The Loebner Prize

In this episode, we focus on the Loebner Prize competition and explore the question of what it means to sound human in conversations. Our host Kyle Polich talks to Bruce Wilcox, a veteran Loebner Prize participant, and Charlie Moloney, a former judge from the 2017 competition. Both Bruce and Charlie were at last year's Loebner Prize competition, and they each share their stories and experiences from either the judge's perspective or chatbot creator's perspective. Bruce, who is a four-time winner of the Loebner tournament, shares his ideas behind the chatbot engine he developed, called ChatScript, as well as what makes a good conversational agent. In addition, we hear from Charlie about what it's like to judge text conversations with humans and computer programs, and whether judges could ever be fooled into thinking a computer is human.

The Loebner Prize is an annual competition in the artificial intelligence community to build the most human-like chatbot. It has been going on since 1991. You may not have heard of the Loebner competition. The format of the competition is based on the Turing Test, proposed by British mathematician and computer science pioneer Alan Turing in his famous 1950 paper, titled "Computing Machinery and Intelligence." Human judges at the Loebner Prize conduct conversations, via a computer terminal, with a chat program and with a human being, known as “confederates” during the competition. After each round of conversations, the judge has to decide which is a human and which is a chatbot. They then have to rank each of their conversation partners from most human-like to least humanlike. The chatbot with the highest average ranking wins the competition.


(Charlie Moloney)[] (Bruce Wilcox)[]