IEEE Technology and Society Magazine - December 2021 - 73

some documents or summaries of its proceedings
[40], and multistakeholder bodies in AI can strive for
transparency with similar or greater measures.
Transnational, consensus-based standards
were the most significant tool to fill governance
gaps for nanotechnologies, ultimately influencing
national-level policymaking around the world [41].
Standard-setting bodies will similarly play an important
role in the governance of AI, through harmonization
across technical areas-including, for example,
terminology, nomenclature, privacy, and security-
critical to the deployment of AI across industries and
jurisdictions. Once published, such standards are
likely to be incorporated into national and supranational
legislation and policies, inform the work of
multistakeholder initiatives, and, over time, be incorporated
into multilateral agreements, framework
conventions, and other instruments of public and
private international law. Already, the ISO/IEC 27701
standard on data privacy and its predecessors have
had global impact, with national and supranational
legislatures adopting principles and practices from
the standard, while still tailoring regulation to local
concerns and policy priorities [42]. Furthermore,
transnational standards more directly targeting AI
should similarly enable global coordination while
leaving room for local policy concerns.
As evidenced by ISO TC-229-Nanotechnologies,
standard-setting bodies may exert their influence by
moving beyond purely technical concerns, including
by considering health and environmental issues,
and extending recommendations to policymakers
on consumer and societal dimensions including, for
example, product labeling [43]. Similarly, the JTC 1/
SC 42 and IEEE SA working groups have begun to
evaluate ethical and normative issues in AI and may
be well positioned to influence national and supranational
policy.
However, as with nanotechnologies, legitimacy
troubles may follow should " pay for play " or other
processes lead to inadequate participation or exclusion
of public and civil society organizations [32].
The ISO, and other standards setting bodies such as
the European Committee for Standardization and
the Standardization Administration of the People's
Republic of China should work to ensure that their
processes of setting and interpreting norms for AI
are transparent and open to participation from interested
stakeholders to build public legitimacy and
design more robust norms [31].
December 2021
Future of global AI governance
Taken together, these lessons from other technological
spaces suggest that multistakeholder bodies
and transnational standards may provide optimal
institutions and instruments for the early global
governance of AI, where legitimacy and participation
concerns can be meaningfully addressed. This
conclusion reflects both a normative determination
based on the potential effectiveness of these
instruments and institutions in the immediate term
(relative to other options) as well as a pragmatic
determination that these mechanisms will provide
the most politically achievable options in the shortto-medium
term.
Based on the ideas presented in this article, we
suggest here that the next five to ten years of global
AI governance will likely be characterized by an
increase in the volume and influence of transnational
standard-setting and multistakeholder body
activities around AI. The use of transnational standard-setting
bodies as mechanisms for diffusing norms
on AI to national and supranational governments
would follow the empirical trends observed around
nanotechnologies [41], while engaging multistakeholder
P3s appears likely given the preference of
large private actors to operate through forums such
as the ICH or WEF. As various influential actors distill
norms for AI governance, such as the OECD and
G20 principles [20], standard-setting institutions and
multistakeholder bodies will likely act as primary
mechanisms for dispersing these norms around the
world [44]. The predominance of these institutions
would create a significant role for softer legal norms
in advancing early global governance for AI.
While the EU has recently proposed supranational
regulation for AI, establishing a comprehensive
and risk-based regulatory framework [17], this
push toward hard law likely reflects a development
that will remain unique to Europe in the shorter term.
The notably precautionary postures toward risk and
technological uncertainty in the EU, existing supranational
infrastructure, and greater regulatory resources
generally available in high-income jurisdictions all
provide conditions that collectively allow EU states to
place AI higher on their agendas. While other states
may express interest in hard international law for AI,
the difficulty of achieving the political conditions to
place these issues on the agenda, combined with the
significant resource demands for doing so, will likely
restrict the success of such efforts in at least the next
73
IEEE Technology and Society Magazine - December 2021

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