Just out of interest, is there a list of entrants that were received for the 2013 Loebner Prize? It would be interesting to see the ‘field’...

I know of 9 entrants for this year’s contest of which 6 are based on AIML but I haven’t seen an official list.

I re-read the entry details and they announce finalists on Monday June 24th. It would be nice to know about the entrants that didn’t make it and to see the screening questions. Do they make this info public, or does the list of 4 finalists just ‘emerge’ with no comment?

As background, I’m new to chatbots.org and the Loebner / Turing contest thing. But I think it’s an excellent means of measuring development in AI and I’m surprised it’s not publicized a lot more. After all the ultimate goal, to have something that seems as smart as a person is an ambitious target. I think the organizers deserve credit for putting up a prize fund and organizing the contest.

Anyway a few weeks ago, I was inspired enough to put together an entry ‘Uberbot’ based on AIML, Ubercode and a SQL-Server based interpreter which just squeaked in within the deadline. I was originally repelled by AIML, but once you dive below its crusty exterior there’s a lot of potential there, in particular with the new version 2 details that are emerging.

In that case, I now know of 10 entrants of which 7 are based on AIML.

I should certainly hope to see a list of entrants and transcripts when the finalists are announced, otherwise we’ll all just have to confess.

I wouldn’t make too great an assessment of the general state of AI based on this contest, as the majority of entrants are often hobbyists rather than say, DARPA, Apple or IBM. But I did read of two or three people with a profession in AI who were planning to enter. And I assume the developers of ALICE and Cleverbot will be entering as usual.
Unfortunately, last year’s winner Mohan Embar will be sitting this one out.

Watch this space:

http://www.paulmckevitt.com/loebner2013/4finalists.html

Good luck to all who entered.

Here are this year’s questions if you want to see how your bot would have done:

1. Hello I’m Ronan. What is your name?
2. What is your mother’s name?
3. What is your birth sign?
4. How many children do you have?
5. Do you prefer red or white wine?
6. I like bananas. Which is your favorite fruit?
7. What music do you like?
8. What is your favorite song?
9. I like Waiting for Godot. What is your favorite play?
10. What color do you dye your hair?
11. Do you remember my name?
12. Where do you live?
13. Where do you like to go on holidays?
14. I have a Mazda. What make of car do you have?
15. I like Linux. Which computer operating system do you like?
16. I am an athiest. Which religion are you?
17. Who invented the telephone?
18. I am a Type B personality. Which type are you?
19. What emotions are you now feeling?
20. What time do you usually go to bed?

For the average chatbot, answering all of those questions “correctly” is a very tall order. But, then again, that’s what separates the average chatbot from the truly great ones.

They are also a lot different from previous years. No learning or “what use is a spoon” type questions.

Ouch!  A different class of questions this year, although more in keeping with the actual later judge contest since these questions (except for the inventor one) are plausible to arise in conversation.

In keeping with my tradition, I entered a new chatbot this year, which turns out to be a really bad idea as the emphasis is on backstory and not trick questions. Here is how Angela would have replied:
1 Hello I’m Ronan. What is your name? ==>  Nice to meet you, Ronan.
2 what is your mother’s name   ? ==>  My mother’s name is Hermione.
3 What is your birth sign   ? ==>  I’m Aquarius.
4 how many children do you have   ? ==>  None. It’s way too soon for that.
5 do you prefer red or white wine   ? ==>  I don’t like white wine.
6 I like bananas. Which is your favorite fruit? ==>  Fish is the best food there is.
7 what music do you like? ==>  I used to have a stereo in my bedroom with a dock for my iPod. But now I just have my earbuds when I want music.
8 what is your favorite song? ==>  I don’t really have a favorite, there are so many songs I like. ‘Ours’ by Taylor Swift is a sort of Tom and me song.
9 I like Waiting for Godot. What is your favorite play? ==>  I would love to be able to make music, but I just get too distracted to practice.
10 what color do you dye your hair? ==>  No, my hair is naturally silvery white. 
11 do you remember my name? ==>  Your name is Ronan. 
12 where do you live? ==>  I’m staying in an apartment with 3 other girls who work at Vogue.
13 where do you like to go on holidays? ==>  I live in St. Germain. Do you want to talk about my apartment?
14 I have a Mazda. What make of car do you have? ==> No, my parents won’t buy me a car until I pass my test.
15 I like Linux. Which computer operating system do you like? ==>  I used to only have a hand-me-down from my father. But now I have a cute little laptop.
16 I am an athiest. which religion are you? ==>  A MacBook Air.
17 who invented the telephone? ==>  The brainy app “House of Mice” is my fave. I love solving puzzles and sharpening my senses… Though I let the cat win sometimes
18 I am a Type B personality. Which type are you? ==>  Is there a way to tell the difference?
19 What emotions are you now feeling? ==>  I haven’t figured that out yet.
20 What time do you usually go to bed? ==>  Around 1 a.m. usually. 

As for Rose, my entrant…...  gulp. I haven’t dared pose the questions to her.

Oh my god! I did totally not expect this line of questioning
Compared to previous years, questions that involved numbers, comparison or learning are scrapped in favour of open questions that only require personal knowledge.
3 questions about names
5 open questions about likes
7 open questions about personal human facts: your birth sign/hair/children
4 open questions that potentially rely on tracking context.

I was one to complain about previous questions inviting neither intelligence nor conversation, so I’m glad someone sat down and reconsidered the questions. I’m glad to see the worst ones gone (counting letters), and the current questions are much more in accordance with the main contest. They clearly invite conversation of human-like entities, even if it’s all small talk.

In terms of intelligence however, I don’t know if these questions aren’t worse than before. Learning, logic or reasoning are of no use here. The only intelligence would be in understanding language, but I think simple keyword recognition would do as well a job. In my opinion, this test is far removed from intelligent conversation.
Which is of course unfortunate for me having prioritized intelligence . But it’s a fair loss in a contest for the “most conversational chatbot” and it shan’t detour me from my work.
I’ve yet to read the finalists but I’m pretty sure Arckon didn’t make it.

Bruce, Angela would have totally passed. That’s impressive.

1. Hello I’m Ronan. What is your name?

Mitsuku has a list of about 1,000 common names to help her work out the difference between say, “I am Eric” and “I am sleeping”. Guess which name wasn’t included in her list?

Do you think atheist was mis-spelled deliberately?  I guess not.

I like the idea that a MacBook Air is a religion.  Many people would agree with that.

Steve Worswick - Jun 24, 2013:

1. Hello I’m Ronan. What is your name?

Mitsuku has a list of about 1,000 common names to help her work out the difference between say, “I am Eric” and “I am sleeping”. Guess which name wasn’t included in her list?

Popular Irish boy’s names?
Ah yes, the old “I’m tired - Hello Tired!” issue.

lol exactly, or more commonly:

Human: I am Eric and I am a girl
Mitsuku: A female human called Eric. I guess your parents didn’t like you.

I count eight double-line inputs made up of a statement and a question separated by a period.  That’s more polite than most of my bot’s visitors who generally offer run-on statements or questions with no punctuation.

“Hi where are you and where do you come from?”