Abstract | ||
---|---|---|
Some have argued that human skills and abilities are essential to effective health coaching and cannot be replicated by conversational agents. We sought to understand the differences in interaction patterns between two group of participants. In one group, participants messaged with a wizard-of-oz (woz) conversational health coach. In the other group, participants messaged with an actual health coach who was given the same script as the agent, but encouraged to deviate from the script when necessary. We found that conversational patterns differed between the groups, with longer conversations in the human coach group. However, participants were not more likely to respond to messages from the human coach than the woz agent, and were more likely to proactively message the woz agent than the human coach. We discuss implications for the design of conversational health coaches.
|
Year | DOI | Venue |
---|---|---|
2020 | 10.1145/3334480.3383081 | CHI '20: CHI Conference on Human Factors in Computing Systems
Honolulu
HI
USA
April, 2020 |
Keywords | DocType | ISBN |
Conversational agents, health coaching, wizard-of-oz | Conference | 978-1-4503-6819-3 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Elliot G. Mitchell | 1 | 4 | 5.88 |
Rosa Maimone | 2 | 50 | 9.07 |
Lena Mamykina | 3 | 1124 | 98.05 |