Our work explores the relation between users and conversational agents from the HCI perspective in an interaction continuum linking humans and agents together. We highlight the need for a common representation space that we name "shared playground". In the shared playground users and agents coordinate through the linguistic notions of competence and performance to reach an "agreement" in order to communicate successfully. The human-agent coordination is possible only if both parties share some preliminary knowledge. we argue that natural language understanding alone is not sufficient to achieve a satisfactory conversation. We elicit the need for level(s) of representation in order to engage the user by ascribing human traits of the agent. We clarify the rise of an Uncanny Valley in conversations and propose possible solutions to mitigate its effects. Finally, we present a set of features to quantitatively describe the eeriness in conversations with the hope to temper distant conversational agents and consolidate closer conversational companions.
Occhiuto, D., Garzotto, F. (2020). Hints of uncanny utterances in a disrupted interaction continuum. In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications VISIGRAPP - (Volume 2) (pp.225-230). SciTePress [10.5220/0009157302250230].
Hints of uncanny utterances in a disrupted interaction continuum
Garzotto F.
2020
Abstract
Our work explores the relation between users and conversational agents from the HCI perspective in an interaction continuum linking humans and agents together. We highlight the need for a common representation space that we name "shared playground". In the shared playground users and agents coordinate through the linguistic notions of competence and performance to reach an "agreement" in order to communicate successfully. The human-agent coordination is possible only if both parties share some preliminary knowledge. we argue that natural language understanding alone is not sufficient to achieve a satisfactory conversation. We elicit the need for level(s) of representation in order to engage the user by ascribing human traits of the agent. We clarify the rise of an Uncanny Valley in conversations and propose possible solutions to mitigate its effects. Finally, we present a set of features to quantitatively describe the eeriness in conversations with the hope to temper distant conversational agents and consolidate closer conversational companions.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
