Ok, that was odd. In the interest of full disclosure, I didn’t get past the first few… whatever they were (can’t really call them “words”, after all) before the Grammar Nazi in me started stabbing the inner parts of my ears, making them bleed. I had to leave the page before permanent damage occurred.

Dave Morton - Sep 20, 2013:

Ok, that was odd. In the interest of full disclosure, I didn’t get past the first few… whatever they were (can’t really call them “words”, after all) before the Grammar Nazi in me started stabbing the inner parts of my ears, making them bleed. I had to leave the page before permanent damage occurred.

It appeared to be a… modified… version of the phys.org article. Though I don’t know why anyone would bother to take the time to do that…

Steve Worswick - Sep 20, 2013:

Well here’s the weirdest press release I have seen so far:
http://www.geekgawk.com/2013/09/17/mitsuku-chatbot-has-good-answers-for-the-loebner-prize/

That reads like a lot of my chatlogs.

Found another one:

DVICE: http://www.dvice.com/2013-9-18/chat-loebner-prize-winning-chatbot-now

Here’s ours at AiDreams !

http://aidreams.co.uk/forum/index.php?page=Steve_Worswick_Interview_-_Loebner_2013_winner

http://article.wn.com/view/2013/09/16/Music_site_chatbot_wins_AI_Loebner_contest/#/related_news

I was also featured on BBC local radio:
https://www.youtube.com/watch?v=VbW9b2sGNPc

And on Austrian radio EO1 (German language). Link courtesy to Paul McKevitt. Too much to translate, but this is the most extensive and civilized report I’ve heard yet, starring Hugh Loebner, Richard Wallace, Steve Worswick and several judges. Even the infamous 3rd judge, who was apparently not opposed to the contest but rather seems to be a case of high expectations and zero patience (I have no idea why an AI professor would have such expectations, lest he was poorly informed). I found this report refreshingly honest, from Hugh Loebner’s reasons for having the contest, to Steve saying that Mitsuku’s program has no intelligence of its own.

Don Patrick - Oct 1, 2013:

And on Austrian radio EO1 (German language). Link courtesy to Paul McKevitt. Too much to translate, but this is the most extensive and civilized report I’ve heard yet, starring Hugh Loebner, Richard Wallace, Steve Worswick and several judges. Even the infamous 3rd judge, who was apparently not opposed to the contest but rather seems to be a case of high expectations and zero patience (I have no idea why an AI professor would have such expectations, lest he was poorly informed). I found this report refreshingly honest, from Hugh Loebner’s reasons for having the contest, to Steve saying that Mitsuku’s program has no intelligence of its own.

Is there a translated transcript of this anywhere?

Jarrod Torriero - Oct 1, 2013:

Is there a translated transcript of this anywhere?

Any transcript of this - translated or not - would be more than likely a copyright infringement. The laws in Austria are usually very similar to those in Germany, so I expected that even Radio OE1 isn’t allowed to offer their own transmissions for download as soon as they’re older than one week, and their streaming page suggests that it is indeed the case.

I wonder if anyone would be so kind as to translate the part where Prof. Noel Sharkey is speaking between 26:26 and 26:55. I heard Mitsuku’s name mentioned and am curious to know what he said, as I can’t quite make out the English part over the German translator.

(Copyright in my country makes an exception for the purpose of reporting news. I’m pretty sure that’s part of the Berne convention)

After four judging rounds, to him there was a clear favourite. “One question I asked which was: ‘which is bigger, a small mountain or a large lion?’. Mitsuku was the only machine to get this right. Another machine told me she came from San Francisco. So I asked where she preferred to go for breakfast, upon which she said ‘I like cheese on toast best’.” So it didn’t take any killer questions to tell man from machine.

But wasn’t Mitsuku the one who said that a large lion was bigger? . And what I’m personally curious about: Was this answered through the knowledge base?

Don Patrick - Oct 1, 2013:

But wasn’t Mitsuku the one who said that a large lion was bigger?

From what I could make out of the logs, Mitsuku came out with a generic, “The obvious one” or “How big are we talking here?” but this was only due to the judge typing too many questions at once and not waiting for the answer. Had he given her a chance to answer, she would have said, “In my opinion, a mountain is bigger than a lion.”

However, I wasn’t watching the final round, as I was doing a feature for French TV.

(thanks for the translation though)

Don Patrick - Oct 1, 2013:

And what I’m personally curious about: Was this answered through the knowledge base?

Yes, Mitsuku’s knowledge base contains relative sizes for objects from 0 for sub atomic particles to about 15 for the Universe itself. On my scale, a lion is 5 and a mountain is 7 (it ignores the large and small parts unless the 2 nouns are the same). The response given depends on the difference between the 2 sizes. As this only had a difference of 2 (7-5), it gives out the above comment.

Had the question been something like, “Which is larger, a mountain or a pea?”, Mitsuku would have said, “Are you joking, a mountain is enormous compared to a pea.”, as the difference was so much bigger.

Must’ve been another chatbot then. It’s hard to tell who’s who from the transcripts. But I guess I can conclude that a knowledge base is (or would have been) of advantage for answering the more classic testing questions.

More interviews at Technology Voice, starring Hugh Loebner, Steve Worswick and judge Noel Sharkey, beginning at 30:40.

I think the running order was:
Round1 Judge1 (IZAR)  JUDGE2 (TUTOR) Judge3 (Mitsuku)  Judge4 (ROSE)
round2 judge2 (IZAR) Judge1 (TUTOR)  Judge4 (Mitsuku) Judge3 (ROSE)
Round3 judge3 (IZAR) Judge4 (TUTOR) Judge1 (Mistuku) Judge2 (ROSE)
round4 judge 4 (IZAR) Judge3 (TUTOR) judge2 (Mitsuku) Judge1 (ROSE)