From tools to friends: where is the borderline?

Adaptivity in human-computer interaction has been focused on the knowledge, plans, interests, and preferences of the user and has not yet considered extrarational factors such as the user's attitude, personality, or emotional state. Recent research findings prove, however, that humans tend to assign personalities to interfaces, even when they are based on traditional techniques like graphics or natural language and do not include any kind of anthropomorphism, and that the usability of these interfaces is affected from the compatibility between their personality and the user's personality. In addition, it appears that interfaces may better serve the user if they recognize the user's changing attitudes (such as doubt or uncertainty) and respond appropriately. If people already perceive computers based on the 'graphical tool' metaphor as 'having a personality', this perception will most probably increase in that interfaces will evolve towards the new 'assistant' metaphor. Agent interfaces that are installed on the more recent commercial products or are demonstrated as research prototypes show some 'implicit' (and probably unplanned) personality which is potentially perceived as being very 'friendly' by some users, but might be met with disapproval by others. In the present interfaces, and even more in the future world of agent interfaces, it is therefore reasonable to assume that adaptivity should extend from the rational aspects of the mental state that were considered so far to personality and emotions, so as to show not only a friendly but also a warm and empathic interaction. In proposing this Workshop, we start from Carbonell's claim that "Understanding personality traits is important for generating expectations about probable behaviours" and from the statement that "Emotion is not regarded anymore as an undesirable consequence of our embodiment that must be neglected, but as a necessary component of intelligent behavior that of fers a rich potential for the design of artificial systems, and for enhancing our interactions with them" . We wish to promote acquisition, in the UM community, of theories about personality and emotions that have been developed in different domains (such as cognitive science, 'believable agents', conversation, entertainment and so on), and investigation of how they might influence adaptation in HCI. We assume that utilization of these theories in our community will contribute to improving human-computer interaction and will produce, at the same time, an advancement of the 'believable agents' field through discovery of new problems and development of new methods