NEWS: Chatbots.org survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..
Embodied Conversational Agents, or ECAs, have been developed for a wide range of applications. One of the most often reported difficulties is to maintain the user’s attention and interest. Most of the studies report that interaction with ECAs does not last more than a few turns.
To overcome this short interaction pattern, a popular approach is to make ECA more human-like, but recent work suggests that some aspects of human behaviour are more important than others. The theme to be explored in this workshop is that the important aspects are those that
make an agent appear to have social intelligence. The social intelligence hypothesis is that intelligence as we know it
is a result of evolution in an environment where cooperation is key to survival. Animals that live in same species groups, including humans, develop protocols for dealing with intra group pressures. These
protocols require the presentation and recognition of cues that express social relations and any agent, human or virtual, that is to operate in a social context must be able to work with these cues. A key question is what protocols and techniques have evolved in human society, and what
must an Embodied Conversational Agent do to be a recognisably social being?
These issues have been studied before in various disciplines. Reeves and Nass show how humans are sensitive to the medium of a message, not just the message content, and Brown and Levinson use the concept of ‘face’ to model politeness. The aim of this workshop is to draw this
work together by showing how it is applied it to the creation of ECA’s. Multi disciplinary and multi paradigm contributions are welcome. Authors are not necessarily expected to have implemented a system, but the consequences of their paper should be apparent to those who wish to create embodied conversational agents that act in a social world.