Wochat Chatbot User Experience Summary (bibtex)
by Carla Gordon, Jessica Tin, Jeremy Brown, Elisabeth Fritzsch, Shirley Gabber
Abstract:
A team of 5 interns at the USC Institute for Creative Technologies interacted with 5 of the 6 chatbots; IRIS, Sammy, Sarah, TickTock and Joker. Unfortunately no one in our team could get the 6th chatbot, pyEliza, working. We found that there were certainly some chatbots that were better than others, and some of us were surprised by how distinct each bot felt from the others. One member commented on how they felt as though each different chatbot had an individual “voice” so to speak. Others were surprised by just how much of a “personality” the bots seemed to have. Most members of our team cited IRIS as their favorite, in terms of being capable of producing naturalistic conversation, with Sammy taking a close second. However, only one member of the team was able to interact with Sarah and TickTock, but that member cited TickTock as a capable conversation partner, and Sarah as being the best bot on a number of measures including appropriateness of responses and overall conversation cohesiveness. Therefore, perhaps if more members had been able to interact with Sarah and TickTock they may have ranked higher. Lastly, Joker was by far our least favorite, with whom no member of our team was able to have anything resembling a naturalistic or even cohesive conversation.
Reference:
Wochat Chatbot User Experience Summary (Carla Gordon, Jessica Tin, Jeremy Brown, Elisabeth Fritzsch, Shirley Gabber), In Proceedings of the 2016 IVA: WOCHAT Workshop, Zerotype, 2016.
Bibtex Entry:
@inproceedings{gordon_wochat_2016,
	address = {Los Angeles, CA},
	title = {Wochat {Chatbot} {User} {Experience} {Summary}},
	url = {http://workshop.colips.org/wochat/documents/ST-281.pdf},
	abstract = {A team of 5 interns at the USC Institute for Creative Technologies interacted with 5 of the 6 chatbots; IRIS, Sammy, Sarah, TickTock and Joker. Unfortunately no one in our team could get the 6th chatbot, pyEliza, working. We found that there were certainly some chatbots that were better than others, and some of us were surprised by how distinct each bot felt from the others. One member commented on how they felt as though each different chatbot had an individual “voice” so to speak. Others were surprised by just how much of a “personality” the bots seemed to have. Most members of our team cited IRIS as their favorite, in terms of being capable of producing naturalistic conversation, with Sammy taking a close second. However, only one member of the team was able to interact with Sarah and TickTock, but that member cited TickTock as a capable conversation partner, and Sarah as being the best bot on a number of measures including appropriateness of responses and overall conversation cohesiveness. Therefore, perhaps if more members had been able to interact with Sarah and TickTock they may have ranked higher. Lastly, Joker was by far our least favorite, with whom no member of our team was able to have anything resembling a naturalistic or even cohesive conversation.},
	booktitle = {Proceedings of the 2016 {IVA}: {WOCHAT} {Workshop}},
	publisher = {Zerotype},
	author = {Gordon, Carla and Tin, Jessica and Brown, Jeremy and Fritzsch, Elisabeth and Gabber, Shirley},
	month = sep,
	year = {2016},
	keywords = {Virtual Humans}
}
Powered by bibtexbrowser