I
m
a
g
e
b
y
J
a
s
o
n
Te
s
t
e
r
G
u
e
r
r
i
l
l
a
F
u
t
u
r
e
s
/
F
l
i
c
k
r
Algorithmic Bosses,
Robotic Colleagues:
Toward human-centered algorithmic
workplaces
We already know algorithms can make our lives and our work
more efficient, but how can we go beyond that to create trustworthy,
fair, and enjoyable workplaces in which workers can find
meaning and continuously learn?
By Min Kyung Lee
DOI: 10.1145/3013498
Algorithms pervade every aspect of our daily lives, more so today than ever. Algorithms decide what we see online, from Google search results and Amazon ads to Facebook News Feed and Netflix recommendations. But they’re not just online. Algorithms are increasingly taking on the roles of bosses, managers, and coworkers.
Companies use algorithms to find candidates for jobs, evaluate the performance of customer
service agents, and connect patients with physicians. Governments use algorithms to
determine when to patrol certain areas and screen immigration applicants.
Algorithms have actually helped
give rise to an entirely new type of workplace and workforce. In emerging digital work platforms, algorithms flexibly
and efficiently match customers with
service providers; you’ve experienced
this type of service if you’ve ever used
Uber or TaskRabbit. Robots are slowly
but steadily entering the workplace,
and working right alongside people.
They deliver goods in hotels, care for
patients and older adults with support
from nurses or caregivers, and assem-
ble machinery alongside employees in
small businesses. Because of these re-
cent changes, along with an industry
push for efficiency, the social impacts
of algorithmic technologies are no lon-
ger negligible.
My goal is to design such algorithms
to better support human values, moti-
vations, and unique capabilities. We
already know algorithms can make
our lives and our work more efficient,
but how can we go beyond that to cre-
ate trustworthy, fair, and enjoyable
workplaces in which workers can find
meaning and continuously learn?
This is an opportune and exciting
time to address the aforementioned
question. Historically we have seen
many first-generation technological
design choices leave long legacies,
even if those choices were less than
optimal. The design principles we use
to create “smart on-demand trans-