Seminar & Forum
Academic SeminarMachine and Human: A Field Experiment of AI Chatbot Disclosure for Conversational Commerce
- Date
- 2018-11-20 ~ 2018-11-20
- Time
- 09:30 ~ 11:00
- Place
- Building 9, 7th #9708
- Department
- School of Management Engineering
- Manager
- Seulbi Lee , seulbilee@kaist.ac.kr
- Major
- IT Management
We would like to invite you to participate in Management Engineering(ME) Seminar.
1. When: November 20th (Thuesday), 09:30~11:00
2. Where: Building 9, 7th, 9708 lecture room
3. Speaker: Prof. Xueming Luo (Temple University)
4. Topic: Machine and Human: A Field Experiment of AI Chatbot Disclosure for Conversational Commerce
5. Research field: IT Management
* Lecture will be delivered in English.
Abstract:
Empowered by AI, chatbots are surging as new technologies to facilitate firm-initiated communications and conversational commerce. However, little is known from the customer side regarding these questions: (1) Will the disclosure of AI chatbot identity negatively impact customer purchases? (2) How would customers perceive A.I chatbot as service agents facilitating purchase decisions? And (3) how to mitigate the negative impact of chatbot disclosure? We collaborate with a large fintech company to conduct a randomized field experiment. Our data analyses find that undisclosed AI chatbots are as effective as proficient human workers in engendering customer purchases. However, the disclosure of chatbot identity before or after the machine-human conversation negatively impacts the purchase conversion by 56% to 83%. We also explore the underlying mechanism. While the subjective survey data suggest that chatbots are perceived as less capable of providing product knowledge and sentiment empathy when compared to human agents, the objective voice-mining data on the conversation content suggest no such gap. These results imply that the negative impact of chatbot disclosure may be driven by human’s subjective outgroup bias against machine. Fortunately, such negative impact can be allayed or recovered by prior experiential learning of AI or a disclosure right after making the purchase decision—a disclosure timing strategy that enables customers to learn the full experience of interacting with the smart machine. These findings have profound implications for tackling AI technology adoption challenges, for consumer targeting, and for company communications.
1. When: November 20th (Thuesday), 09:30~11:00
2. Where: Building 9, 7th, 9708 lecture room
3. Speaker: Prof. Xueming Luo (Temple University)
4. Topic: Machine and Human: A Field Experiment of AI Chatbot Disclosure for Conversational Commerce
5. Research field: IT Management
* Lecture will be delivered in English.
Abstract:
Empowered by AI, chatbots are surging as new technologies to facilitate firm-initiated communications and conversational commerce. However, little is known from the customer side regarding these questions: (1) Will the disclosure of AI chatbot identity negatively impact customer purchases? (2) How would customers perceive A.I chatbot as service agents facilitating purchase decisions? And (3) how to mitigate the negative impact of chatbot disclosure? We collaborate with a large fintech company to conduct a randomized field experiment. Our data analyses find that undisclosed AI chatbots are as effective as proficient human workers in engendering customer purchases. However, the disclosure of chatbot identity before or after the machine-human conversation negatively impacts the purchase conversion by 56% to 83%. We also explore the underlying mechanism. While the subjective survey data suggest that chatbots are perceived as less capable of providing product knowledge and sentiment empathy when compared to human agents, the objective voice-mining data on the conversation content suggest no such gap. These results imply that the negative impact of chatbot disclosure may be driven by human’s subjective outgroup bias against machine. Fortunately, such negative impact can be allayed or recovered by prior experiential learning of AI or a disclosure right after making the purchase decision—a disclosure timing strategy that enables customers to learn the full experience of interacting with the smart machine. These findings have profound implications for tackling AI technology adoption challenges, for consumer targeting, and for company communications.
Contact : Lee, Jisun ( jisunlee@kaist.ac.kr )
