A.I. - Artificial Intelligence

[amjoie]

I think we need a new category in the forum for discussing AI, both now and in the future. :)

The actual discussion surrounding a future life form that borders on sentience is an interesting discussion, because it involves the idea of whether or not man is capable of a genuine "creation of life" -- in a sense making man a "god," of sorts.

As a "god" what responsibilities does man have toward his creation, and what responsibilities does his creation have toward the "god" who created it?

But I would sincerely doubt that anyone at Ugobe is concerned about whether or not creating Pleo nudges them onto a "god" level. LOL

More to the point regarding Pleo is what are the boundaries of his AI? What level of intelligence can we expect of the little critter soon to share our homes? And that answer is a moving target. It all depends on the development of his software.

The development of his software, in turn, depends on the extent his physical body can accept software commands, and what those physical limitations mean to the advancement of First Hatch level Pleos. It is a question of form following function and function following form.

Roughly speaking, Pleo is an "herbivore" with a lumbering form of locomotion on thick limbs. He is unlikely to advance beyond a "beast" level of intelligence, regardless of his computer chip -- simply because he does not have opposable thumbs, for instance. Or soft pliable mouth parts that might allow him to use some sort of tool to enlarge his capabilities, in the absence of fingers/thumbs.

In fact a really sophisticated and high level AI would be wasted on a form like Pleo, unless you wanted your pet to engage in challenging conversation as the extent of the use of his intelligence. Which I suppose is a laudable goal, in its own right. :)

Still, it seems to me, we must content ourselves with this fact: Pleo was designed for nothing more than our enjoyment. He is, and always will be, a pet-like diversion designed to delight and entertain humans. And I predict he will be very good at it. So as an artificial life form, he will have a successful niche. :D

But that doesn't mean he needs more intelligence than, say, a cockroach. LOL ;)

[PixiePleo]

Hi amjoie

Ok, so first of all if man creates artificial intelligence that is self aware I would imagine we would have the same obligations as we do to children. However, seeing that you can program A.I. you would never have the helpless years, your artificially intelligent creation could be theorhetically smarter than you. For instance you might have the skill to program your new AI child and upload all sorts of information before ever flipping the switch to bring that intelligence on line.

I'm kind of thinking of Star Trek: TNG here and Commander Data. Data learned and grew through the show as far as understand humans and human emotion went, but he was a super intelligent computer as far as the amount of information he held in his "Positronic Brain."

So as a parent of this AI are you obligated to humankind to upload morals? manners? From what culture?

So I am jumping WAY ahead than we are technologically capable currently. But hey speculation is the fun part right ?

[amjoie]

Hi amjoie

Ok, so first of all if man creates artificial intelligence that is self aware I would imagine we would have the same obligations as we do to children. However, seeing that you can program A.I. you would never have the helpless years, your artificially intelligent creation could be theorhetically smarter than you. For instance you might have the skill to program your new AI child and upload all sorts of information before ever flipping the switch to bring that intelligence on line.

I'm kind of thinking of Star Trek: TNG here and Commander Data. Data learned and grew through the show as far as understand humans and human emotion went, but he was a super intelligent computer as far as the amount of information he held in his "Positronic Brain."

So as a parent of this AI are you obligated to humankind to upload morals? manners? From what culture?

So I am jumping WAY ahead than we are technologically capable currently. But hey speculation is the fun part right ?

Ah, now you get to the meat of the problem. You see, I am of the opinion that your responsibility to the life form you create includes giving it a real life -- by that, I mean freedom of choice.

So, the life form will require the ability to learn by making mistakes. Which means it must start young and stupid, learning slowly enough that the mistakes it makes won't be so damaging or dangerous that it destroys itself or others.

And all during this learning period, you must provide adequate lessons in a number of ways, both by direct training and also by providing learning opportunities that it can use to discover things on its own, making sure it never becomes bored. It must stay motivated to continue to learn and grow.

You absolutely cannot give it super powers unless you take away all freedom of choice.

Hopefully, your training and guidance will allow the life form to develop the moral fiber it needs to survive, and to enjoy its survival, while causing no harm to others in its environment.

It would not be possible to hardwire morals, aside from a few simple directives, like the need to survive, balanced by the need for others to survive. Think in terms of the base instincts of a species. Otherwise, morals must be learned within a particular environment.

During this entire process of growth and learning, you must provide the equivalent of food, shelter, and clothing, along with the promise of security so that the creature feels safe and sure of its future.

*But,* you cannot take away its incentive. So, you must give the creature a function -- a reason to exist. A job to do, with some kind of upward mobility in the field. A feeling of self-worth. In other words, a vision for its own existence that is compelling enough to keep it looking forward to each day.

See? The responsibility of a creator is much more complex than you might imagine.

Not easy to be a "god" is it? LOL

[hardyfoster]

you could have several pleos around the house that could make great security guards. no one would ever suspect them.

of course it would be a low-end security system, but with some slight modifications you could up the standard..

here's something for you as well
http://www.incx.nec.co.jp/robot/english/robotcenter_e.html

and this... these robots are able to form different shapes and such depending upon how many are working together, then figure out a way to move itself...
http://www.isi.edu/news/news.php?story=160

[PleoWorldHostSauri]

hardyfoster the Superbot and the Papero are definitely cool robots. Thank you for sharing the links!

[PixiePleo]

amjoie I have plenty of reasons for an AI to exist!! First it could take over all my duties, then it could bring me nice cool drinks while I sit on the beach watching the surf, then .... lol

Ok so now comes another question, they are machines, do you give them days off?

[amjoie]

you could have several pleos around the house that could make great security guards. no one would ever suspect them.

of course it would be a low-end security system, but with some slight modifications you could up the standard..

here's something for you as well
http://www.incx.nec.co.jp/robot/english/robotcenter_e.html

and this... these robots are able to form different shapes and such depending upon how many are working together, then figure out a way to move itself...
http://www.isi.edu/news/news.php?story=160

amjoie I have plenty of reasons for an AI to exist!! First it could take over all my duties, then it could bring me nice cool drinks while I sit on the beach watching the surf, then .... lol

Ok so now comes another question, they are machines, do you give them days off?

Interesting. So as a gods, the first thing you would create would be a servant class of life forms, designed to protect you, relieve you of unwanted chores and also serve you. I assume, in addition, you would want them to have a somewhat worshipful attitude? ;)

You realize that for your life forms to enjoy their work, they would have to find the things you *don't want to do* both fun and challenging; with the added vision of doing it for the remainder of their lives and the incentive of serving you in an "ever better way," giving them the desire to wake up each day happy? LOL :rolleyes:

Actually, you could achieve this balanced life for them, by limiting their intelligence to a certain level. But they are never going to end up being Data, that's for sure. :D

So, not only does form follow function, and function follow form -- but the life form's intelligence itself has to be adjusted depending on what the creature is designed to do. We have just added another level of complication that we, as gods, must consider for each of our creations. What is the limit of their capacity for intelligence?

..............

I saw a demo of the superbots on satellite TV. It was fascinating, and reminded me of the replicators from the SG1 sci-fi show. I would say that they have been given some survival skills, on a rudimentary level -- and I suppose some could argue that they would have the beginnings of rudimentary "self awareness" as a "group of bots" creature.

But I don't think it qualifies for actual intelligence. They still need to be directed by a superior mind, who gives them a mission and reaps the reward of their service. That relegates them to being nothing more than sophisticated tools.

edit: I forgot to address the "toddler with tubes" bot. It looks like a Pleo "brain" on steroids, in a different form. Still no fear of eventual sentience, at this level of programming. However, the humanoid form means it has a higher potential for development. But notice the AI is starting at a baby level, and their goal is to get it to the three year old stage -- and it will take quite a while to get it there.

I can't really think of a good reason to have a three year old humanoid hanging around, other than to keep me busy, attempting to keep *it* out of trouble. I would view it as a recipe for disaster. :P

I raised human toddlers, and the good thing about that stage of life is that they eventually grow out of it. LOL

So, while this is a fascinating experiment in the current limits of technology, it is far from useful at this stage of development. It gives a whole new meaning to "taking baby steps" toward a technological goal. :)

[PleoWorldHostSauri]

Has anyone seen the developments being made with A.I and cars? There was a whole show on the Science Channel about the cars, it was really amazing.

[hardyfoster]

the valerie android,

if you scroll down to around the middle, there will be a link to give an example of some AI programming.

also, there is a CD of some of the acutal AI. I'm sure you can read, so you'll see it...

they already have a robot that can wash dishes and such by the way.

http://www.androidworld.com/prod19.htm

[hardyfoster]

oops.... this got me to thinking...
i wonder if a speech thing could be put into Pleo, then put in valeries AI data CD and then you could ask pleo a question and get an answer....


this is perhaps for another thread... but i think the next hurdle with robots should be:

1. lengthen the battery charge
2. reduce the charging time
3. improve the robot recharging itself
4. perhaps some kind of self-charging while in use. I remember Honda talking about one of their prototype cars (i wish they would make it, it's look COOL !) being battery operated and the batteries would be recharged somewhat from the energy used when braking.

Perhaps as a robot moves, some of the movement or energy can supply the battery with some power......

then, it would make a great pet for people in nursing homes.....


link to honda car....(the drivers seat is modeled after a fighter jet cockpit (click on picture 3) ... i would have so much fun in this car. http://honda.motorcentral.net/cars/2001_Honda_Dualnote_Concept/pictures/ )

[amjoie]

the valerie android,

if you scroll down to around the middle, there will be a link to give an example of some AI programming.

also, there is a CD of some of the acutal AI. I'm sure you can read, so you'll see it...

they already have a robot that can wash dishes and such by the way.

http://www.androidworld.com/prod19.htm

This project started in 2005, and right now they are selling fingers and a software program. The pictures of Valerie are just mannequins. They have no working robot, just parts -- unless I missed something on the website. And all those parts have been available for a very long time. I would have to see a working prototype of the entire robot to be impressed. I'm not even sure this thing is for real. (shrug)

edit:
Has anyone seen the developments being made with A.I and cars? There was a whole show on the Science Channel about the cars, it was really amazing.

I missed that show. What were they using the AI to accomplish, and how did it interact with people? Tell us more. :)

[PleoWorldHostSauri]

amjoie, the AI was so that the car could navigate itself and not hit other cars, animals, people. It was doing the actual navigation via gps but the responses to obstacles were AI. The driver didn't have to do anything other than tell the car where they wanted to go. They said it was still in development and talked about other features but I am having a tough time with my recall today LOL.

[hardyfoster]

here are a few links to get you started....
i'm not really impressed with talking to ALICE, but the site gives you alot of information.
http://www.alicebot.org/
you can click on "what is aiml" on the left for more information.

I was/am more impressed with ELIZA. At the time i first encountered eliza (years ago) i came up with a programmable way you can teach the computer to learn how to hold a conversation with a person. i never pursued programming it though. it would be better to have a team to accomplish something like i have in mind.
for example, if i mentioned "soap". the bot would first need to analyze the conversation to see what and why i'm talking about soap (we'll assume it has already learned the associations). Soap can be associated with washing your car, taking a shower, washing the dishes, how soap is made, etc. THEN the bot would be able to make a proper statement or ask a question to continue on with the conversation.

anyways, here's a link to Eliza.
http://www.chayden.net/eliza/Eliza.html

wikipedia chatterbot
http://en.wikipedia.org/wiki/Chatterbot

lots of different bots
http://www.knytetrypper.com/ain.html

[amjoie]

here are a few links to get you started....
i'm not really impressed with talking to ALICE, but the site gives you alot of information.
http://www.alicebot.org/
you can click on "what is aiml" on the left for more information.

I was/am more impressed with ELIZA. At the time i first encountered eliza (years ago) i came up with a programmable way you can teach the computer to learn how to hold a conversation with a person. i never pursued programming it though. it would be better to have a team to accomplish something like i have in mind.
for example, if i mentioned "soap". the bot would first need to analyze the conversation to see what and why i'm talking about soap (we'll assume it has already learned the associations). Soap can be associated with washing your car, taking a shower, washing the dishes, how soap is made, etc. THEN the bot would be able to make a proper statement or ask a question to continue on with the conversation.

anyways, here's a link to Eliza.
http://www.chayden.net/eliza/Eliza.html

wikipedia chatterbot
http://en.wikipedia.org/wiki/Chatterbot

lots of different bots
http://www.knytetrypper.com/ain.html

Oh wow, those are great links. I'm still having fun reading them, and trying out a bot chat, but I wanted to come back here and thank you for posting the links. :)

[PleoWorldHostCami]

I agree amjoie, some great information there hardyfoster. Thank you so much.

[amjoie]

Asimo is being discussed in another thread, but I think some links should be here, since Asimo is a good example of AI put to use:

http://asimo.honda.com/asimotv/

http://world.honda.com/ASIMO/

[ryane]

i think mit really has the upper hand in the ai field, because they understand that it would take centuries to code all the information we hold in our minds. they are building robots that can learn. you all probably have read about the robot cog or seen him in the news, if not check it out. i really think this approch will pave the way for trully intelligent robots. it's only a matter of time...

[pleoguy101]

Has anyone seen the developments being made with A.I and cars? There was a whole show on the Science Channel about the cars, it was really amazing.

oh ya I saw that!!!

[June03]

I'm in the process of testing and trying to incorporate some kind of 'extra AI' into my aibo's main personality (on the ers-210). The aibo is connected to my home computer via my wireless network and the computer is the server (database) where the commands are interpreted and acted upon. Kind of like Eliza or Alice.

See here: this is the HomieGate that can be used with aibo and spoken commands/questions
http://www.aibohack.com/homie/overview.htm

Wouldn't it be nice if Pleo one day could use something similar? And maybe even connect to the internet via the computer to search for informaiton if it isn't in the database...

[pleoguy101]

I was wondering, it says pleo will get used to things like surprises in this pagehttp://www.pleoworld.com/discover/emotions/emotion/scared do you think my pleo will get to be ok around loud noises like my drums and when my friend comes over and plays the electric guitar while I'm playing drums, do you thinkl pleo will become ok around those things??

[cmadmacs]

He might want to join in.

[pleoguy101]

LOL, good point, maybe he could dance:D

[hardyfoster]

pleoguy, are you and your friends planning on getting the "rock band" game ?

[mweed]

I'm sure we could program his head to bob back and forth like a metronome to help your band keep time, or maybe whack a small drum with his tail to set a beat... :)

[pleoguy101]

pleoguy, are you and your friends planning on getting the "rock band" game ?

well actually, my friend, or should I say pleoguy102 is planning on getting that guitar hero game.. the new one thats coming out

[pleoguy101]

I'm sure we could program his head to bob back and forth like a metronome to help your band keep time, or maybe whack a small drum with his tail to set a beat... :)
thats a good idea mweed, pleo could play drums with me, or he could be backup singer:D

[hardyfoster]

yes, guitar hero is great. which "rock band" is a singer, 2 guitarists, and a drummer. since you mentioned you play drums, i'm sure you'd like the drum portion. it has 4 pads and a pedal for bass.

[pleoguy101]

oh, thats gonna be cool:D

[Lu Bu]

I think Pleo will have enough AI to get used to those things right out of the box, though if he can't, a software called LIFE OS is going to upgrade its AI to make it smarter and learn overtime.

[Lu Bu]

I think AI will be greater in complexity than human intelligence in maybe the next decades or so. Because robots like ASIMO, QRIO, ETC. Already have enough AI to learn, throw, detect human faces, recognize and learn words, have conversations, etc. which all humans are capable of doing.

[radioastronomer]

I think AI will be greater in complexity than human intelligence in maybe the next decades or so. Because robots like ASIMO, QRIO, ETC. Already have enough AI to learn, throw, detect human faces, recognize and learn words, have conversations, etc. which all humans are capable of doing.


We are still a ways off (and quite possibly will never achieve) from the type of "AI" that most envision. Here is the rub - why would we want to make machine intelligence mimic the way a human brain works when silicon is structured differently? Yes they are both finite state machines, however, machine intelligence most likely will take its own (least resistance) path. We may not even recognize it as such.

[Caaahl]

I just watched "A.I." and Boy, i am moved. And i want my Pleo.....now! :D

[Lu Bu]

Pleo ain't as complex as you thought. His recognition of color and reactions to color is very inconclusive and unresponsive, due to the tests and vids I have seen so far. Took a guy 25 tests (videos too) just to try to see if Pleo can react to color spontaneously and constantly but turns out, he only reacts to a color sometimes, most of the time he only looks at a color for only 1 second and ignores it.

[Caaahl]

Pleo ain't as complex as you thought. His recognition of color and reactions to color is very inconclusive and unresponsive, due to the tests and vids I have seen so far. Took a guy 25 tests (videos too) just to try to see if Pleo can react to color spontaneously and constantly but turns out, he only reacts to a color sometimes, most of the time he only looks at a color for only 1 second and ignores it.

I am in contact with this guy and new tests are on their way. ;)

*edit* ah ok, the new tests are already there.

[brian55127]

We are still a ways off (and quite possibly will never achieve) from the type of "AI" that most envision. Here is the rub - why would we want to make machine intelligence mimic the way a human brain works when silicon is structured differently? Yes they are both finite state machines, however, machine intelligence most likely will take its own (least resistance) path. We may not even recognize it as such.

That would be sad if we didn't recognize it because we would probably inadvertantly erase any evidence of machine intelligence with a hard drive wipe or system restore.

[InNM]

Speaking if A.I., do you know about Ray Kurzweil and his articles on computing power this century? Here is a link to his site: http://www.kurzweilai.net/index.html?flash=1
Ray has plotted the evolution of computing power from the beginning of electronic computers and found that it is increasing at a double log rate. Based on this, a computer will have the processing ability of all the humans on earth around the year 2060. This will be the singularity where it is esentailly infinite. Human evolution has stopped. This is the century of Human/machine evolution. Barring a major destructive event, humans will be part machine during the next 50 years. The question of the "God" computer comes up. we can build super intelligent machines (and most likely will), but what happens if they think we are not needed? Of course, humans will be infinitely smarter due to machine interfaces. see article http://www.kurzweilai.net/articles/art0134.html?m=1 about Rays projections. After the first part it gets really interesting.
BTW, when I started using the internet, it was all text and the modem was 300 bits/ second. You could watch the line move slowly across the screen. Now, in Japan they get up to 60 Million bits/S at home and look at the internet.