Page 69 - AI Governance Day - From Principles to Implementation

AI Governance Day - From Principles to Implementation

The role of AI safety institutes

The idea of AI safety institutes was another key topic of discussion. These institutes aim to
create a network for monitoring and mitigating the risks associated with AI deployment. While
the concept is appealing, the implementation may vary based on a country’s resources and
existing infrastructure.

Ms. Falco highlighted the importance of standards and guidelines in ensuring the safe and
ethical use of AI.

"Standards help us to be able to regulate, that's the work of policy makers in general,
and companies when they develop their own standards, […] have references,
guidelines which help them go down the right line. That just simplifies everyone's
work […] and then we can bring in the participation of tech technicians experts and
other stakeholders in the ecosystem so that we can best […] create clear rules."
(Mercedes Aramendia Falco)

Uruguay’s approach involves incorporating these standards into national policies and
collaborating with international bodies to maintain a balance between innovation and risk
management.

Ms. Theofelus offered a thought-provoking perspective on the terminology used, suggesting a
more positive framing might be beneficial. For countries like Namibia, with limited resources,
integrating AI safety functions into existing institutions could be a more feasible approach.

"Just even the wording 'AI safety institute' already presumes that there's some threat,
as opposed to looking at the positive side of things." (Emma Inamutila Theofelus)

Mr. Verma supported the idea of AI safety institutes, particularly for a large and diverse country
like India. He emphasized the role these institutes could play in providing safeguards and
setting standards that startups and other entities could follow.

"It is a good idea to have an AI safety institute […]. In my country there are a lot
of startups and there would be some guidelines [and standards] to follow." (Niraj
Verma)

5�6 The critical conversation on AI safety and risk

Panelists:

– Professor Stuart Russell: Professor of Computer Science at the University of California,
Berkeley

59