Page 123 - Artificial Intelligence for the Internet of Everything

Trust and Human-Machine Teaming: A Qualitative Study 109

I think for it to become more of a teammate, it would have to do far more than it
does already. While it seems very personable, I know what kind of “data” it can tell
me, and as far as that goes, it’s not too personal. To be more of a teammate, it
could recommend places to eat out of nowhere, randomly talk to me about things,
chime in on conversations and just generally exhibit more human-like behavior
(Communication richness)
… To be a teammate it would need to “understand” better. That is, instead of me
anticipating it, it would need to anticipate my needs. It’s smart, but still lacking
sometimes (Synchrony)
I consider it more of a tool. It would have to become either an actual human being,
or an AI that was so human like you couldn’t tell it wasn’t, in order for me to con-
sider it a teammate (Humanness)

6.3 RESULTS

Four-hundred and nine participants (68%) reported that they believed their
technology was tool-like versus teammate-like. The remaining 32%
reported the relationship as more teammate-like. Forty-one percent were
home technologies, 31% mobile technologies, 15% navigation aids, 3%
automotive, 3% robotic systems, and 7% were classified as “other.” The
technologies were further broken down based on brands. Twenty-two per-
cent were Amazon products (Alexa, Echo), 15% were Apple products (Siri,
iPhone), 11% were Google Maps, 6% were Androids or Google Assist, 5%
were Google Home, 3% were Nest, and less than 1% for Tesla and iRobot
each. An additional 36% was classified as “other.” These classifications and
brands were categorized a priori.
As shown in Fig. 6.1, reliability was the primary trust antecedent men-
tioned by participants. This was followed by predictability, helped solve a
problem, and proactively helped, in order of frequency. The remaining trust
antecedents were mentioned fewer than 50 times by the participants. There
were no notable differences in trust antecedents based on whether the par-
ticipants viewed the technology as a tool versus as a team member.
As shown in Fig. 6.2, humanness was the most noted reason for viewing a
technology as a team member. This was followed by agency and commu-
nication richness.
However, given the imbalance between participants who noted a team-
mate relationship versus a tool-like perception, as opposed to reporting the
absolute frequency of the dimensions by teammate versus tool perception,
we also reported the percentage of the dimension reported by the participants.