April 25, 2012
Student Tom Joyce and his Chatterbot Earn 4th Place in Chatterbox Challenge
Tom Joyce, a science, math and technology major studying through the college’s distance-learning program, and his robot, Vira, at left, a downloadable chatterbot (a.k.a. “chatbot” or “bot”, a chatterbot is a robot designed to simulate small talk and answer simple questions), earned a top spot among robot masters in the prestigious international competition, Chatterbox Challenge 2012.
Last year, Joyce was in 17th place; he said he is “pleased and surprised to have moved all the way up to fourth, which shows that my work is moving forward and will provide a solid foundation for new and exciting innovations in the future.” Vira outchatted 34 other English-speaking bots vying for a cash prize.
Joyce, who lives in New Jersey, is working with mentor Jacqueline Bishop. In addition to being a student, he is a statewide certified technology-education teacher whose advanced source code is featured in college textbooks published by Prentice Hall. His work primarily is in advanced computer programming and web-application development and database and network administration, including assembling very expensive blade-server hardware and installing network operating systems. He also holds two industry certifications in computerized accounting through general ledger and payroll. Vira was a final project for his course in discrete mathematics.
“The impact of A.I. (artificial intelligence) is multifaceted,” Joyce said. “I have worked in artificial intelligence most of my career. I can tell you that A.I. will continue to impact both the present and the future. For example, press an elevator button, and the doors open immediately, as though the elevator was waiting there for you personally. That is ‘reinforced learning,’ which analyses traffic patterns in and out of the building in order to serve people more efficiently, as well as to reduce wear and tear on the equipment. This not only saves money on equipment maintenance, but it increases safety too.”
He added, “When you have an EKG at the hospital and it gives a diagnosis by machine, which is confirmed by a physician, that is A.I. called ‘EKG interpretation,’ fighting our No. 1 one killer, heart disease. Medical A.I. alerts physicians of medical emergencies, and is saving lives right now.”
A third example Joyce offered is the GPS system, which “speaks” to give driving directions. “Drivers may not realize that avoiding wrong turns and unnecessary driving tasks, such as stopping and reading street signs, can save gas money and increase safety by reducing accidents caused by these avoidable distractions,” he said. “Artificial intelligence often saves money in many ways. A.I. is everywhere -- in military contracts, government agencies, law enforcement, medical devices, digital-signal processing, data acquisition, building engineering for controller area networks in office buildings and airports and forensic reverse engineering for low-level analysis of malware and spacecraft control.”
Vira has 150,000 records in her robot brain, reported Joyce, and she uses a number of algorithms, not just for A.I. stimulus/response, but also for speech synthesis and voice recognition. Vira controls her own computer graphics for 3D modeling and animation to very rapidly map phonemes to visemes.
"It has been my pleasure to work with Tom, and to learn about his creative endeavors in the use of technology,” said mentor Bishop. “He is making, and has made, a positive impact with his successes in the field of artificial intelligence development."
The CBC began in 2001 as an annual contest for chatbots, with cash prizes. Minimal restrictions are placed on the type of technology used in the creation of the bot. Botmasters from across the globe submit their chatbots for evaluation and competition. Every entered bot is asked a series of questions and scored on its responses by independent judges. The top 10 bots move to a final round where an additional series of questions is posed to the finalists. The winner is selected by the judges as the bot who has scored the highest from among the finalists. A number of chatterbots who have entered the competition in the past have become the foundation for commercial technologies.
“A.I. is at least 100 years old,” explains Joyce. “A good place to find A.I. is on museum computers, some of which, although very old, are still in service today. Finding ways to enjoy A.I. is important because designing it usually involves pushing the limits. Luckily, building a machine that simulates human intelligence and talks right back to you is an experience most find very entertaining and rewarding.”
Talking about his great strides at the CBC, Joyce is unequivocal. He said, “My wish came true.”
Artificial intelligence chatbots, also known as chatterbots or conversational agents, developed in a series of three stages over the past 60 years.
The First Wave
The late Alan Turing, known as “the father of artificial intelligence,” conceived of the talking computer in his 1950 paper, “Computing Machinery and Intelligence.” For his famous test, Turing imagined an artificial intelligence that communicates in natural language through a text-based medium, such as a teletype. This year in the U.K. The Turing Centenary will celebrate his 100th birthday.
In 1966 MIT Prof. Joseph Weizenbaum became the first botmaster, or chatbot author, when he created the famous ELIZA program. ELIZA was the first A.I. to apply the concept of stimulus-response pattern recognition to natural language understanding. ELIZA was also the first bot to employ conversational logging as a means for the botmaster to review and refine the bot.
Dr. Hugh Loebner began sponsoring the first real-world Turing Test in 1991. To the surprise of many, the winner of the first contest was based on the ELIZA psychiatrist program.
In 1994, Michael Mauldin created a bot named Julia in an online MUD environment. He coined the term “chatterbot” to describe his conversational programs.
The Second Wave
The advent of the World Wide Web marked the beginning of widespread access to chatterbots. By exposing their bots on the web, botmasters collected a huge amount of conversational log data to help them improve the quality of the bots. Better and faster computers led to the development of large knowledge bases for bots.
Dr. Richard Wallace launched the free software ALICE project in 1995. ALICE led to the development of the open AIML standard for creating chatbots. An alphabet soup of AIML interpreters and servers appeared.
The first commercial chatbot companies, Neuromedia and Virtual Personalities, were launched in the heady early days of the dot-com boom.
In the late '90s, two prominent websites emerged to index and promote chatbot projects and companies. These were the Simon Laven page, and Marcus Zillman’s Botspot.com.
The Third Wave
Today chatterbots have been adapted to nearly every ecological niche on the internet. Bots appear on Web pages, in instant messaging and respond to email and forum posts. They can be found in Second Life, online games and social networks such as Facebook and Twitter. Bots support marketing and advertising and are used in education.
New technologies for automated learning vastly reduce the time and effort needed to create convincing bots. At the same time, a series of new commercial opportunities have opened for bots and their botmasters.