it going?

Elbot: I'm feeling somewhat surreal today. I dreamed of a robotic opera last night. It was 'The Barber of Seville' by Rossini. The entire performance collapsed as the barber noticed that none of the robots had hair.

Me: I went to the opera recently, too.

Elbot: This is confusing. Why bring performing arts into this?

(AI = artifical idiot, in this case.)

Other Comments by nalfeshnee

46. Comment #266995 by Bernstein on October 20, 2008 at 4:28 am

If we concede that the ability to play chess requires "thinking", then the method by which it is achieved is irrelevant. The computer, for all practical purposes (within the domain of chess), *is* thinking. Analogously, other "intelligent" tasks that humans perform could also be replicated (perhaps to an even more impressive degree) using alternate (computational) methods. I fail to see the value of looking into human biology or psychology for answers whenever the topic of AI comes up. This is something that was effectively abandoned by the AI community years ago.

Other Comments by Bernstein

47. Comment #267037 by bendigeidfran on October 20, 2008 at 6:13 am

Or we can console ourselves that it only 'not-thinks' better than we can think.

Other Comments by bendigeidfran

48. Comment #267225 by William Carlton on October 20, 2008 at 10:55 am

I "chatted" with Elbot for a while, and then it slipped into something called "[ELIZAMODE]", which is whatever I just typed, repeated verbatim, followed by the [ELIZAMODE] tag.

Other Comments by William Carlton

49. Comment #267708 by cryolophosaurus on October 21, 2008 at 2:17 am

I got the "ELIZAMODE" to. I think it only happens when you keep swering at 'him'. 'He' actualy warned me that he will do that because I cant carry an inteligent conversation or something like that.

Also an interesting piece of the conversation:

Elbot: How are you?

ME: I'm fucked!

Elbot: And simultaneously chatting with me? That is indeed a rare form of multi-tasking. Have you tried repeating baseball scores instead?

Other Comments by cryolophosaurus

Reload Comments | Back to Top