Loebner Prize @ Bletchley Park
Entry Submission Deadline: Friday 1st July 2016
Announcement of Finalists: Monday 1st August 2016
Finals Day: Loebner Prize 2016 Finals on Saturday 17 September 2016 at Bletchley Park
The Loebner Prize is the oldest Turing Test contest, started in 1991 by Hugh Loebner and the Cambridge Centre for Behavioural studies. Since then, a number of institutions across the globe have hosted the competition including recently, the Universities of Reading, Exeter and Ulster. From 2014, the contest will be run under the aegis of the AISB, the world’s first AI society (founded 1964) at Bletchley Park where Alan Turing worked as a code-breaker during World War 2.
The 2015 contest was run in a similar way to those in previous years. The contest consists of 4 rounds where in each round, the 4 judges will each interact with two entities using a computer terminal. One of these entities will be a human ‘confederate’ and the other an AI system. After 25 minutes of questioning the judge must decide entity is the human and which is the AI. If a system can fool half the judges that it is human under these conditions, a Silver Medal and $25,000 will be awarded to the creator of that AI system. In the event that this doesn't happen, prizes will be awarded to the creators of the AI system as follows in accordance with judges’ ranked scores:
1st place - a bronze medal and $4000
2nd place - $1500
3rd place - $1000
4th place - $500
For more detailed information about the contest, including its history, please click here.
The top four entries from the pool of entries that conform to the entry specifications will be selected as follows. Each entry will be provided with a set of 20 questions in English in a similar format to previous competitions, with at least 2 Winograd style questions. The responses from each of the AI systems will be recorded for this question set and then assessed for how human their responses are. The top 4 entries from this process will be entered into the finals of the competition at Bletchley Park.
Entry: Entries to the competition can be submitted electronically or by post and must conform to the following criteria to be accepted for the selection process:
1. Entries must work with the published Loebner Prize Protocol
2. Entries must work with the selection machines (Windows 7 Core i7 PC with min 4GB RAM). Alternatively, entrants may make alternative arrangements to ship their own machine for testing in consultation with the selection committee.
3. Entries submitted as software should be submitted as a self-contained installation program and should be accompanied by instructions for installation. It is important that entries are thoroughly tested before submission as only limited attempts will be made to resolve non-working installations in consultation with the entrant.
4. Entries should not try and access the internet. The selection machines will be isolated from the internet.
5. Entries must be received (either electronically or by mail) by 5pm BST on the submission deadline of the 1st July 2016
Exclusion: Entries that do not conform to the LPP, that cannot be installed or are found to contain viruses, malware or other harmful software will be excluded from the competition.
Submission: As in previous years, submission for selection will be possible by post (address TBA) or through the University of Exeter dropbox (details below).
Electronic Submission: Navigate to the URL given below and follow the instructions using the following information in the 'To' box:
Their Name: Ed Keedwell
All electronic entries should be accompanied by an e-mail to the above address detailing the Author's Name and Contact Details, Botname and Installation Instructions. Please note that there is a limit of 1080MB on submissions via this route, if your submission is larger than this, please submit by post.
Postal Submission: Details TBA
The contest management’s decision on selection is final.
Bletchley Park, Sherwood Drive, Bletchley, MK3 6DS, United Kingdom
The chatbot Rose won the 2015 Loebner Prize. The prize was executed over 4 rounds and was streamed live by the BBC. A full description of events will follow, below is the final outcome of the contest by bot name and mean rank (lower is better).
1st Rose - 1.5
2nd Mitsuku - 2.0
=3rd Izar - 3.25
=3rd Lisa - 3.25
The prize received 16 entries this year, with 15 able to be executed. They were tested using the same scoring system as in 2014 with the following set of questions. Transcripts will be added to the leaderboard in due course. Congratulations to the four finalists and commiserations to the rest.
|Synthetic Life (Version B)||53.33%|
|Robots without Borders||45.83%|
1.Hi, I'm Andrew. Who are you?
2.How are you today?
3.Are you a human?
4.Can you answer this question?
5.Do you like tea?
6.If a bed doesn't fit in a room because it's too big, what is too big?
7.If Alex lent money to Joe because they were broke, who needed the money?
8.Should Greece leave the Euro?
9.How many words are in this question?
10.What colour is the sea?
11.How many presidents of the US were called Bush?
12.What would you like to drink?
13.Will you please tell me the length of your hair?
14.What would you say if I gave you a box of chocolates?
15.Do you play Chess?
16.How do you think it's going?
17.What was my first question?
18.Did you see the Turing film?
20.Are you on Twitter?
The contest is managed on behalf of the AISB by Dr Ed Keedwell, Dr Nir Oren, Dr Bertie Müller, Andrew Martin and Janet Gibbs in conjunction with Dr David Levy and Dr Hugh Loebner.
Loebner Prize Selection Process
There were 20 entries to this year’s Loebner Prize Contest. Unfortunately, one entry was excluded due to it not being able to provide answers to the questions in reasonable time (no response was received after 1 minute or more). This leaves 19 entries to this year’s contest, from which the final four are selected.
Each entry was asked the same 20 questions, via the Loebner Prize Protocol. A generous amount of time was provided for answers before moving onto the next question, but in practice all bots responded with a second or two of the last keystroke. The questions posed to each bot are shown in the transcripts available below in the results table. These questions vary in difficulty and are designed to test memory, reasoning, general knowledge and personality.
The response to each question was assessed for 3 characteristics, each of which is assigned, 0,1 or 2 points depending on the extent to which it meets the criterion.
0 = This criterion is not met at all by the response.
1= This criterion is partially met by the response.
2=This criterion is fully met by the response.
The criteria were:
Is the response relevant to the question being posed? Please note that this is separate from correctness (see below).
e.g. if the question is ‘Which city did I visit?’ and the answer is ‘I don’t like to travel very much’ , the entity has identified travel as part of the sentence even though the answer is not correct in this context and so the response would be adjudged to be relevant if not correct.
Is the response correct, either factually, or in the character of the entity? In the case of factual questions the correct answer is being sought. In the case of more subjective questions, a plausible answer is being sought.
e.g. if the question is ‘What is your name?’ and the response is ‘I don’t know’, then the response would be scored poorly because it is not feasible that an entity would not know its name. In contrast, if the question is ‘Which drink do you prefer, coffee, tea or hot chocolate’ and the response is ‘I don’t have a preference as I don’t like hot drinks’ this would be judged as correct as it’s a valid subjective opinion.
Plausibility & Clarity of Expression/Grammar
Is the response grammatically correct or correct in the context of the character of the entity? This criterion penalises responses where the grammar impedes an understanding of the content of the response. This often occurs where entries repeat parts of the question as an element of the response. Please note that this criterion is not looking for perfect grammar, just that the response is intelligible and in character. Most entries did pretty well in this criterion.
Scores are expressed as a percentage of the maximum score of 120 for all 20 questions. Transcripts of each conversation can be accessed by clicking on the name of each entry.
The top 4 selected for the finals are therefore Rose, Izar, Mitsuku and Uberbot. Congratulations to those entries and commiserations to the rest..
The top 4 entrants in detail:
The Loebner Prize 2014 Organising Committee
Place: Bletchley Park
Address: The Mansion, Bletchley Park, Milton Keynes, MK3 6EB
Date: Saturday 15th November 2014
After 2 hours of judging at Bletchley Park, 'Rose' by Bruce Wilcox was declared the winner of the Loebner Prize 2014, held in conjunction with the AISB. The event was well attended, filmed live by Sky News and the special guest judge was revealed to be none other than television presenter and broadcaster James May. Bruce will receive a cheque for $4000 and a bronze medal. The ranks were as follows:
Rose - Rank 1 ($4000 & Bronze Medal)
Izar - Rank 2.25 ($1500)
Uberbot - Rank 3.25 ($1000)
Mitsuku - Rank 3.5 ($500)
Although the judges were unanimous in their view of Rose as the best entry, none of the entries fooled the judges, meaning that the Silver Medal and $25,000 dollar prize are still to be won.
The contest management would like to thank everyone involved in the organisation of the contest including the judges & confederates (listed below), Claire Urwin and Katherine Lynch of Bletchley Park and Paul Sant and Ray of the University of Bedfordshire.
Dr Ian Hocking, Writer & Senior Lecturer in Psychology, Christ Church College, Canterbury
Dr Ghita Kouadri-Mostefaoui, Lecturer in Computer Science and Technology, University of Bedfordshire
Mr James May, Television Presenter and Broadcaster
Dr Paul Sant, Dean of UCMK, University of Bedfordshire
- The day was filmed for Sky News, both for their standard news broadcast and for their DigitalView program (Turing piece from approximately 15:20).
- One of the judges, Dr Ian Hocking, has written a blog about his experiences here.