IEEE Technology and Society Magazine - September 2022 - 41

Connectivity and competency in SSH
Connectivity is a critical dimension for SSHs.
Institutionalized and folkmedical SSHs reflect forms
of distributed cognition wherein human and nonhuman
agents share information and complete
specified tasks, for example, monitor, regulate, categorize/classify,
and decide [28]. Whether explicit
or implicit, the interaction between the agents in
an SSH requires epistemic dependence, with each
agent trusting the output or information provided
by another. While in small, well-defined social networks,
agents can gain direct experience about
their respective abilities through interaction (i.e.,
transactive memory), larger social networks require
monitoring and regulating beyond the capacities of
agents, leading to an increasing reliance on information
that can support indirect exchanges between
agents, such as reputational information (e.g., gossip
and formal titles) [29]. Within a social network, status
markers like prestige can therefore be substituted
for direct knowledge when evaluating the competency
of agents [30].
Collective competence and EPAs. In healthcare,
there is growing recognition that HCPs' competencies
need to be operationalized and directly
measured. A competency reflects a distinct set of
knowledge and skills associated within a particular
area of practice, that is, domain-specific intelligence.
These competencies can be assessed and used to
judge a learner's progress against a set of clearly
defined milestones. For instance, the Royal College
of Physicians and Surgeons Canada defines competencies
such as medical expert, communicator,
collaborator, scholar, leader, and health advocate.
Each competency can be further subdivided into a
set of EPAs, essential tasks within a specialty or subspeciality
that clinicians' must be able to perform
independently. If A/ISs are used to supplement or
replace human agents, they must also be assessed
in terms of whether they can perform EPAs autonomously
and determine how much human supervision
is required.
No practitioner is an island. Along with individual
competency assessment, there is growing recognition
that healthcare communities are defined
by collective competence. In healthcare, collective
competence reflects the intelligence of a set of HCPs
that are (directly or indirectly) coordinating their
activities to deliver healthcare within a social network.
For instance, in surgery, operations require the
September 2022
coordination of nurses, surgeons, and anesthetists
as well as supporting technologies that monitor the
patient and provide diagnostics. Thus, when using
an A/IS, we must identify and assess the mechanisms
supporting the distribution of trust within an SSH.
Distribution of trust within SSH. As early adopters
of expert systems and A/ISs, healthcare delivery systems
have been one of the first areas of practice to
consider the design criteria required to facilitate
trust [31]. Contemporary SSH frameworks for trust in
A/ISs address these issues in terms of both professions
[32], [33], [34] and users [35], [36]. In communal
settings where agents are undifferentiated (i.e., equivalent
and interchangeable), social goods such as
prestige and trust are likely to be evenly distributed
among community members. However, in social
networks defined by highly specialized knowledge,
trust will likely be asymmetrically distributed. Furthermore,
in contrast to hierarchical structures which
are defined by a single criterion used to judge the
products of group members, healthcare is defined by
interprofessional or interdisciplinary teams creating
a heterarchical structure defined by multiple assessments
and evaluation criteria [37]. This heterogeneity
can create challenges for determining the accuracy
and reliability of each agent within the SSH leading to
the dependency on prestige.
Regardless of the specific criteria, interdependency
between agents (trust) should be allocated
in proportion to the accuracy and reliability of an
agent's attitudes, beliefs, and behavior within a
specified domain. In humans, transactive memory
reflects an agent's knowledge of another's skills
and competence developed through repeated
interaction. Transactive memory is essential to the
successful completion of tasks [38]. In addition to
transactive memory, the fidelity of communication
channels (e.g., noise) and differences in the knowledge
of the agents might result in errors or systematic
biases, respectively. Moreover, as the social network
becomes larger, and multiple agents must coordinate
to perform specialized tasks, the probability of
error must increase.
IEEE Technology and Society Magazine - September 2022

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - September 2022
