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..

Bragging rights and maybe a couple of bucks
 
 
  [ # 106 ]

Finishing up the formatting of the movie conversations and noticed a big gaffe. The original transcript for my conversation with Mitsuku was overwritten somehow.  I went back and had another conversation, but I would have to say the first conversation was better. So Steve when you read this, please send me a link to the original transcript. So, these will be up in a few, lets give everyone a chance to review and update ant gaps they see in the transcripts, then we will tally the final scores and issue a decision. And god help us if its a tie, because I just noticed in the rules that there is no taking that possibility into consideration.

VInce

 

 
  [ # 107 ]

I’ll try to find the original chat transcript and email it to you.

I noticed the “holiday” comment on the transcript page. Mitsuku only failed because it was spelled as “hoilday” which she doesn’t understand and so fell back on a default answer. Had it been spelled correctly, she would have answered it.

 

 
  [ # 108 ]

Hi Vince,

I just had a quick check and cannot find the original log. I get many visitors and have nearly 3000 logs from this week alone.

If you go to Mitsuku on the PC you used to talk to her and say, “What is my id?”, she should give you a code like A72635CGE87. If you post that here, I can hopefully find your log.

 

 
  [ # 109 ]

Client id is 89915afb7e19e718.  And again I have to apologize, I simply copied a transcript into notepad and saved what should not have been saved, where it should not have been saved. Twice actually. Actually I thought the holiday (Even with the miss-spelling) was a success as I used it referencing Halloween and that was how Mitsuku took it. If you get a chance to see comedian Trevor Noah, do it. His take on the difference between American English and how the rest of the world speaks English is absolutely hysterical.

Ok guys, Im going to give anyone else who wants to provide a link to primary transcripts the rest of the day to do so, I will link to them, and (after enough sleep and a proper amount of decent coffee) I will finish calculating the scores and tomorrow we have a winner. I cant thank you enough for participating, I had a blast, thanks for your patience through our many little problems


Vincent Gilbert
CTO RISOFTDEV inc

 

 
  [ # 110 ]

And thank you Vince for your dedication in running this contest. I know from first hand that it takes up a lot of time and I salute you for that sir.

Unfortunately, the client id you were given is a new chatlog and so I cannot retrieve it. However, I remember you saying “I was expecting something a little more robotic” in another log and so I will search for that.

 

 
  [ # 111 ]

I found your original transcript on movies and have just emailed it to you.

 

 
  [ # 112 ]

Vince, I just got in. I will send you the logs tomorrow morning East Coast Time.

 

 
  [ # 113 ]

Thanks Steve, it turned out to be a lot of work, I wish the robotics club had been able to judge (those kids are pretty amazing) but it was also a lot of fun, and really the credit goes to everyone who participated.  The questions submitted were crazy difficult, and all the entries did well, particularly in the conversations. By contrast, go back just a few years and look at the types of questions which were being asked, the progression of the field is readily apparent.

Merlin, I’ll hold off on the scores until everyone who wants to provide backup transcripts has done so.

Vince

 

 
  [ # 114 ]

Just noticed that Izars interog transcript was incorrect. This was not a reflection of the questioning, it occurred when I was preparing it for the web. Izars transcript runs bottom to top, we want it to go top to bottom, and somewhere in the lack of sleep fueled translation 2 of the questions were left off.  This has been corrected.


Vince

 

 
  [ # 115 ]

Ok, the scores are in. I have to tell you that I sort of agonized over this because we included subjective scoring in the total score. If we or someone else does it again, I think the scoring should be broken up so that the bots are judged in each of the 4 categories separately. If your bot is a dark brooding overload of the universe, and someone else cracks jokes, it probably is not fair that they gain an advantage by adding the humor score to the total. In this case, it didn’t affect the outcome. According to the strict interpretation of the rules, Mitsuku had a clear advantage in the number of questions she hit in the logic category and so came out on top. The subjective scores would not have affected the outcome. I’m going to announce the scores here but I should have the complete scoring sheets and my notes up in a bit.

Mitsuku 77
Skynet AI       60
Izar       58
Laybia 58

Thats right a tie for third

Louise Cypher 54
Johnny 46
Aidan 31


So about the judging. I did my best to maintain some reasonable level of continuity in how I scored things like originality and humor, but in the end it is subjective. So if you have a complaint please list your complaint in the space provided here—> ▢  Please write legibly. See I thought that was funny, others… maybe not so much wink Bottom line was to have a showcase, and that I believe we did that and I believe everyone did great. There are some special awards that will go up noting certain bots who did things that I thought were pretty incredible, even though they did not win, place, or show.  Our now famous .JPG special awards at that. Genuine imitation gold colored pixels.

Vince (Or whats left of him)

 

 
  [ # 116 ]

Apologies for my absence.  My day job consumed me for this last week.  Vince, I agree the bottom-up transcript logs are hard to copy/paste from, and for future endeavors, I’ve updated my http://www.appsentience.com/btchat interface with a top-down transcript, so you won’t have issues anymore!  Did I miss any other follow-up items?

I also had to jump over to see Mitsuku scream!  Oh my - Happy Halloween Steve!  Izar already has fangs so I didn’t need to modify the character any on the Android interface for Halloween.  One of these days I need to get around to putting the character on the web interface like everyone else.

And thank you Vince for all your hard work!

And congrats Steve on winning this round! 

 

 
  [ # 117 ]

To be honest, I fully expected to win the Bragging Rights contest and it still puzzles me why the rest of you bothered to enter tongue wink

Seriously though, well done to all entrants and I’ll say it again, well done to Vince who ended up running this contest single handed. While his crew were abandoning ship, Captain Vince stood on the deck and regained control.

It was a great contest and was good to warm Mitsuku up to defend her crown at the Loebner’s in a couple of weeks. Thanks again Vince. Same time next week? wink

 

 
  [ # 118 ]

Next week? You just made me regret not having the strength to tease you about only my only kidding about the prize money LOL

Ok the score sheets and the badges for various awards are up along with some of my notes. I also have some notes that I made on how to do the contest better if it does run again, although more than likely not next week. When I have time (hopefully this weekend) I will put up some additional information and may even have time to pretty up these results pages which are quite frankly…pretty ugly.

Thanks to all for entering, in addition to being a lot of fun I think there was some good science here.

Vince

as for Steves post above…thats why the contest was called

BRAGGING RIGHTS

well played sir, well played wink

 

 
  [ # 119 ]

Cool jpegs! *tries to find a way to make it his chatbots.org sig*

I just looked at the score sheets and found Mitsuku scored nothing at all for:

Human: John ate a pizza. What did he do?
Mitsuku: Sounds yummy. He ate a pizza.

Shurely worth at least a 1?

Oh and yes, the post above was strictly in line with the contest name. grin

 

 
  [ # 120 ]

Personally, I’m a firm supporter of having forum sigs, even for sites like this. Sadly, mine is not the power, or we would have the option to have them.

 

‹ First  < 6 7 8 9 10 > 
8 of 10
 
  login or register to react