IEEE Technology and Society Magazine - December 2021 - 52

Soft Law Governance of Artificial Intelligence
AI Insurance: Risk
Management 2.0
Lucille Nalbach Tournas
School of Life Sciences, Arizona State University,
Tempe, AZ 85281 USA
 ArtificiAl intelligence (Ai) allows a computer
program to supplement or, in many cases,
replace the role of human decision-making or action
[1]. This often includes gathering and analyzing data
to make more efficient and/or accurate decisions.
While some AI is fairly simple, other AI applications
rely on sophisticated machine learning (ML) that
allow machines to learn and improve from experience
without being continually reprogrammed by
humans [2].
AI has become ubiquitous in most industries and
technologies, including smartphones, automobiles,
healthcare, home appliances, and major machinery
[3]. As it continues to develop, no sector will be left
untouched and there are unique risks associated
with AI applications and tools. Many of which will
be unknown a priori to decision and policy makers
as they shape the regulatory framework. Specifically,
AI may leave organizations liable for physical
harm, data breaches, property or brand damage,
and business failure without an adequate system
to catch and rectify these repercussions [4]. The
combination of fast-paced, ever-changing technology
with new and potentially unknown risks makes
effective regulation challenging. In this way, AI can
be viewed as a " wicked " problem for policy makers
who wish to regulate the technology and the tools
that it enables [5]. In planning and policy, a wicked
problem is one that is difficult or impossible to solve
because of incomplete, contradictory, and changing
requirements that are often difficult to recognize
[6]. At the same time, AI is rapidly evolving, making
Diana M. Bowman
Sandra Day O'Connor College of Law, Arizona
State University, Phoenix, AZ 85004 USA
it difficult for legislatures to pass laws that successfully
govern the technology in the medium- to longterm.
This pacing problem known as " legal lag " is
further compounded by the breadth of applications
that the emerging technology enables and the fact
that it does not respect national or supranational
borders [7]. Given these characteristics, traditional
state-based command-and-control mechanisms
such as legislation and binding rules are unlikely
to serve as an immediate solution to manage AI's
effects on society.
This article argues that alternative mechanisms
that draw upon the concept of " soft law " will be
needed to manage the risks associated with AI
deployment and build confidence and trust in those
who interface with these systems. In particular, it
focuses on the role that the insurance sector may
play in mitigating risks associated with AI. While
often not a focus of scholarship writ large, insurance
groups (for the purposes of this article meaning both
primary insurance and re-insurance) are frequently
at the forefront of rulemaking, thus shaping organizational
behavior.
Soft law: A primer
Limitations of legislation and regulations-comDigital
Object Identifier 10.1109/MTS.2021.3123750
Date of current version: 7 December 2021.
52
0278-0097/21©2021 IEEE
monly referred to as command-and-control instruments-have
resulted in policy makers, regulatory
scholars, and practitioners to increasingly explore
" softer " mechanisms to shape the behavior of individuals,
entities, and/or sectors [8]. As defined by
Marchant, soft law refers to frameworks that " set
forth substantive expectations but are not directly
enforceable by government " [9]. These mechanisms
include, for example, codes of conduct, certification
schemes, and industry standards. Soft law
mechanisms can be initiated and implemented both
IEEE Technology and Society Magazine
IEEE Technology and Society Magazine - December 2021

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - December 2021
