AI Zone Admin Forum Add your forum

NEWS: survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..

Unofficial AISB Loebner Prize 2017 Finalist selection
  [ # 31 ]

I’m both pleased that Steve won, and rather bummed that the rest did not. but I’m not one of those “everyone gets a trophy because you’re ALL special” kind of folk, so I’ll just leave it at that. smile

BTW, if anyone who was actually there might have found a small green thumb drive, I think the owner is looking for it.


  [ # 32 ]

I dont think the green thumb drive got there. It arrived at at the college and was never transferred to Andrew.


  [ # 33 ]

Thanks guys. I’ve just got back home and am a little tired but I’ll write my usual report in the next few days. The results were:
1 - Mitsuku
2 - Midge
3 - Uberbot
4 - Rose


  [ # 34 ]

Oversight: A cautionary tale.

I’ve always enjoyed the conflicting meanings of the word “oversight”.
#1 the action of overseeing something.
#2 an unintentional failure to notice or do something.

You might wonder what happened to Rose. Well, it’s like this…

Dave Morton graciously created a mechanism for Rose to meet the new loebner protocol. It was a webserver on a USB stick, that relayed AI webpage to PHP webpage to Chatscript Server. It fully met the protocol requirements and had the extra features relating to callbacks that ChatScript utilizes. A week or two before the contest, it was broken by an unannnounced and unneccessary change in the protocol, but Dave fixed it and we were good to go.

AISB had promised Loebner finalists help in meeting the new protocol. This eventually consisted of releasing a bridge that converted new to old protocol. This was not released until a week before the tournament, and it was incorrect. Dave Morton debugged and fixed it at the start of the week.

I fedex’ed the USB stick to the given address, emailing the loebner prize committee that it was due thursday by noon, and providing the tracking number. Tt arrived and was signed for on thursday morning. It then “disappeared”, in that it never
reached it’s recipient Andrew. Nor did he or anyone else inquire as to its whereabouts on Friday. So Saturday morning in response to my email prompt, I was told “I’ve received your zip file sent via WeTransfer, but not the USB stick.”

In addition to that stick, which was my primary entry, I sent a backup zip that didn’t have the full interface capability, and instead used the new and relatively untested bridge to the old protocol. The University’s dropbox was down, so I was told to send it directly to Andrew’s dropbox, which in turn refused my file multiple times, so I shipped it via a 3rd party transfer service. That was successful and what was entered into the tournament. And in the first two conversations I saw, Rose responded with “huh?” roughly half the time. Why? I have no clue. Maybe the logs will tell me when they send them to me. But since Rose doesn’t normally do that, I have to presume something went wrong outside of Rose.

It brings to mind the word oversight. How well their oversight of this tournament worked, and which definition I apply.


  [ # 35 ]

Webcast of the event:

For the sake of discussion…

Creation Date: 2017-09-13T14:04:48Z

How is a new domain name created on the 13th, ready to webcast by the 16th?

Bruce, if it was signed for, then it sounds like they had your thumbdrive all along.

Dave, Congratulations on your first Turing test contest, joining Team ChatScript.



  [ # 36 ]

Congratulations Steve, and also Merlin who seems to have done quite well for a first time in the finals.
I am saddened to hear that Rose’s standing was again due to a technicality rather than her actual performance. At least she was responsive once and a while in each round for a brief impression. From Bruce’s story I don’t know how to fix any of it for the future, but changes within two weeks before the contest are not okay by my standards. For the selection round, I also had difficulty getting zip files across to Andrew, or even for emails to reach his university inbox, ending in me contacting him through a personal channel on the last possible day. It would seem university-hosted services are not to be relied on for this contest, and the voluntary nature of the organisation seems to limit their activity to the weekends.
Much appreciation to Dave, who has apparently done a lot of legwork behind the scenes to compensate.


  [ # 37 ]

Congratulations Steve and Rich. I can go back in my mind to 1995 when Dr. Thomas Whalen wrote an essay about his experiences while losing the Loebner Prize Contest that year. He “reassessed” what it meant to be human. It seems like every year somebody has pause to reflect on it all. I know what it’s like. You work all year, getting ready for what your bot missed last year. It’s disappointing for sure.

If I believed in the anomaly, I’d expect it to occur from the rapid application development of bots, under “perpetual beta testing” and there’s always “one more thing.”

Thanks and congratulations to Dave.


  [ # 38 ]

Congratulations to Steve for winning the bronze - well done!

Uberbot had his own share of problems - for two of the four rounds he made no reply or only replied ‘undefined’. Nir was very helpful and we traced it back to some Javascript errors which were quickly fixed. So Uberbot was on line for rounds 3 and 4. The coding errors were my responsibility, and could have been found if we did a full dummy run in the morning, or the previous day. On a positive note I met Richard, Steve, Andy and the others, and had a look round Bletchley park and the National Museum of Computing. A good weekend.


  [ # 39 ]

All that follows is for the sake of discussion… Under what circumstances would it be right to demand the unofficial results be declared invalid?  For example, once a contestant pays for certified delivery, and it is signed for, from that point forward, shouldn’t the competition take full responsibility?

Would circumstances such as theft suggest a contestant’s certified delivered thumbdrive be located, or redelivered at the competition’s expense,  and that the competition be run all over again?

As a solution proposal, I recommend consulting Dr. Huma Shah, an expert, who is efficient and extremely organized at international Turing test competition management.



  [ # 40 ]

CONTEST RULE: One entry per contestant.

1. Contestant A makes one entry.
2. Contestant M makes one entry.
3. Contestant A entry joins Contestant M entry.

Does the CONTEST RULE disqualify Contestant A entry and Contestant M entry?


  [ # 41 ]

Loebner trivia:

If you want to win the Loebner prize, it helps if your last name begins with “W”.

Wallace, Weintraub, Whalen, Wilcox, Worswick: 15 wins
All other names: 12



  [ # 42 ]

hey, that’s pretty coool!!!! Congrats to all winners and finalists Mitsuku, Rose, Uberbot & Midge!!! Let’s continue to improve our bots and even be better next year!!


  [ # 43 ]

As is the case with parcels / packages that are signed for here in the USA, The person who signs for the article is responsible for it until it is handed over to the addressee or proper intended recipient.

One would think it would be an easy exercise to determine who signed for the thumbdrive.

Yes, at this point is is moot but certainly an unfortunate happening.

Why can’t the contest be kept to a simple set of rules instead of having so many bells and whistles?

I find it also odd that although participants have an entire year to prepare, test and retest their entries, then have them “fail” at the 11th hour. Hmmm.

Just an observation and certainly not meant to ruffle anyone’s feathers (if anyone has feathers).


  [ # 44 ]

The path from the reception desk to the recipient is always precarious. If Bruce’s experiences were like mine, then Bruce’s email was probably blocked and I don’t know Andrew to be at the university on Thursdays. Either way several things were cut too close and I’m amazed FedEx even arrived within two weeks.

Art: The bridge between old and new protocol wasn’t released until a week before the contest and was buggy, and the full network setup isn’t tested until the actual day as far as I know. That said, the protocol itself had been available since January and was barely looked at until I brought it up two months ahead.

Everyone who I’ve told the outcome of last year said that contest should have been redone. However, I understand that press, promotion, judges and materials aren’t easily rescheduled and there’s no guarantee that there won’t be another problem then. In a way the contest is rerun, every year. In practice it just means that Mitsuku was really only up against one opponent in the finals this year and last year.

I hope all problems can be found and improved upon for future use.


  [ # 45 ]

In every contest that I have ever entered, once the Judges do the final voting, the results are set in stone.
The logistics of rerunning a contest are just too daunting.

What we can do is provide input and feedback to insure future contests are better than past ones.


 < 1 2 3 4 5 >  Last ›
3 of 7
  login or register to react