ITRI 616 Trenton Arthur 26518104 Loebner Prize PDF

Preview

Summary

Assignment completed for subject....

Description

The Loebner Prize History Firstly, what is the Loebner prize? In the year 1950 Alan Turing proposed the question of “Can machines think?”. This led to him creating the Imitation Game which referred to this question. The Imitation Game was a test between an interrogator and a computer/machine. The interrogator will ask questions to the machine without any visible contact. For the computer to be regarded as intelligent, the interrogator should not be able to realise that it is a computer/machine answering. It was later adopted as the Turing Test. Dr. Hugh Loebner who was an American inventor together with the National Science Foundation as well as the Sloan Foundation started the Loebner Prize Competition in 1991. This was a contest held every year where scientists/inventors/even ordinary people could sign up for to identify the most “human – like” programs. Loebner had said that he would provide $100,000 and a solid gold medal to the first person who can prove a computer intelligent in the Turing Test by making use of textual, visual as well as auditory components. $25,000 was awarded to the first bot that passes a text-only Turing Test where between $2000 and $3000 for the most human - seeming contestants for the year. In 2017 the prize was a bronze medal as well as $4000. Loebzer Prize winner 2017 In 2017 the winner of the Loezber Prize was Steve Worswick who created his bot/program called Mitsuku. Mitsuku is a three-time winner of the Loebzer Prize in the years 2013, 2016 and 2017. This bot has a female “appearance” and is said to be of the age 18 years old. The bot was first created in 2005 by Steve. Mitsuku makes use of AIML (Artificial Intelligence Modelling Language) technology which allows for the development of artificial systems. It provides the possibility to create human interfaces while still being simple to work with. Mitsuku contains all the files used by Alice (Artificial Linguistic Internet Computer Entity) which was the winner of the Loebzer prize in 2000, 2001 and 2004 and was developed by Richard Wallace. The way Steve has developed Mitsuku is in the way that it is able to source out information for itself if it is unsure of the question. Here is an example, the person who is chatting asked Mitsuku a question, “Are you able to eat a car?”. Mitsuku will then look up any properties of a car and thus the value of what the car is made of is thus changed to metal. Thus, Mitsuku will reply no as this property is not edible. Mitsuku can search for images, if you ask to see a rose, a picture of a rose will be shown by her. By making conversation with millions of users all over the world, Mitsuku can become more intelligent as she learns new meanings to words and stores all the information in logs. There are smaller versions of Mitsuku available for use by Pandorabots which can be used as a service. Mitsuku operates under the username “Pandorabots” on several platforms. The original Mitsuku can be found online.

Mitsuku provides a great step forward in the world of AI as it is a bot/program that can learn from the conversations had to improve on itself. If Steve can implement audible conversations then this will be a step closer to completely fulfilling Hugh Loebner’s goal of having the first bot to be proven fully intelligent using the three components; auditory, visual and textual. Mistuku has also won a chatbot showdown against Siri. This is the original Mitsuku bot http://www.square-bear.co.uk/mitsuku/home.htm. On this page, https://pandorabots.com/mitsuku/ ,is the latest of Mitsuku, one can chat with her on several platforms:      

Facebook Messenger. Kik. Telegram. Skype Messenger. Twitch Group Chat. Android App (Beta).

References Mauldin, M.L., 1994, August. Chatterbots, tinymuds, and the turing test: Entering the loebner prize competition. In AAAI(Vol. 94, pp. 16-21).

Ex er ci se1. 14 Ex ami net heAIl i t er at ur et odi scov erwhet hert hef ol l owi ngt askscancur r ent l ybesol v edby comput er s : 



Pl ayi ngadecentgameoft abl et enni s( Pi ngPong) For a game of table tennis, a robotic arm can be fitted to the paddle which will then make use of sensors to detect where the ball is and thus move the paddle to hit the ball. It can be solved by computers. Dr i vi ngi nt hecent r eofCai r o,Egypt Google is currently in the works of developing self-driving cars which will make use of many sensors to function correctly as well as a road map stored within the car to assist with travelling. It can be solved by computers.



Dr i vi ngi nVi ct or vi l l e,Cal i f or ni a Google is currently in the works of developing self-driving cars which will make use of many sensors to function correctly as well as a road map stored within the car to assist with travelling. It can be solved by computers.



Buyi ngaweek’ swor t hofgr ocer i esatt hemar ket I believe this would be difficult at the current stage of computers even though there are robots out there, the robot needs to know what you want and needs to know what the items look like and where there are exactly stored. I believe at a later stage it may be possible but currently cannot be solved.



Buyi ngaweek’ swor t hofgr ocer i esont heWeb This is solvable by a computer as all that is required is the information of what you want, how many etc. is entered onto the site and stored. All you need to do is pick it up and provide payment details for the transaction to occur.



Pl ayi ngadecentgameofbr i dgeatacompet i t i vel evel There are already computers which can play chess and many other board and card games. So, bridge will be solvable by a computer.



Di scover i ngandpr ovi ngnew mat hemat i calt heor ems This task will not be solvable now. Computers can solve mathematical theorems but being able to discover them will be a challenge as they would need to be intelligent to do so.



Wr i t i ngani nt ent i onal l yf unnyst or y Computers will not be able to write a funny story, they do not know humour so they won’t be able to do so. Jokes can be inputted to the computer but writing a funny story on the other hand would not be possible.



Gi vi ngcompet entl egaladvi cei naspeci al i z edar eaofl aw For a computer to provide legal advice, it would need to know in detail every area of the case within the certain law area. A computer can easily provide quantitative advice but not qualitative.



Tr ansl at i ngspokenEngl i shi nt ospokenSwedi shi nr ealt i me This is solvable by computers.



Per f or mi ngacompl exsur gi caloper at i on Computers would be unable to do so now due to the complexity involved and knowing what to do if an error occurs and many situations....