AI Zone Admin Forum Add your forum

NEWS: Chatbots.org survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..

Neukom contests in story writing & poetry
 
 
  [ # 16 ]

Maybe the victory conditions weren’t online when I first posted this, because I wouldn’t have given these things much consideration if I had realised how much of a “Turing Test” they were going to be:
http://phys.org/news/2016-05-dartmouth-contest-good-poets.html

“The judges were hunting for machines so they are not looking at a Hallmark card and reading the poem inside”

 

 
  [ # 17 ]

I kind of have a real problem with the condition of having to “reveal the source code” to something like this, especially if it worked well.
Perhaps the word Proprietary doesn’t come into play in these “competitions”.

Then again, maybe it’s just me….

 

 
  [ # 18 ]

I thought they changed that rule to say something like “it would be nice to see the source code”?

 

 
  [ # 19 ]

Yep, it would be nice to see your secret workings of Mitsuku or Skynet-AI or any other prominent entry. Sure it would!

If the rules were changed as you assume, then one can simply say…“Sorry but No thank you!”

 

 
  [ # 20 ]

Exactly. I don’t plan on giving Mitsuku’s source files away on the chance of winning $3000.

Just checked the site and it says, “We prefer that submissions include source code”. I wonder how many entries actually did that?

 

 
  [ # 21 ]

Go ahead Steve, then in a relatively short time, Mitsuku will have lots of “cloned” brothers and sisters! Wow! What a legacy! Hehe!!;)

 

 
  [ # 22 ]

http://www.pcworld.com/article/3072669/analytics/can-robots-make-art-yes-but-dont-ask-them-to-write-a-poem.html
It’s much worse than I imagined. Only 2 entries for poetry and 3 for story writing. And the organisers are clueless. I think it’s pretty obvious why there was little interest: First you’re asked to send in your code, then the rules change, then winning becomes a blatantly uninteresting matter of whether the entries are human or not, rather than a matter of quality. Only fools would draw conclusions from a mere two entries.

Organizers of Dartmouth College’s “Turing Tests in Creative Arts” announced on Wednesday that there were no winners in their competition for AI algorithms to create “human-quality” short stories, sonnets and dance-music sets.

The goal was to show whether human judges—in the form of literary readers and party-goers on a dance floor—could distinguish which creations were generated by machines and which by humans. The contest’s organizers expected dozens of entries in each category but received far fewer submissions than they expected—“a testament to the difficulty of writing sophisticated code that creates another dimension of AI—creative intelligence,” they said.

In the poetry contest, there were two entries that each generated two sonnets; in the literature contest, there were three entries that each generated a short story. Some human-generated entries were included as well.

 

 
  [ # 23 ]

While randomly working on a rhyme algorithm, I came across one of the participants of the poetry contest explaining their approach:
https://www.ischool.berkeley.edu/news/2016/teaching-computer-write-poetry

 

 
  [ # 24 ]

Nice find Don. I found that article quite interesting but was also able to spot the deficiencies of the computer generated poetry.

While some / most of the words rhyme at the end of each sentence / line, those words are often used twice in the same sentence or stanza without regard to being previously used. This is something I’ve noticed of the years of reading computer generated poetry. Let not even get into Haiku which to me is nothing more than a license for a computer to be able to screw up and hope to go unnoticed. Mark it up similarly to being an alien or speaking a foreign language so what I say might not make sense.

Come on…those have all been tried and postured. People are somewhat wiser than they were when the first Eliza program was released.

The programs have gotten much better than those old days of trying to fool people but obviously, they still have a few small fences to vault.

 

 
  [ # 25 ]
Art Gladstone - Feb 20, 2017:

those words are often used twice in the same sentence or stanza without regard to being previously used. This is something I’ve noticed of the years of reading computer generated poetry.

Interesting observation. It makes sense that word variety is also important for the entertainment value of poetry, like the element of surprise in a joke. I’ll make note of that because next time there’s a (better) poetry contest I’ll probably give it a try.

 

 
  [ # 26 ]

Apparently this is now an annual contest. Here are the results of 2017: http://bregman.dartmouth.edu/turingtests/node/1

My conclusion is that I don’t get poetry. The entry B4 at least seems to have a versatile rhyming algorithm.

So highly famous Sony classical,
    A very pleasant happy television.
    That lovely goodness gracious Simon rattle!
    And then an awesome righteous folk musician.

    Are loving songs about the counter culture,
    But never ever ready steady go!
    This style is really timeless Donna summer,
    With any very groovy audio.

    A better lucky lady Marmalade!
    Who likes to play an old piano roll!
    For whether jazz or fame or being played,
    Clever like a psychedelic soul.

    The music sings a song of love and Bach,
    I love the sound of psychedelic rock.

 

 
  [ # 27 ]

Announcing the five 2018 contests to generate poetry, limericks, more poetry, kid’s stories, or music.
Additional requirement: All entries must be accompanied by a two-page description of the approach.
For each contest stands a first prize of $1000.
Deadline May 15.

http://bregman.dartmouth.edu/turingtests/node/72

 

 < 1 2
2 of 2
 
  login or register to react