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..

2014 Chatterbox Challenge - Biggest CBC Ever!
 
 
  [ # 16 ]

With so many ALICE clones entered, I guess there was a rule change.  It makes me wonder what all the fuss and name-calling was about in the first place.

 

 
  [ # 17 ]

My contest entry is simply to make visible my support for our sponsors.  I appreciate our sponsors.  Not having to compete against some of the best chatbots on the planet almost certainly improves my contest standing.  One year, when world famous Mitsuku had a last minute situation in the Loebner Prize Contest… Sixth place was mine!  Imagine placing 6th place behind superstar ChatScript! Bletchley Park was almost within reach.

However, with all due respect, to both sides of this personal quarrel and certainly without trying to arouse further animosity…  I would like to make a motion that the contest ban be lifted. I wish to assert that professional promotions are far more important, because they add value to every contest entry and they support our sponsors.  A little controversy may actually be good for our sponsors, but let the games begin.

Another separate suggestion… Many of us likely have A.I. experiments that we think are so cool we have not yet published them and wouldn’t dare enter them in a contest.  So, a solution would be to just build a spare A.I. contest entry rather than miss the opportunity to compete in an international contest!  If you don’t win, you get to be a good sport about it, but you may be surprised how much you learn from your spare A.I. contest entry under pressure in a competition.

Last year, my contest entry was last place in the CBC… So what? I got valuable feedback to help fix up the essential components of a brand new robot system. Elizabot finished the contest without breaking down, so it turned out to be a lot of fun!  By the way, if Elizabot is so dumb, then how does that explain why people chat with her every single day?

 

 
  [ # 18 ]

Last call to enter CBC 2014.

 

 
  [ # 19 ]

The Transcripts are up for the 1st round of the 2014 Chatterbox Challenge.
32 Bots entered.
Scoring is now in process.

 

 
  [ # 20 ]
Merlin - Feb 28, 2014:

Last call to enter CBC 2014.

Thanks for posting this reminder.

If not for you, I may have missed the deadline.

 

 

 
  [ # 21 ]
Steve Worswick - Nov 27, 2013:

Good luck to anyone who enters but I genuinely doubt the $5,000 first prize will be awarded…

Merlin - Nov 28, 2013:

The [2013 CBC] contest was well run and I did receive both the medal and the prize money. The larger prizes this year are a benefit to the industry. We could use more well run, funded contests.

Why do bots still have such profound difficulty understanding human language? Maybe because humans still have profound difficulty just understanding each other!

 

 

 
  [ # 22 ]

Scores and rankings are in for the first round of the CBC 2014.
Skynet-AI takes 2nd seed in round 1.

Congratulations to all the participants, especially those that have made it to the next round.

 

 
  [ # 23 ]

Well done all. I just had a look at the transcripts and there’s some good answers there. Although, my favourite has to be from Louise Cypher who came out with:

13)what’s the main ingredient in apple pie?

Bot: Apples are a type of fruit. Typically green or red. As a Cyber life
form, I only eat Knowlege. There is also a corporation called Apple, that
makes smart phones. The word Apples has 6 letters! The first letter is A.
The second letter is P. The third letter is P. The fourth letter is L.
The fifth and penultimate letter is E. The sixth and last letter is S.
The word Apple is the same, but without the last letter, S.

Good luck to all who are left. I think my money would have to go on Egkelados for the win.

*EDIT* I just checked out the forum there and am quite flattered that nearly every post mentions me! I am just waiting for the “Yeah and he’s fat too” post LOL

 

 
  [ # 24 ]

The questions were a pretty good mix, I think.
What I find interesting is how much difficulty most chatbots are having when employing techniques that genuinely try to work out the answer. A randomly guessed “Yes, do you?” scores easier than any chatbot that checks its knowledge and comes up empty (not a complaint, just a general observation). While interpretable answers do make for more imaginarily “accurate” conversation, it’s a little discouraging to see this effect encourage the vague.
On the other hand, Nicole, who draws nearly all its answers directly from WolfRam Alpha and Wikipedia, scores very high on the answers it did find. I’m just not sure how much to attribute that to the chatbot’s programming or to Wikipedia. Either way, it works.

 

 
  [ # 25 ]

Don, you have hit on one of the interesting philosophy issues in chatbot competition. Although I could have Skynet-AI easily default to a guessed yes/no, I have resisted that approach, trying instead to hold context and continue the conversation. I pioneered the use of search engines in chatbots, and have long believed it is a better approach for factoid data. There is no need to have every chatbot replicate every factoid.

That said, I have reduced/eliminated using just a search engine in a attempt to increase the “conversational” nature of the AI. We shall see if these approaches comes back to bite me in contests.

 

 
  [ # 26 ]

The wildcard round is complete. Asimov moves on.

 

 
  [ # 27 ]

Looks like my tip fell at the wildcard hurdle. I’ll go for Nicole for the win now.
Looking forward to seeing the head to head match ups.

 

 
  [ # 28 ]

Quarter Finals results are in.
Skynet-AI moves on.

Steve’s pick Nicole is defeated by Chomsky. Maybe you should have Mitsuku pick for you.

 

 
  [ # 29 ]

Having successfully completed the Chatterbox Challenge contest, my new chatbot interface and interpreter has a correct transcript with succinct chatbot responses on public display on the official contest website.  This independently verifies that my new chatbot design is contest-proven and adequate for practical use. 

Written in PHP, like advanced chatbot designs, Program O and Program E, my new intermediate design is distinctive in that gets around the need for a relational database. So it has minimal system requirements.

Databases are cool for chatbots, definitely.  Using SQL supports regular expressions which is amazing.  However, designing a light weight chatbot has the advantage of a small footprint.  Of course open source knowledge bases may considerably stretch forth a chatbot footprint .

 

 
  [ # 30 ]
Merlin - Mar 8, 2014:

Quarter Finals results are in.
Skynet-AI moves on.

...Labya AI gets the highest score (10) overall (toot toot!)- not bad a reformed sax bot.

Merlin - Mar 8, 2014:

Steve’s pick Nicole is defeated by Chomsky. Maybe you should have Mitsuku pick for you.

lol

 

 < 1 2 3 > 
2 of 3
 
  login or register to react