IEEE Technology and Society Magazine - September 2020 - 23

easily foreseen and farther afield
but conceivable?
The combination of overly deterministic and overly positive thinking steers both the applicants and
the funders away from considering
the harm that a given project might
make possible. We read every day
about destructive uses of the Internet in society - and specifically
about huge platforms such as Facebook and Amazon - that are surely
not what their inventors and early
developers had in mind. But that's
likely in part because no one asked.
After all, science fiction writers were
able to imagine some of the current
craziness even in the 1940s.
In this way, the National Science
Foundation selection process -
and many similar decision-making
processes in the world of technology - is the opposite of "algorithmic
self-governance." Only the voices
of those who will benefit - or who
maybe will benefit - are clearly
heard. The voices of those who will
or might be harmed or who will be
left out are ... left out.
It's long past time to ask researchers and technology developers to
think in advance about the social
harm their work might cause down
the road and not simply to enumerate good things they directly anticipate from their research. That would
not, of course, prevent all possible
harm - the only way to do that
would be to put a permanent stop to
all scientific and technological progress. But it might give society a head
start in controlling or ameliorating
some of that harm.
Some humility about how little
can be predicted and some willingness to predict what can be predicted even when it doesn't make for
good selling points for a proposal -
these could make our headlong scientific and technological progress
at least somewhat less dangerous.
The National Science Foundation is
SEPTEMBER 2020

in a particularly important, indeed
unique and critical, position to en--
courage such thinking about technology and society.

A More Holistic View of
Technology and Society

Governments, for example, must
mandate balanced and nuanced
analysis. But so too must every
s t a keholder. We should not let
either too much optimism or too
much pessimism cloud our vision:
the goal of social impact analysis
should be scientific realism, so that
we as a society get more of the best
and less of the worst of influential
new technologies.

This is not a call to swing from technophilia and technothusiasm to
technophobia. Exciting opportunities are afoot. But as public bodies,
corporations, and individual
consumers, our enthusiasm
should be tempered. To
any sufficiently important
innovation, there is bound
The goal of social impact analysis
to be a dark side. There is
a dark side to pain-reducing
should be scientific realism, so
and life-saving medications;
that's why they are tested
that we as a society get more of
and why they have sidethe best and less of the worst of
effect warnings and why
some require a doctor's preinfluential new technologies.
scription. A more complete
view of technology's social
impacts would replace the
search for positive impacts with a
Here are two strategies that can
search for all conceivable impacts -
help:
positive and negative. The main
1) Aggressively seek diverse perplayers in today's advance of techspectives from different stakenology - both the private compaholders. The most challenging
nies and the governments - need
circumstances are often comto lead by example.
pletely unexpected because we
In the specific case of autononever even made an effort to look
mous cars, a dangerous barrier
for them. These are what Donald
to such analysis may be, perhaps
Rumsfeld, the former U.S. Secresurprisingly, the ubiquity of levelstary of Defense, memorably
of-driving-automation frameworks.
called "the unknown unknowns."
While levels-of-driving-automation
One reason an array of diverse
frameworks can surely be useful to
perspectives is useful is that it
inform discourse and policy, they
allows for a more thorough set of
are at their core, a set of engineering
social impacts (positive and negspecifications. As such they do not
ative, easily foreseen and less
necessitate consideration of broader
discernable) to be surfaced. In
goals, including social and humanisevery field, there are implicit
tic ones. For example, as researchassumptions about "how things
ers have noted, by their very nature,
work." Typically, those assumplevels-of-driving-automation frametions hold true - that is why they
works may rule out more creative
have been adopted and internalforms of cooperation between vehiized. But they can still backfire
cles and their drivers, i.e., humans
under new conditions. One way
and their machines [9], [10].
∕

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

IEEE Technology and Society Magazine - September 2020

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - September 2020