Chatbots have received paramount attention in the last year. E-commerce retailers increasingly incorporate text-based AI chatbots in customer service. However, it is not uncommon for these chatbots to make mistakes during today’s customer-chatbot encounters. Based on a mixed- method approach, this research delves into consumer-chatbot interactions in order to discover how cognitive (expectations) and behavioral attributes (usage intention) are affected by erroneous conversational outcomes with AI chatbots. The analysis also takes into account pre- and post-purchase customer journey phases and age of the users. In the context of retailers selling consumer electronics, the findings indicate a three-dimensional approach to expectations and usage intentions, described by chatbot-related, user-related and environment-related characteristics. Erroneous interactions lead to a decrease of high expectations and usage intentions for the same chatbot and affect users in their post-interaction attitude towards other chatbots in the short term. These effects change when a longer time frame is considered. Additionally, prior attitudes towards AI and chatbots provide insights into variations within the sample (n=16).
I chatbot hanno ricevuto un'attenzione fondamentale nell'ultimo anno. I rivenditori di e- commerce incorporano sempre più spesso chatbot IA basati sul testo nel servizio clienti. Tuttavia, non è raro che questi chatbot commettano errori durante gli attuali incontri tra clienti e chatbot. Basata su un approccio di tipo misto, questa ricerca analizza le interazioni tra consumatori e chatbot per scoprire come gli attributi cognitivi (aspettative) e comportamentali (intenzione d'uso) siano influenzati dagli esiti errati delle conversazioni con i chatbot IA. L'analisi tiene conto anche delle fasi del customer journey pre e post acquisto e dell'età degli utenti. Nel contesto dei rivenditori di elettronica di consumo, i risultati indicano un approccio tridimensionale alle aspettative e alle intenzioni d'uso, descritto da caratteristiche legate al chatbot, all'utente e all'ambiente. Le interazioni errate portano a una diminuzione delle aspettative e delle intenzioni d'uso elevate per lo stesso chatbot e influenzano gli utenti nel loro atteggiamento post-interazione verso altri chatbot nel breve termine. Questi effetti cambiano quando si considera un arco di tempo più lungo. Inoltre, gli atteggiamenti precedenti verso l'IA e i chatbot forniscono indicazioni sulle variazioni all'interno del campione (n=16).
To Err Is Not Only Human – Exploring User Expectations and Usage Behavior in Al-Chatbots: Effects of Errors on User Experience in Customer Service
VASSILIADOU, NIKOLETTA
2022/2023
Abstract
Chatbots have received paramount attention in the last year. E-commerce retailers increasingly incorporate text-based AI chatbots in customer service. However, it is not uncommon for these chatbots to make mistakes during today’s customer-chatbot encounters. Based on a mixed- method approach, this research delves into consumer-chatbot interactions in order to discover how cognitive (expectations) and behavioral attributes (usage intention) are affected by erroneous conversational outcomes with AI chatbots. The analysis also takes into account pre- and post-purchase customer journey phases and age of the users. In the context of retailers selling consumer electronics, the findings indicate a three-dimensional approach to expectations and usage intentions, described by chatbot-related, user-related and environment-related characteristics. Erroneous interactions lead to a decrease of high expectations and usage intentions for the same chatbot and affect users in their post-interaction attitude towards other chatbots in the short term. These effects change when a longer time frame is considered. Additionally, prior attitudes towards AI and chatbots provide insights into variations within the sample (n=16).È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/3337