Slashdot Mirror
MIT AI Acts Childish on Purpose
garibald gave us the link to an article in the electronic Telegraph about researchers at MIT who have built an interactive AI robot called "Kismet" that is as cute as any
George Lucas character, and is supposed to function on the emotional level of a
human two-year-old. The cuteness is not gratuitous. As the article makes clear,
there is a sound, scientific reason for it. (For pictures, and more technical depth than the Telegraph story, you may want to go directly to Kismet's Web Page.)
When Kismet was vigorously over-stimulated with a stuffed toy, it fell asleep. Kismet is obviously narcaleptic.
I can't wait until this thing serves me in my local burger restaurant and tells me whether I 'have a problem with that'.
-kojak
Did someone *cough* say head? *cough*
Not every meaningful judgement is based on completely known factors. I think there's certainly a lot of room to mistakenly judge a new development like this one, based on a context that really doesn't fit. Because people are unused to emotions from non-biological things, there'll be room to forget what's actually what. People will be reluctant to cause 'pain' to things that have no real capacity for pain -- or ignore the pain of something that does indeed have that capacity.
But that doesn't mean that some concern aren't valid. This has the potential, down the line, to produce the ultimate in "junk-food relationships": satisfying the hunger, without feeding the actual need. If you don't think a lot of people will choose no-effort emotional interaction with a computer, over the hard sustained work of building a relationship with another human -- just look around you, and see how many people are wearing the marks of Haagen-Dazs rather than balanced meals.
...But we need some fake skin on it to make it look like more human.
Yeah, idiot, shouldn't you have linux a few more
times in your signature?
AI at MIT is a joke since Brooks lost DARPA funding and now has to appeal to the mass media with cute robots like this and Cog. At least he is not down at the level of BEAM robotics. I think Brooks might still be able to regain his credibility.
Be sure to visit the Dead Kennedys Shrine.
The pictures of her with the robot caught my eye at first, but when I went to her home page I was disappointed. She's not hot. Must have been the lighting and camera angle or something. Don't you hate it when you see a girl from behind and you think she's beautiful, only to have her turn around and reveal that she's ugly? Pisses me off dammit.
I think Slashdot needs a pr0n section.
"It's unfathomable how moralists can get away with claiming that their opposition forces their viewpoint of free inquiry onto others as some sort of oppressive yoke."
Are you suggesting it's immoral for them to make moral judgments in this context?
Face detection isn't that difficult. There're well known algorithms for face detection that work with varying degrees of success. In fact it's a project for some machine vision classes offered at my university. I've done similar work myself (Hand detection actually).
I don't think it actually detects a toy. It seems to just detect motion, which is quite trivial.
Apparently, MS has been heavy on AI for a long time. Windows acts infantile, burps, farts, and cries all the time, to get your attention. That which keeps your attention will be hard to ignore, remove, or forget. The ultimate marketing and technician training tool: the O.S. that adjusts its crashing to your skill level!!
Visit http://www.os2hq.com/ for more "Warped Perspectives."
Neil W. Van Dyke said:
"I don't think you realize that a thread of comments on someone's appearance is not appropriate in this context"
What do you expect from someone with the last name of "dyke"...
"...animals that have emotions but little if any concious experience of them."
/. is focused on computers, the 'AI interpretation' of psychology and philosophy (Dan Dennett as an example of the latter) tends to be accepted much more easily than elsewhere. While popular, it is certainly not representative of the majority, which has recently (the last few decades) begun turning away from computation as a paradigm of the mind (and reality for that matter).)
I'm not sure sure that 'unconscious emotion' is coherent. Obviously there can be unconscious physiolgical effects similar to those accompanying emotions, but 'emotion' presupposes experience (though maybe not complete consciousness). However, the physiology should not be confused with the emotions themselves.
"Remember that the evolutionary/adaptive power of an emotion is that it makes you react well in a given situation (e.g. fear -> fight/flight, etc)."
Again, I question this. First off, fear may not be an emotion, though it may be a component of some emotions. Secondly, many emotions *don't* make us react well in many situations. In many cases they actually degrade performance. Anger is an excellent example of this--specifically, the tendancy of people to do stupid things when they are angry, things which sometimes degrade their chances of surviving. And Kismet has not 'evolved' anything--he has been engineered to act in certain ways in response to stimuli.
Your conclusion that Kismet is experiencing emotions is behaviorist--it appears to have emotions, thus it does. This position has been largely abandoned outside AI, where it retained primarily because it is the most AI-friendly option available. However, convenient as it may be, it is still wrong. (Actually, it isn't even behaviorist; you have no reason to believe that it is even 'behaving', yet you accept that as the reality of the situation. For all we know it could be a puppet. I think some skepticism is warranted.)
(A personal gripe: since
-jcl
For good-looking robotics: Check out Björk's new video love...something. They are even frenchkissing each other, how's that for emotion!
we had a demo of a backprojected head on a curved dome (small semi-opaque white plastic) a while back at tomorrows world in birmingham..would be kewl if they could do that..the face when backprojected is a lot more lifelike and uses less power too..
coders all look like dorks. besides we arent interested in checking out torvalds vital statistics...we would if he was female and looked like cindy crawford tho.
I don't think you've stopped to consider some of the large differences between seals and cows. Cows as we know them are very much bio-engineered creatures. We've spent the last ten thousand or so years breeding them to be docile and tasty. The creatures we've ended up with wouldn't even be able to survive without human support. One of the problems with the idea of abandoning the eating of cows is that it would probably mean the near extinction of the species.
On the other hand, seals are hunted, not raised for slaughter. They live without human support. The only real change humans make to their populations is to reduce them. Maybe they would let people raise seals for slaughter like they do with cows. That's what they do with Caymans for example. Of course, the problem with that is that most people are just out to make a fast buck, and it's cheaper for them to have a small farm as a front and hunt wild caymans. The same thing could be assured to happen with seals.
Sure, cuteness counts, but a lot of non-cute species are protected as well. Would you hug a Komodo Dragon? There are protected fish as well. In fact, almost all fish are protected. There are many laws (often violated as a matter of course) about what volume and what size you can catch for all sorts of fish. These laws exist for a good reason. Animals can be hunted to extinction by human beings. It's happened before, it's happening right now, and it may well continue to happen.
So, just to sum up, what I'm saying is that the cow supply is quickly replenished and quite large, while the seal supply is not quickly replenished and is smaller. We'd probably run out of seals, but the only way we're going to run out of cows is if everyone suddenly decides to eschew beef, or if they get wiped out by a plague or some such.
Sunday's NY Times talks about S.K.'s film project
about an A.I. in the form of a child.
Harvey Keitel, Kirk Douglas, Farrah Fawcett; Saturn 3.
Can Cyc tell you about anything outside the realms of mathematics and text? Can Cyc tell you what an apple is like besides basic facts? Will it ever be able to?
The Cog lab researchers have some interesting ideas about embodied intelligence. They believe the complete lack of embodied experience is hampering Cyc and similar projects. Robots like Cog include many small subsystems working together in an attempt to copy animals. Cog cannot accomplish much yet. They think by following an approach rooted in developmental psychology, it will be better someday.
Maybe they know what they're doing with their approach, or maybe not. Direct comparisions with projects such as Cyc are difficult, however.
RFC, eh? - get some sleep.
j/k
Seriously, I don't know what is more unsettling:
1. expending effort and resources to develop emotive responses in mechanical devices, or
2. the image of programmers cooing at a robot that looks like a gremlin - "Daddy loves Kismet, does Kismet love daddy?"
Saw the video, doesn't look too impressive.
Seems that the algorithm is basically: See moving object, get excited (or some other expression) but look disgusted or fall asleep if it goes on for too long. Respond a little differently when human face is detected. They blather on about 'learning ' but I don't see any evidence of that either.
Looks like a lot of hype for precious little substance.
You're thinking on the same track they are, except they want skin to make it act, not just look, more human. Dr. Anne Foerst is a post-doc at the lab working on skin. If you fish around the MIT AI site, you may find a bit about the work. Cog, Kismet's "older brother," has some fake skin on his stomach. Judging from what I've read, developing skin for the robots has proved very difficult.
I saw Dr. Foerst speak, and she is amazing. She has a joint appointment with MIT AI and Harvard Divinity. Her work on skin has a theological basis from Genesis informed by modern studies of human development. Her exegesis of Genesis, based on the work of a 14th century rabbi, was something totally new to me. She is both a theologian and a computer scientist with formal education in both fields. This article includes a brief background.
Cynthia could make a porn video and make people pay for it to raise money. I can see why that robot gets mad when she leaves :)
i think they couldn't get the thing smarter than a talking ape so they settled on a 2 year old...
Large print giveth, and the small print taketh away
So you're saying by expressing signals that appear to humans as reflective of internal emotions, Kismet is experiencing emotions?
Spoofing is a term applied when a process transmits information to another process reflecting some status about itself that is, in fact, not reflective of the true state. For example, IP Spoofing transmits false IP Source Header information. Samba does not spoof SMB, though--it really speaks it.
It's a bit of a grey line.
In humans, it turns out that the mere expression of an emotional state is enough to generate some aspects of that state internally. On a personal note, this is why I spent Senior Year of High School smiling like a hyena. Acting happy created happiness--strange but true. But Kismet doesn't act out externally as an expression of internal emotional states, or vice versa--there *are* no emotions to reflect, not in the human sense of the word. It's a spoof.
But try telling that to someone who is looking at a cute sad little doll that asks not to have its power disconnected.
What was it that the Star Trek authors argued? That sentience required consciousness, self-awareness, and intelligence? Good values. Nice values. But the ability to effectively convey emotions (and perhaps complex concepts) effectively is probably required for any creature to be recognized as sentient. Call it a bug in the Human OS if you need to. But it's true.
--Dan
Once you pull the pin, Mr. Grenade is no longer your friend.
Wow.
Sometimes I think computer people Just Don't Get It.
Then I see this.
Now, alot of us geeks might whine and moan. "Oh, it's so cute. Die Jar Jar Die!" "The future of technology lies with...Terminator Furby." Whatever.
Outside of the cold, sterile world of gigabit routers and tab completion, a convincingly emotive device has been developed--artificially.
The importance of this is not to be underestimated. Large chunks of government policy are designed to protect animals with emotive properties. As a classic comedy routine went, "Sometimes I think the animals just all got in a line. 'What are you?' 'I'm a seal.' 'You're cute, honk that horn, we'll make sure nobody beats you with a club. Now what are you?' 'Cow. Moo.' 'GET ON THE TRUCK! You're a baseball glove.'"
Pet Rocks were quite the subversive satire on this maternal instinct.
Tamagotchi began the trend, and Furbies proved the consumer attraction, but I think it's the progeny of Kismet that will whip up quite a media frenzy. Who wouldn't think that the media has been waiting to report that a machine built by man doesn't want to die? It's a pent up desire; one that will be released at the first credible moment.
And along comes Kismet.
The key to Kismet really is that it spoofs emotion. Think about it for a second. The MIT guys "sniffed" humans with cameras, copied the protocol stream down to the transitions from one emotive signal to another, and (most importantly) parsed enough of the incoming emotistream to generate a seemingly interactive experience.
Shades of Eliza? Obviously. Eliza spoofed "Rogerian Psychology", where a person does nothing but ask the minimum amount of questions to keep you talking. Eliza took some aspect of human psychology and looped it on itself to create meaning to the user at the least possible computational cost.
Give Eliza and Kismet a love child equipped text to emotive speech and speech to text, and toss in a degree of anger, rage, and fear if the robot "believes" 1) it is to be deactivated(put a video sensor near the off switch) or 2) it is ignored for excessive periods of time, and a non-zero part of the population will believe it alive and as worth protecting as a cute baby seal.
Is it just me, or is it scary how much Hackerthink(spoof, parse, etc.) fits so many different situations? If you can talk to a single person in biotech for more than twenty mintues about their job and not realize they're utter hackers, you aren't paying attention. The same applies to psychologists. How many psych papers read like security bulletin? ("God refuses to patch. F1zRR has released Prozac 1.0 to compensate.")
If Kismet ever goes mainstream, the psychologists are going to have a field day. The technological revolution eliminated the need for menial workers. Kismet, scarily enough, could make shallow friendships far more awkward. "We don't do much more than talk like Kismet."
All this will go on until the media decides to flip public opinion around on its ear(thus making everybody tune in) saying "Are we nuts? This Kismet thing is NOTHING compared to humans! Deux Ex Humana!"
That's just my thoughts. It's late. I'm tired. I have to get up in four hours. Joy.
Send me comments. Or don't.
Once you pull the pin, Mr. Grenade is no longer your friend.
Share and enjoy
Well, don't worry about it too much. She's married. If you scroll down to the end of the MIT page and look at the publications, you'll see her last name changes sometime between 1996 and 1998. ;-)
Hello!!! The PPro was not the main deal here. If an array of Ti DSPs isn't serious CPU throughput, I really would like to know what YOUR definition of throughput is ... :-)
/dev
"There's no secret. You just press the accelerator to the floor and keep turning left." -- Bill Vukovich
Perversion exists only in a cultural context. There is no cultural context for "where this is going", because automata that behave like humans have never existed before.
Consequently, any value judgements that head-in-the-sand universal moralists might want to make as such developments unfold are just complete unmitigated rubbish.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
> Automata that behave like humans in one way or another have existed hundreds of years
Not in human culture, they haven't. It takes a large amount of public interplay before an idea germinates and becomes a cultural meme -- most ideas just wither and die without trace except in the minds of researchers. In the area we're discussing, there is one huge big ZERO of cultural heritage to draw upon as a basis for value judgements that have any applicability.
> I don't see why you can't accept someone's current perception of automaton-human interaction to be valid, just because we have seen so little of it and it is not something society has got used to
I can't accept it because apples aren't oranges, and the fact that you haven't seen many apples but have seen lots of oranges doesn't entitle you to judge apples to be bad oranges. You have to get used to apples before you can validly say whether they are good or not.
Every single one of your arguments is laced with the desire to impose your current judgements on a new state of affairs, and there is no arguing with that, so I won't, because you won't entertain the notion that your previously acquired views might not apply.
The closed mindsets of universal moralists are legion, and it's no surprise to find their perpetual arguing against the open-minded position of making no a-priori judgements even here. If you had been in control, there would be no free Internet because it dismisses most pre-Internet values as irrelevant. Sheesh, I'm glad you're not in control.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
but the interaction of automata with humans is something that can and has to be judged within a human's value system as well.
Human can and will judge anything and everything, of course, but that doesn't mean that the judgement is valid. In this case it's not valid, because there is no cultural context in the automaton+human universe against which a system of values can be built.
Judging something out of context is only for the brain-dead, just like judging that an apple is a very poor quality orange. Not having such a context available is not a valid excuse for applying a previous inappropriate context in its place. The logical approach is to let a new context develop naturally over time, lead where it may.
But with 95% of humans happy to follow "moral leaders" like sheep, we're in for a field day of a-priori moralistic universalism, especially when nanotech begins to change the rules built up over millenia.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
Kismit reminds me less of a gremlin and more of Hector from that 1970's science-fiction movie with Charlton Heston!
---
seumas.com
I'm sorry if my post was misunderstood - it was only half-serious, and was referring to the sex bot article some time ago.
---
> The pictures of her with the robot caught my eye at first...
I don't think you realize that a thread of comments on someone's appearance is not appropriate in this context. Having to put up with stuff like that can get really old, and discourage women from participating in the "community". In case you hadn't yet realized it, Mrs. Breazeal has damned impressive techie credentials. (And she probably has better aesthetic sense than you do, as well.)
I'm afraid that you weren't thinking very clearly when you posted, but I'm sure that you will find it helpful to closely re-read and thoughtfully contemplate the posts in question (especially #53 and #60), consulting a dictionary if you encounter any unfamiliar terms.
I wonder if they have implemented a 'familiarity' algorithm in the robot,...
I visited the MIT AI Lab in early June (I took a one week course -- it was fun). I asked Cynthia this question, and the answer is no, the robot does not remember. It identifies facial features and reacts according.
I believe she said it was on the to do list.
Steve M
Like in mimicry. I came across a factoid about this bird, apparently would feed the carp in the pond. Apparently the gaping mouth of the carp was enough to trigger the 'mothering instinct' of the bird...
Oh fucking give me a break. Academia is, if you haven't noticed, a male dominated institution--meaning we weed the women out when they're younger. Moreover, who the fuck are you to tell us what is and is not appropriate discourse on slashdot? Are you honestly saying that if a Cindy Crawford lookalike came into your lab, the first thing you'd notice would be her curricula vitae (no, her OTHER vitae, you letches)? Yeah, that's a real nice touch at the end, adding in your analysis of her aesthetic sense. What happened to her vaunted credentials, Einstein? Maybe they melted away in that hypocritical moment, as you waxed poetic about how she lights up a room.
Obviously you DON'T work in academia, since any one who does know that it's ALWAYS important to package grant proposals in the best light, technical merits be damned.
This makes me worry that our first consumer-friendly robots are going to be like Marvin, the Paranoid Android.
That Cynthia is quite a babe. I'd like to have her try to make ME smile...
---
Put Hemos through English 101!
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
It reminds me more of a Mogwai robot that is naked (C3PO:why are saying that I am naked?)
"The obvious mathematical breakthrough would be development of an easy way to factor large prime numbers." Bill Gates,
As much as everyone is fawning over Kismet, Dr. Brooks as much admits that it is a hack when he lets slip that it exploits "human" programming. Much more impressive (and operating at a 5-year-old level, not 2-year-old) is CYC, which has been being built in Austin Texas for about 10 years now by Dr. Douglas Lenat. See Cyc here. It is based on Dr. Lenat's work going back to the 80's to build a computer that can learn and reason. Dr. Lenat previously had built AM and Eurisko. All /.ers should have at least some familiarity with his work.
DNA is a Turing machine. You, however, being dynamic and emergent, are not.
Personally, I'm going to arrange some Kismet faces to make a "How Do You Feel Today?" montage...
great idea, but DAMN! that thing's ugly. Looks like a cross between a furby and Short Curcuit's Johnny 5.
--
Rare Window - free your photos
Looks like pretty antiquated hardware on the PC side, I saw a PPro processor and the interface uses ISA slots. On the movs, the reactions looked lagged, coulda been the mov format though. You need some serious processing power to handle true fuzzy logic, maybe a hardware upgrade is in order? :)
I wonder if they have implemented a 'familiarity' algorithm in the robot, one that uses recognition technology to "remember" facial and structural contexts, so that Cynthia could conceivably be perceived as a 'mother figure', whereas another unfamiliar person in the field of view could possibly cause a negative reaction without any overstimulation. Maybe that's next.
I remember seeing on some cheesy PC tv show there were some programs floating about that used facial recognition with a camera to verify identity, maybe this could be incorporated in the experiment to make the robot's reactions more 'human'.
doc.
This is a sig. It is like every other sig in the world, except that it is mine, and it is different.
Yep, Cynthia could probably make a furby smile! :)
They need to check kismet with Janet Reno, to see if he's really just reacting to faces!
I believe that nature evolved emotions before consiousness, and that there have been (and almost certainly still are) animals that have emotions but little if any concious experience of them. Remember that the evolutionary/adaptive power of an emotion is that it makes you react well in a given situation (e.g. fear -> fight/flight, etc). Kismet is like a primitive animal that has evolved to have emotions that benefit him, but has not evolved the brain architecture necessary for conciousness and hence the introspective awareness of his own emotions.
Kismet is not spoofing emotion - he is exhibiting emotion through his resultant behavior. There is nothing fake about this, or fundamentally different to the way emotions work in ourselves. If a human observer anthromorphises Kismet and assumes he is experiencing anything, then that is a human shortcoming, and not a reflection on the reality of Kismet's emotions (internal state).
Ben
I disagree that emotion implies experience (i.e. concsiousnes). From www.dictionary.com : "disturbance or agitation of mind caused by a specific exciting cause and manifested by some sensible effect on the body". If an animal enounters an unusual and/or rapidly changing stimulus (say I jump out in front of you!), then this will cause adrenaline to be produced, which will have a certain effect on it's mind & body and short term behaviour. In another situation (say the sensations of grooming), endorphins might be released which will have the effect of temporarily reducing the animals aggression level, and making it more apathetic.
The other poster in this thread brought up the example of a bird feeding a gaping mouth carp due to a hard-wired respose. My Canadian fiancee saw a moose in their backyard trying to mate with a PLASTIC deer! Things like this make it hard to deny that much behaviour is simply a hard wired responce to certain stimulii. I guess you could make the distinction between emotions (biochemical) and hard-wired neural behaviours, but obviously a robot like Kismet could easily parallel either via global state that is modified by stimuli and effects his behavior (as they have implemented), or via more specific behaviours such as wide eyed and slack jawed as a response to girls like Cynthia (not yet implemented).
Incidently, when I said that emotions make animals "react well", I didn't mean that it makes them react optimally all the time, just that it makes them react in ways that are statistically good for the species (i.e. that statisticaly have survival benefit).
FWIW, my own hobbyist "artificial animal" research is based on evolution, neurology and child development, rather than AI. Cog/Kismet are more along the right direction that projects like Cyc, but nonetheless still rather misguided..
Ben
Emotions are a function of one of the oldest parts of the brain, and almost certainly are hard-wired evolutionary adaptive "knee-jerk" responses to external stimuli.
Would it be any more fake if Kismet was programmed to find "baby faces" (big eyes, rounded face, etc) cute, than humans finding Kismet cute?
This isn't spoofing. This is the way emotions work.
Ben
I suppose that means it won't be long before they train it to post to Slashdot.
In the area we're discussing, there is one huge big ZERO of cultural heritage to draw upon as a basis for value judgements that have any applicability.
I want to make sure I understand your argument: you have a problem with saying that human intercourse with a robot is perverted because we have no experience to use to make this judgement. The problem with this argument is that, although we have never had robots capable of sexually stimulating humans, we have plenty of experience with sexual acts other than human+human. Any of these acts could be called perverted, but the word "perverted" is a very subjective term.
We have invented an endless list of mechanical devices, the sole purpose of which is sexual stimulation. Porn shops litter the highway selling these devices as well as video tapes of sexual acts. We have an online army of porn peddlers that, combined, probably out gross many countries. You can even buy a life sized doll made out of silicon that has an artificial internal skeleton (Check it out). We have personally witnessed the way in which technology has evolved sexuality and now you say that we can not use that experience to judge what seems like the next step in its evolution, particulartly one that seems eminent?
As a species we do this all the time: use past experience to predict and judge the future. We are doing it right now in relation to the cloning debate. If we can not use history as a guide for the future, then we are lost. Note that I said use history as a guide. No one is saying sex with robots is perverted because sex with animals is perverted. We are using reason, logic, history, and personal experience to make a judgement on a scenario. The fact that we have never actually been in that scenario does not make the judgement wrong, it simply means we have to be more careful in making it.
-
-
It is possible for your mind to be so open that your brain falls out.
Am I the only one who thought Kismet looked like he could be Crow's ancestor (of mst3k fame :)
XML is like violence. If it doesn't solve the problem, use more.
The hard part is still how the robot learns and applies its knowlege. Are they making robots or the next generation of Furbys? ;)
Geeky modern art T-shirts
Linus Torvalds looks like a dork. I'd figured he was at least some skanky, pierced cyberpunk type based on all the hype he's generated but he looks like a home economics teacher. I guess that invalidates his life's work, too...
Unless it learned how to be "cute" itself, it's just an AIBO. Why would MIT be pushing this when they've almost certainly got more significant projects running?
"HORSE."
-Flaming Carrot
> How's that for robotic news?
Pretty poor.
But JFKj is merely missing, not confirmed dead yet.
OFTC: By the community, for the community
No I should run aroud hiding my identity because I'm afraid that my nick name will be associated with me and everyone will be out to get me.
ACs.
OFTC: By the community, for the community
Of course, I'd be prepared to fight that poster
:-])
in a duel of Quake II to defend the ladys honour.
If only he'd given his name...
Of course, that would be more knightly than geeky, and perhaps the lady would take more
satisfaction out of it if she'd frag him herself.
(Repeatedly. With a blaster. From behind. With
eyes shut and one hand bound to the back
But still, the offer stands. (En garde!)