J.A. Jacko (Ed.): Human-Computer Interaction, Part II, HCII 2011, LNCS 6762, pp. 533–536, 2011. © Springer-Verlag Berlin Heidelberg 2011 Older User-Computer Interaction on the Internet: How Conversational Agents Can Help Wi-Suk Kwon 1 , Veena Chattaraman 1 , Soo In Shim 1 , Hanan Alnizami 2 , and Juan Gilbert 2 1 Department of Consumer Affairs, Auburn University, Auburn AL 36849, U.S.A 2 Human-Centered Computing Division, School of Computing, Clemson University, Clemson SC 29634, U.S.A {kwonwis,vzc0001,szs0029}@auburn.edu, {hanana,juan}@clemson.edu Abstract. Using a qualitative study employing a role-playing approach with human agents, this study identifies the potential roles of conversational agents in enhancing older users’ computer interactions on the Internet in e-commerce environments. Twenty-five participants aged 65 or older performed a given shopping task with a human agent playing the role of a conversational agent. The activity computer screens were video-recorded and the participant-agent conversations were audio-recorded. Through navigation path analysis as well as content analysis of the conversations, three major issues hindering older users’ Internet interaction are identified: (1) a lack of prior computer knowledge, (2) a failure to locate information or buttons, and (3) confusions related to meanings of information. The navigation path analysis also suggests potential ways conversational agents may assist older users to optimize their search strategies. Implications and suggestions for future studies are discussed. Keywords: Conversational agent, older users, Internet, interaction. 1 Introduction Only 35% of seniors aged over 65 years use the Internet in the United States [1]. The primary cause for this underrepresentation of older users is that hardware and software interfaces have not been designed to accommodate their needs [2]. Conversational agents (CAs) are animated embodiments in computer-mediated environments that respond to users through verbal and non-verbal communication [3]. CAs facilitate ‘intelligent’ retrieval of information based on users’ individual needs. Applying findings from research on pedagogical agents [4], we propose that CAs have a significant potential to improve older users’ interactions in computer-mediated environments. The purpose of this study was to identify potential roles a CA can play to enhance older users’ Internet interactions in consumer environments, through a qualitative study employing a role-playing approach with human agents. The researchers assume that real interactions with humans can be used to model interactions with CAs.