A vision for human-machine mutual understanding, trust establishment, and collaboration

Abstract

Human-machine interactions are likely to require synergistic multidisciplinary research efforts for supporting a paradigm shift towards collaborative-oriented use cases. An essential aspect of collaboration is trust and in order to establish it there is need for human-machine mutual understanding (HMMU). We argue that achieving HMMU will require evolving from an approach that reduces human factors as uncontrollable environmental elements, to one that repositions human emotions not only as a central part of an integrated control paradigm, but also as interpretable and steerable through appropriate information flows and mutual learning cycles. On the strategic decision-making side, we argue conflict resolution will require anticipating multiple trade-off situations that include human factors. On the operational level, symbiotic human-machine cognitive architectures should embed detected human emotions as inputs in shared machine control models. Trust measurements will play the role of mediating task coordination by pinpointing and dynamically composing appropriate situation-aware interaction protocols. In addition to a vision for HMMU, this paper proposes a multidisciplinary research strategy that attempts to unify the isolated efforts of different communities. The proposed vision is contextualized within a high-level research roadmap to support near and long-term activities in HMMU.