Global Format
​
Joint sessions will be held across European and North American participants where time zones overlap. Duplicative sessions will be held separately in North American and European time zones for some sessions to allow for all day participation.
​
Content Organization
​
The symposium will be held over two and a half days:
-
Day 1 (March 22) will focus on active and open discussions on the 7 core topics (see below).
-
Day 2 (March 23) will complete discussions on the 7 core topics, as well as bring the conversations to life in the real world with demos and supporting discussions including research agenda, corporate agenda, and the technology ecosystem of AI.
-
Day 3 (March 24) will focus on summarizing the information shared throughout Day 1 and 2, including summarizing any duplicative sessions across time zones. The community should develop a consensus around key recommendations, action plan, and research agenda for different stakeholders.​
7 Core Topics
Each core topic has a dedicated discussion session, as well as linked resources for review beforehand to inform the discussions.
-
AI Principles: In order to implement AI ethics, one must first understand and define what is ethical. This discussion may include classical approaches and their recent updates on ethical foundations, including utilitarianism (Mill, greatest good for greatest number, optimizing benefits and probabilities), rules, norms and virtues (religious traditions, Kant), fairness (Rawls, etc.), political consensus, other schools, and combinations. AI principles may focus on accountability, transparency, interpretability, explainability, safety, security, privacy, robustness, diversity, human agency, and more.
​​
-
Machine Ethics: When artificial intelligence is incorporated into machines and autonomous systems, the need for ethical AI becomes even more apparent. One does not need to look far for recent examples of AI systems that have caused harm, or even turned deadly. The discussion on Machine Ethics may include policies and frameworks for implementing ethics in software and autonomous systems: Values trade-offs, planners, optimization, ontologies, theorem-provers, machine learning, data structures, combination architectures, non-standard methods, specific applications and domains.​
​
-
Automating Machine Ethics: In order for AI to be able to scale successfully and provide the hoped for benefit to humanity, we need to have ethics incorporated into the development process at the initial stages, not as an afterthought. A clear and meaningful way to continuously measure the performance of AI against ethical standards, similar to how systems are now monitored for performance, will be key for building sustainable ethical systems. Topics for exploration may include ethics by design and automated bias detection.
​​
-
Enterprise AI: AI provides a wide range of opportunities for organizations, but with it brings an array of challenges. Discussions will focus on current enterprise AI strategies, business use cases, software and model development processes, adoption of AI/ML models, standards, compliance, governance, skills, and capabilities.
​
-
Public Sector AI: Organizations should not be the only ones benefiting from AI. There are many use cases for AI in the public sector, which may bring about unique challenges and considerations. Discussion may include use cases of AI in the public sector, national AI strategy and regulation, and AI for military.
​​
-
Operationalizing AI Ethics: Many can agree on the high level tenants and goals of ethical AI, however things can get messy when putting pen to paper, or rather code to computer. This topic will focus on concrete tools, technologies, and real world examples of how ethical AI can, or has, been achieved in real world scenarios.
​​
-
Standardization and Certification: In order to achieve success as a whole, the community needs to continue to place the utmost value on advancement through standardization and high quality training. Education, training, standards, third party AI certifications, tools, platforms for risk management and compliance, as well as AI/ML model development methodology and practices will be showcased in order for continued education after the Symposium conclusion. Topics may also include the current state and future for AI ethics adoption and governance methods.​
​​​
​