Page 127 - Artificial Intelligence for the Internet of Everything

Trust and Human-Machine Teaming: A Qualitative Study 113

this limitation of the current study by using structured vignettes to manip-
ulate features of technology to carefully investigate the nuances of various
technologies. This will be particularly true for the social-interaction con-
struct, as not all technologies considered by the participants had this feature.
The six dimensions of the teammate-likeness model were all invoked in
the explanations for why the technology was (or could be) viewed as a team-
mate, though not equally. Among the a priori set of dimensions, agency was
most common followed by communication richness and synchrony. This
shows the importance of viewing the technology as possessing some level
of decision authority for a human to view the technology as a teammate.
Furthermore, the findings for interdependence correspond to the manage-
ment literature in terms of the importance of interdependence among indi-
viduals on a team. These are clearly important features for humans when
considering the teaming relationship with technology. While the current
results are interesting from a research perspective, care should be taken so
that decision authority and interdependence with automated systems is done
only when it has been carefully considered along with the potential limita-
tions of such technologies. Human-machine teams may prove to be more
effective than either humans or technology alone; however, great care must
be taken in the design and implementation of technology in the workplace
to avoid overreliance on reliable technology, as such, overreliance can result
in negative consequences if and when the technology (or the human) makes
mistakes (see Onnasch et al., 2014 for a review).
Relationship-building was the least common explanation provided by
participants. It is possible that the affordances provided by the existing tech-
nologies did not allow for relationship-building. While many of the technol-
ogies listed offer interactive features, many of these technologies lack the
capacity to develop relationships. This dimension may be more relevant
for more advanced future technologies. As noted above, this limitation
could be addressed in future research by structuring the types of technologies
considered (i.e., surveying participants with a wider gamut of technology
affordances).
Interestingly, an unexpected dimension, humanness, was the most com-
mon response for what makes (or what should make) a technology be
viewed as a teammate. The mere notion of a teammate may invoke anthro-
pomorphic perceptions—engendering comparisons to a more traditional
human partner. What remains unclear is whether or not the humanness
dimension was a conglomeration of the other dimensions noted in the
Autonomous Agent Teammate-Likeness Model (Wynne & Lyons, 2018).