AI Loebner Prize

2023 Winner – Mitsuku

The last winner of the Loebner Prize in 2023 was the chatbot Mitsuku, created by Steve Worswick. Mitsuku changed its name in late 2020 to Kuki as it was just easier for all. You can chat with Kuki here https://chat.kuki.ai/chat.

Kuki has a history of performing well in the competition. Kuki was first entered in the Loebner Prize in 2013 and won the award for the most humanlike A.I. as a conversational program that year. The chatbot was also entered into the competition in 2014 and 2015, placing fourth and second respectively. In 2023, Kuki outperformed its competitors significantly. In the qualifying round, Kuki scored 90 percent, with the second-place contestant trailing behind at 78 percent. In the final, Kuki scored 1.25, where the best score is a 1 and lower is better, while the second-place bot scored 2.25.

Kuki’s strengths lie in its ability to mimic human conversation using artificial intelligence markup language, a method known as “pattern matching”. It excels in natural language processing and can carry on conversations in a manner that often doesn’t feel like you’re interacting with a bot. It can handle a wide range of topics and offer relevant responses, even when it doesn’t fully understand the user’s input. Instead of constantly asking for clarification, Kuki often provides a more generic response that keeps the conversation flowing. It also makes an effort to inject humor into the conversation when possible.

Kuki has an array of features that make it entertaining and informative. It can provide information on a variety of subjects, from global warming to music charts. It can predict the weather, provide information on local movies (in the UK), deliver your daily horoscope, and even tell you about historical events that occurred before your birth. Despite its strengths, there are areas where Kuki could improve. It sometimes struggles with maintaining context across multiple iterations and lacks the ability to effectively handle sensitive topics, especially emotions and relationships.

Loebner Prize

The Loebner Prize was an annual competition in artificial intelligence that awards prizes to computer programs considered by the judges to be the most human-like. The competition was started in 1990 by Hugh Loebner in conjunction with the Cambridge Center for Behavioral Studies, and it was based on the Turing Test, a test of a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human.

In the competition, independent judges would engage in a text conversation with the computer programs and rate their performance. The judges did not know which entities they were interacting with — they could be either human or chatbots. The goal for the chatbots was to convince the judges that they were the humans. The program that was most successful at persuading the judges that it was human would win the competition.

The original scoring system for the Loebner Prize was based on the Turing Test, which involves fooling judges into thinking that the computer program they are interacting with is a human. However, the exact scoring methodology varied from year to year, and more recently, the competition focused not just on fooling judges, but also on the quality of the conversation.