Laugh Out Loud!

[Sakonosolo]

I don't know if it actually puts a real person on the other side or if it takes things that people have typed in and tries to find the best match for a response. So say you type in "How are you?", the computer would search for something that a real person used earlier as an answer to that question and then puts it as a response. With probably millions of people using this it can build up quite a large database and eventually get really good at conversation. At least that's what I always thought, though earlier it might have been person to person just to built up responses. I'm not quite sure anyway.

[MDL]

ah, that would actually make a lot of sense.

Spoiler:

[fedcom]

I don't know if it actually puts a real person on the other side or if it takes things that people have typed in and tries to find the best match for a response. So say you type in "How are you?", the computer would search for something that a real person used earlier as an answer to that question and then puts it as a response. With probably millions of people using this it can build up quite a large database and eventually get really good at conversation. At least that's what I always thought, though earlier it might have been person to person just to built up responses. I'm not quite sure anyway.

I'm pretty sure that's how it works too. I've experimented with it before and it feels like I'm talking to a person with incredibly short term memory or something. Once I asked it if it spoke french and it said 'yes'. Seconds later I asked it if it would like to live in France and it said "no because I don't speak french" or something along those lines.

Has anyone here heard of a Turing Test?

[Sorikai33]

Has anyone here heard of a Turing Test?

Yeah. That's where they try to determine if a machine can exhibit intelligent behavior. A human judge has a conversation with both another human and a machine. If the judge can't tell the difference, the machine passes the test.

Cleverbot took one last year, and passed.

http://www.newscientist.com/article/...-is-human.html

[Tofufu]

Chuck Norris has betrayed us!

Spoiler:

[RoxasMP1000]

[dasia]

Loool

[Nobodyman]

Chuck Norris has betrayed us!

Spoiler:

http://i.imgur.com/HrDkn.jpg

You fools, he's just luring him into a false sense of security.

[dpara]

Yeah. That's where they try to determine if a machine can exhibit intelligent behavior. A human judge has a conversation with both another human and a machine. If the judge can't tell the difference, the machine passes the test.

Cleverbot took one last year, and passed.

http://www.newscientist.com/article/...-is-human.html

That is because most people don't know how to ask "hard" questions;

If I shoot someone do go to prison?
If I pick my nose till it bleeds will one of my fingers be bloody?
Can I sit in a chair made of smoke?
If I whack someone with a fresh fish will it make a BABOOMBOOMBANG sound?

..these are still a bit easy since by guessing yes or no the PC will still get it right for some people.

But something like:

How many fish would it take to fill a full bin with fish?
Why don't we build everything out of ice?
Why doesn't lots-of-water make a great fossil fuel supplement?
If something releases water under pressure then it is likely?

or something with a wordplay, humor or analogies will break (very very certainly) all current AI applications.

[bartholemew kuma]

[Shuhan]

[Wagomu]

I decided to glance at one of the many NSFW webcomics that advertise on MSPA for shits and grins. And so did I reap said shits and grins, especially on this page (which is actually pretty SFW). I am not sure what to say to this:

Hidden:

[fedcom]

Yeah. That's where they try to determine if a machine can exhibit intelligent behavior. A human judge has a conversation with both another human and a machine. If the judge can't tell the difference, the machine passes the test.

Cleverbot took one last year, and passed.

http://www.newscientist.com/article/...-is-human.html

Re-really? I've never once been given the impression I was talking to a real human on cleverbot.

or something with a wordplay, humor or analogies will break (very very certainly) all current AI applications.

Haha yeah. I can't imagine bots being able to handle puns or onomatopoeia.
Cool thread btw:http://www.reddit.com/r/IAmA/comment...an_award_at_a/

[Sorikai33]

Re-really? I've never once been given the impression I was talking to a real human on cleverbot.

My explanation of the Turing test was the way Alan Turing originally designed it. It's use in standard practice is not nearly that simple. For example, the one with Cleverbot had 30 human participants, to reduce human error. At the end the responses were rated, and Cleverbot was determined to be 59.3% human. 50% is needed to pass. As dpara said, it is still easy to break most AI. All the test proves is that Cleverbot can use various algorithms and a huge database of phrases to imitate basic human intelligence.

Edit: Cleverbot just told me the test they took at school today was tasty and I will destroy the Earth in an hour. Basic human intelligence.

Edit2: And a picture so the thread stays somewhat on track.

[zachri]

[bartholemew kuma]

Zach, your Epic!
Nice to see you back

[zachri]

Zach, your Epic!
Nice to see you back

Why thankyou ;D

[zachri]

I don't even know what I have just witnessed and I doubt you will either.

[Kaoru]

Darth Vader in a Kild, on an Unicycle Playing the bagpipes...

your argument its invalid!


....

OK I don't know WHAT is it that I just Saw, but It made me crack on laughter!

[Nobodyman]

In the spirit of St. Patrick's Day.