I tried it Carl but got:

*************************************
******spambot protection protocol******
*************************************
*you must enable cookies to use this site*
*************************************
*********1377810666******************

Steve Worswick - Aug 29, 2013:

I tried it Carl but got:

*************************************
******spambot protection protocol******
*************************************
*you must enable cookies to use this site*
*************************************
*********1377810666******************

Just hit reload on your browser and it should work.  Sorry, but I have to use cookies to prevent comment spam bots from overrunning that page.

Yes that worked. Thanks Carl.

Don Patrick - Aug 29, 2013:

... and you should be aware that all art is copyrighted.

If this is directed at Mitsuku, I can assure you that the artwork is original and was created especially for the project.

I remember a bot from UC,
whose humour was well as could be,
but much as it knew,
its best wouldn’t do,
it turned out the joke was on me.

A robot that understands humour.

Steve Worswick - Aug 30, 2013:

Don Patrick - Aug 29, 2013:

... and you should be aware that all art is copyrighted.

If this is directed at Mitsuku, I can assure you that the artwork is original and was created especially for the project.

I presumed as much from the looks of her lining, hats off to you, sir . My remark was mostly made in regards to the pink bikini-clad image on Carl’s site, and forgettable recollections unaffiliated with this forum. What did spring to mind at seeing Mitsuku’s avatar is that it looked familiar. Could it be that I happened across it in 2004?

More than likely, as Mitsuku has been online for around 9 or 10 years now.

Merlin - Aug 29, 2013:

IPod-AI—I sync, therefore I am.

that was hysterical Merlin

@Don, that link you posted had knock-knock jokes… If you ask me (and still building on my ‘store phonetic information with your knowledge base if you want jokes’), those should be pretty easy to generate as well. The UC project thought among the same lines clearly, basing their ‘recognize a joke’ algorithm on homonyms, though they didn’t seem to have taken it as far as I think is possible.

Somebody, get that Taylor dude a postdoc grant!

Skynet-AI includes a number of jokes and puns. But it is not well suited to stand-up comedy competitions.
I have been thinking a lot about the process of humor and how I come up with something I think is funny.

The “Sneeze the day”—Jimmy Durante joke is a good example.

Wouter did a good job detailing a process, and CR points out the difficulties.

How I went from seize to sneeze is pretty straight forward, but the process of relating sneeze to nose to entertainer with a big nose, is all about knowledge management in a neural net. I have not thought about Durante in about 30 years (bot master shows his age). If not for creating this joke, I may never have thought about him again (which is a shame, because in my childhood I found him very entertaining).

How do we store gigabytes worth of data and free associate it in near real time to come up with something intelligent, let alone funny.

http://www.buzzpatrol.com/google-new-algorithm-for-humor/

http://en.wikipedia.org/wiki/Computational_humor

How do we store gigabytes worth of data and free associate it in near real time to come up with something intelligent, let alone funny.

People in search for the mind’s internals worry about this too much imho. Algorithms first, performance second - when it comes to performance, technology always catches up.

How do we store gigabytes worth of data and free associate it in near real time to come up with something intelligent, let alone funny.

When it is indexed and available via an API?

Wouter Smet - Aug 30, 2013:

People in search for the mind’s internals worry about this too much imho. Algorithms first, performance second - when it comes to performance, technology always catches up.

Precisely, and the more minds working on codifying something like humor the better the results will (eventually) be. 

Regarding the whole anime chick avatar background, have to agree it is old- something I’ve been meaning to update for a while… the “matrix” layout is my personal preference, but it can scare off some of the more timid folk.

Carl B - Aug 30, 2013:

How do we store gigabytes worth of data and free associate it in near real time to come up with something intelligent, let alone funny.

When it is indexed and available via an API?

I agree, I am thinking about preprocessing big data and then indexing into the results.

Wouter Smet - Aug 30, 2013:

Algorithms first, performance second - when it comes to performance, technology always catches up.

Even now, storage and cloud processing eliminate many performance limitations. But finding smart ways to work with the data is still paramount.

I have a standard mental procedure for generating wit on the fly. For example, if someone gets kicked, one could say “He got a real kick out of that.”. Like phonetic similarity, this is a statement that can be read two ways. What my brain commonly does is generate a handful of words that describe the event, e.g. “kick” and “foot”, then it goes through my database of common expressions and movie quotes, trying to find expressions based around the same word (or same sound) that could also apply to the event. Thus ending up with “He got a kick out of that.” or “The shoe is on the other foot / putting his foot down / fancy footwork.” etc, etc. Here the trick is to know the figurative meanings of the expressions and why they are appropriate, but you probably wouldn’t do half bad without even that.

Personally I suspect that laughing is a sort of glitch after perceiving contrasting information. It’s fun, but at the same time I think it is odd to regard the human brain as superior to machines on the grounds that humans are less capable of processing conflicting information. That’s probably irony right there.

I have a big ontology of common words, along with meanings, colours, rhymes, what it is made of, how big it is and so on. To get Mitsuku to make a joke, she looks through the different properties to see which fit best in a selection of rules I wrote. For example, for anything blue, she would probably say:

Why was the sea/sky/blueberry sad?
It was feeling blue

Similarly, for musical instruments:

Why did the trumpet/drum/piano lean against a wall?
It was musically inclined.

Some other rules:

What sort of music does lead/iron/tin like?
Heavy metal

What TV crime show does six/twenty three/four like?
Numbers

What did the ham/beef/pork say when he saw his friend?
Pleased to “meat” you.

and so on.