With over 40 years of experience, UNIDIR is an autonomous institution within the United Nations that conducts independent research on disarmament and related problems; particularly international security issues.
Our specialized focus on disarmament and arms control sets us apart from other think tanks. Our status within the United Nations system enables us to engage with and support Member State deliberations. Our role is to provide ideas and advice and to facilitate dialogue that can advance multilateral arms control and disarmament.
Description of Activities on AI
Project 1: Security and Technology – AI and Autonomy Workstream
The UNIDIR Security and Technology Programme’s AI and Autonomy workstream conducts original research and convenes international events to promote a fact-based, technologically sound dialogue between policymakers, the tech community, the private sector and other stakeholders working on AI technology and its implications for security. This project directly supports the Convention on Certain Conventional Weapons Group of Government Experts on Emerging Technologies in the Area of Lethal Autonomous Weapon Systems in its efforts to advance multilateral debate on concepts such as human control and responsibility, the human-machine interface, and the predictability and reliability of AI-enabled conventional weapon systems (among other considerations). This project also seeks to address considerations related to broader applications for AI in military systems—particularly in decision-making support tools, cyber operations, and nuclear command and control— which themselves raise novel concerns about understandability, reliability and predictability; the potential for unintended interactions or outcomes; and susceptibility of these systems to manipulation. The rate of technological progress in this space requires, as the Secretary-General has described it, a “broader consideration of the impacts of introducing autonomy and artificial intelligence into other military systems, and how effective governance and risk mitigation can be achieved”. The implications of AI for digital, physical and even political security require a fundamental reassessment and, in some instances, re-equipping of the multilateral arms control toolbox. In the period 2019–2021, UNIDIR’s AI and autonomy workstream will seek to a) support understanding of the implications of autonomy in weapon systems and b) explore the options available for AI arms control.
Challenges and Opportunities
Complexity and constant evolution of the research domain, uncertainties regarding possible future applications of military AI, low technological literacy of many policy makers and a reluctance to adopt multistakeholder approaches (particularly in cooperation with the private sector and technical community) to international security challenges
Demand is high for UNIDIR primers, briefings and events, such as the Innovations Dialogue (https://unidir.org/events/2020-innovations-dialogue) where we seek to create spaces to build knowledge, raise awareness among policy makers and convene multi-stakeholder discussions on new technology issues, as mandated by the Secretary-General in his Agenda for Disarmament.
- Need and demand for focused research on specific topics, clarification of the scope and exact meaning of broadly used terms or concepts, as well as description of process pertaining to the use of AI in the framework of military operations or the weaponization of AI.
- The multi-stakeholder approach continues to be valuable for finding common ground and engendering constructive approaches among stakeholders holding divergent or competing points of view.
- Neutral expert analysis is very much welcomed by stakeholders and policy makers.