[Freddy]

Here you go : Bruce Wilcox Interview - Loebner 2014 winner

Don, I included your question in the interview.

We're getting a nice collection of interviews now.

[Don Patrick]

Thanks! Though I had guessed the answer by now . Always a pleasure to read about ChatScript's abilities.

[8pla.net]

What was Don's question?


FIXED TYPO:  Oops!  Edited out an unneeded "the" in the question.   

[Freddy]

It was this one...

Don : Did you use any A.I. techniques like semantic analysis, emotion analysis, learning or inference, and how much use were they in this contest ?

Bruce : I use all of these, but they were of no real use in this year's Loebner.

[8pla.net]

For the sake of A.I. technology discussion only, without disparaging any person, company or product... In the interview, what Bruce referred to as a "permanent household bot" may not qualify as that, in my opinion.  When any household bot is dependent on a remote server and/or the cloud for 99% of it's intelligence, it usually means it is running bare bones on a thin client in the household.

Before I forget... Congratulations Bruce!
...
...
...
...
...
...

[Art]

Interesting but I fail to see how the judging criteria was carried out.

[Freddy]

8pla, are you saying the household bot has to be a self contained unit ?

Art, I have yet to look at the transcripts, but will do shortly.

[Freddy]

Well I tried to read the transcripts but some of it is a bit confusing. For some we're getting one half of a conversation it seems, then a massive response all in one go, presumably the other half. Then there are long sections where the response is a single letter.

I can understand the complaints people have made about how the protocol did not work very well this year. I gave up.

Seems like James May was judge one though.

[Don Patrick]

Loebner transcripts are always confusing. Last year they showed both stretches of conversation side by side and you basically had to guess in what order things were said. They clearly tried to improve on that this year with speech bubbles (I like that), but it is of little avail: as long as people don't press return inbetween responses it's impossible to tell who said what when.
The complaints were about the bots receiving the messages in pieces because of mid-sentence network delays. It looks like they cleaned up the transcripts and pieced the responses back together, so things were actually worse than they look now.

[squarebear]

Don is correct. The transcripts on the site have been amended for readability and are not an accurate representation of the actual conversations.

A more accurate version of Rose's transcript is here:
https://www.chatbots.org/ai_zone/viewthread/1867/P15/#19063

What was actually said to Mitsuku is here:
http://www.square-bear.co.uk/mitsuku/loebner2014.htm

Yes, James May was judge 1.

[8pla.net]

Freddy asked, "8pla, are you saying the household bot has to be a self contained unit?"

If by "the" you mean any A.I. product, well I would have to say, "No", since I support companies and academia in our A.I. community.  I fully support commercial success for leading A.I. designers.

However, if you mean, for the sake of A.I. discussion, then...  In Isaac Asimov's short story "I, Robot", the main robotic characters, are self contained units.  So, science fiction has long since inspired me to prefer a self-contained household bot, over mass produced automated mechanisms.

[8pla.net]

On another forum, https://www.chatbots.org/ai_zone/viewthread/1867/P15/#19063, Hugh mentioned, "a second hub".  I own an old hub and a newer switch.  I use the hub ONLY for emergencies when I run out of ports on my switch.   A hub may be better than nothing but not much better.  Frankly a hub can be lousy compared to a switch.   And a small switch is not expensive.

Now, as a former Loebner Prize judge, I am NOT criticizing the Loebner Prize Competition in any way, but I am going to crack a joke to increase the entertainment value of this discussion.  Hugh has always been nice to me.  So I don't mind helping out Bletchley Park in a friendly manner.  Based on the quote by Hugh, I would simply suggest, that the solution for next year, may be to use a switch instead of a hub.   

Seems from the following quote, that  those poor chatbot entries got blasted with a broadcast of every conversation in the contest simultaneously.   Thank goodness they were machines.  If they were human that might have drove them insane...  ( That was just a joke! )

Technically speaking, hubs operate using a broadcast model and switches operate using a virtual circuit model. When four computers are connected to a hub, for example, and two of those computers communicate with each other, hubs simply pass through all network traffic to each of the four computers.

Switches, on the other hand, are capable of determining the destination of each individual traffic element (such as an Ethernet frame) and selectively forwarding data to the one computer that actually needs it. By generating less network traffic in delivering messages, a switch performs better than a hub on busy networks.

Quoted from: http://compnetworking.about.com/od/hardwarenetworkgear/l/blfaq011.htm