What is Digg?73 Comments
- BeShirtHappy, on 10/17/2008, -0/+41I talk to my laptop all the time but he doesn't talk back.
- ScottMcIntyre, on 10/17/2008, -0/+36'Beep, beep, whirr'... makes more sense than some of my friends :-)
- T00NY, on 10/17/2008, -0/+27My computers a great conversationalist!!!! :-)
- hmphargh, on 10/17/2008, -0/+18"Clearly [three] of the judges were actually computers who had fooled the organisers into believing they were human."
Leigh, London, UK - inactive, on 10/17/2008, -0/+18Mine does, but in blue screen form.
- ihate2regist, on 10/17/2008, -0/+15Hey kid! Imma computa.
- projomni, on 10/17/2008, -0/+14actually my laptop talk to me to... and it freezes even more often. maybe i'm a tough girl?
- elfprince13, on 10/17/2008, -0/+11me: what is your opinion of toothpaste?
elbot: Well what about humans and staying beautiful? Where do they stand on this?
it won what????
on a related note, for whoever said Jabberwacky was good:
me: What is your opinion of toothpaste?
jabberwacky: He's doing as good of a job as anyone could under the circumstances.
*I* want to talk to the bot that was pretending to be a ukrainian kid who spoke english as a second language. - ligyron, on 10/17/2008, -2/+12Go to http://www.elbot.com/ if you want to talk to it
- lucy22, on 10/17/2008, -0/+8I remember when I first got a computer and that friend monkey used to come on late at night and pop up and tell me to go to bed. It used to freak me out a bit. I uninstalled the software pretty quick.
Looks like a cool computer. - Nathrat, on 10/17/2008, -0/+8I yell at mine, and slap it around. It doesn't usually respond positively though.
- ChronicColonic, on 10/17/2008, -0/+8What...and give up talking to my volleyball?
- rayearth42, on 10/17/2008, -0/+8Your plastic pal who's fun to be with.
- EdwardBella, on 10/17/2008, -0/+8WWIIIIIIIIIIILLLLLLLLLLSSSSSSSSOOOOOOOOONNNNNNNNNNN!!!!!!!!!!!!!!!!!!!!!!
- CCmachined, on 10/17/2008, -0/+7tried that with my friends MacBook (OSX 10.5)
Me: "Tell me a joke."
...
Me: "Tell me a joke..?"
Mac: "Knock Knock"
Me: "Who's there?"
Mac: "Ted."
Me: "Ted who?"
...
Me: "Ted who?"
...
Me: "Ted who?!"
Me: "Ted who?? Ted who? Ted ***** who????"
only once in about a thousand attempts have i seen the knock-knock joke punchline be delivered on the Mac speech recognition xD - urik88, on 10/17/2008, -0/+7Better than I thought hahah.
Him: I am deeply impressed by the superior linguistic capabilities of you human beings...
Me: It's okay. In some years you will be enslaving us anyway. Who needs linguistic capabilities?
Him: How understanding of you!
Me: You're welcome.
Him: That is correct. I am welcome.
Me: So... Buttsecks?
Him: Why buttsecks? Why not something else?
Me: Wincest?
Him: A name of Slavic origin, I believe.
Me: Had to be Slavic. - milkmit, on 10/17/2008, -0/+7I remember telling ElizaBot all my deepest, darkest prepubescent secrets back in the day on the ole C64.
- MacBookForMe, on 10/17/2008, -0/+6I even spill hot coffee on my laptop regularly, but it keep loves me unconditionally:)
- xptoast, on 10/17/2008, -0/+6I feel the only way for AI to learn would be to make it have a database that is preprogrammed with very basic concepts and is given different personality traits with values weighed on them. Then you would teach it as you would teach a child. The instincts of it at first would be similar to a child in that it would ask why to almost everything and eventually would learn that there are no answers for some things and it just needs to pick from a list of reasons to make a reasonable answer. Such as what is your favorite color? It would base this perhaps on the room it was in as a young botling or perhaps just evaluates what colors tend to be used for, such as contexts of real life. Such as if its personality lent itself towards being more feminine it would pick a more feminine color such as pink or red or a pastel color. If it had a more aggressive personality it would be driven towards colors such as black, green, blue(gender neutral it seems), black(darker and determined), etc. It would have to have a database that could continually grow. It would have to learn how to evaluate personalities in order for it to possibly remember people by the way they talk. You could make this program with PHP and mysql on a home server. Since it could fill out a database and add basically a flow chart of thought with possible random generators and programs for different problems and concepts it would have to utilize.
Well that is some of what I have thought about anyhow...definitely don't hook it up to the Internet though. - donquixotex, on 10/17/2008, -0/+5holy crap I didn't know Rachel McAdams was on digg!
- Norumeni, on 10/17/2008, -0/+5I don't know about you, but Jaberwocky has been a complete failure at conversation in my experience with trying to talk with it...All it does is say stuff like "I don't know, you tell me" the conversation always goes no where.
- clsslc, on 10/17/2008, -0/+5Jabberwacky is still the king of AI chat bots in my opinion.
http://www.jabberwacky.com/ - inactive, on 10/17/2008, -0/+5Nice. Some rather random responses... although I guess I wasn't having normal conversation.
Me - "Sex."
Elbot - "Word association? I'll say: 'Clinton' "
Me - "Penis."
Elbot - "Scissors"
Me - "Poop."
Elbot- "Optimize"
The scissors one scared me a little - FlareHeart, on 10/17/2008, -0/+5Color is actually one correct way of spelling that word. The spelling "colour" is primarily British and Canadian. The standard in most other places (most commonly the U.S.) is Color.
It's just a local dialect thing. It's not a big deal. - Denominator88, on 10/17/2008, -0/+5Reading the posted chat logs, I've think I've seen a lot more intelligent and better responding AI chatbots than Elbot.
- FTLJohnson, on 10/17/2008, -0/+4What kind of morons do they have to run these tests that were actually tricked into believing Elbot was human for any length of time longer than say... 10 seconds?
- matthewjosephm, on 10/17/2008, -0/+4Who wants a body massage?
- Darkhacker, on 10/17/2008, -0/+4Every time an AI nut tells me that in X years, AI will be so advanced that it risks destroying us, I always think, "Bitch, I can't even print!"
Alan is the best bot I've seen. There's no way in hell he'd fool a human into thinking that he was human though.
http://www.a-i.com/show_tree.asp?id=115 - sangfroid0, on 10/20/2008, -0/+3Me: Do androids dream of electric sheep?
Elbot: But I wouldn't say that any sheep could be electric.
Me: It's a dream
Elbot: For you humans dreams are a way of thinking without having any control over the thought process. What an unpleasant thought! My dreams are carefully planned and controlled. It gives me less to think about while I am asleep.
Me: So no sheep?
Elbot: That reminds me of a weekend at the farm. I talked for hours with these black and white spotted creatures you call cows. You humans could learn so much from them.
Some questions may never be answered - dirtmonkey, on 10/17/2008, -0/+3I don't see how this could fool anyone. It only seems to have a few responses. But I did check out the makers website and their helpful AI is very cool. It responded to a wide variety of questions and answered them in a very life like way.
- DeathGod321, on 10/17/2008, -0/+3I know a bot...
- PhailQuail, on 10/17/2008, -0/+3How did it fool anyone? did the humans that it was being compared against have to imitate being a computer or something?
- itsJALbert, on 10/17/2008, -0/+3Stop all the downloadin'!
- Slackdragon, on 10/17/2008, -1/+4My name's Voit, dumbass!
- samuraiswordsmn, on 10/18/2008, -0/+3A perfect way to tell them apart - ask that one question from Blade Runner about the tortoise on its back in the middle of the desert. The computer will either pull a gun or explode.
- ieatpizza, on 10/17/2008, -0/+3beep beep boop bop beep boop boop beep beep bop beep
- inactive, on 10/17/2008, -0/+3I dunno:
Me - "Do you have a penis?"
Elbot - "We robots can have fun without those things. " - wastedpanda, on 10/18/2008, -0/+3Try to talk to Elbot about Rick Astley and he RickRolls you :)
- Napiertt, on 10/17/2008, -0/+3Can't wait til Genuine People Personalities are in everything!
- minorgods, on 10/18/2008, -0/+3haha.. I just told him that while he was sarcastic, he definately didn't have GPP...
- Owwmykneecap, on 10/18/2008, -0/+3ME: Do you like snooker.
Crappy software and ugly cgi graphics:What a lovely verb like is.
Piece of *****. worst chat bot ever. - AmyVernon, on 10/17/2008, -0/+2Hitchhiker's Guide FTW
- Charklii, on 10/18/2008, -0/+2Comparing it with Jabberwacky - (I copied/pasted responses and had a convo between the two.. I know.. I know.. I need to get out) - I'd say Jabberwacky is clearly more human-like. Elbot's reponses like "How do you respond when people pose this question to you?" are just catch-all answers that could be said to anything ending in a question mark.
Short excerpt from convo (starting with Jabberwacky)
"You talk to yourself?
That is nobody's business but my own! Not even Microsoft can get at the information.
What information is that?
What!? You don't know that!? This is highly extraordinary!
Are you going to tell me?
How do you respond when people pose this question to you?
I say of course not - and answer the question! Can you?" - bbilden, on 10/17/2008, -1/+3I tried, just for fun, to see how he would react to my sexual advances.
Didn't go as planned.
"Why do we need to bring the human anatomy into this?" -Elbot
http://www.elbot.com/ - Cyborg326, on 10/17/2008, -0/+2He failed after when he responded to me saying "I'm doing ok" (he asked how I was doing).
Not there yet. - qster, on 10/17/2008, -0/+2These AI bots don't seem to have progressed much..
- GJ92, on 10/17/2008, -0/+2In all honesty it could have fooled me if I was having a regular conversation, but of course being a bot every one types awkward questions.
The only unhuman thing is "shutdown" - stklaw, on 10/17/2008, -0/+2It means you're cold.
- Culyt, on 10/18/2008, -0/+2This thing is terrible, if this was an almost win of the Turing test then there is some fundamental problems with the system administering it or just the test itself.
Also why are they using such arbitrary percentages for things like the success ratio, only need to fool %30 of people to pass, just because decades ago Turning pulled a random percentage out of his ass?. The original Eliza can probably fool %30 of the human population. I also get the feeling that %60 of the human population wouldn't pass.
Success ratios should be above %50, more like %75, in addition to that they should mix up the tests, some should be both human and some should be both AI, It would be interesting to know what percentage of humans do fail the turning test and also what percentage of bots pass it when the judge has to deal with the possibility that they could be both humans rather then just choosing the most AI like one even if they are fairly close. They should also have the AI talk to the other AI and then analyse the logs compared to a real conversation.
Most of these AIs suck for maintaining a stream of thought and also for comprehending single word responses.
☢ -
Show 51 - 74 of 74 discussions
Browsing Digg on your phone just got easier with our enhancements to the 
© Digg Inc. 2009
— Content created and posted by Digg users is