Topics covered:
Week One: What is AI? – This session provides the attendee with a grounding in AI necessary to understand the purpose and impacts of the EU AI Act. It will cover types of AI and potential AI use cases, examples of real-life negative impacts from poorly developed or deployed AI, ethical principles in AI, and an introduction to various international efforts to regulate AI.
Week Two: The EU AI Act - A risk-based approach to AI regulation. This session focuses on the core of the EU AI Act helping the attendee to understand the intended purpose and structure of the legislation. This will include what technologies and uses of AI fall inside or outside the scope of the regulation, and how different categories of AI are to be regulated based upon their risk. It will also cover the types of AI that the Act prohibits.
Week Three: EU AI Act accountability framework – different actors, responsibilities and required documentation. The AI Act creates multiple accountability requirements for providers and users of AI systems. This session will examine these requirements, including the different responsibilities of AI providers and users, the types of documentation that will need to be created and links to other pre-existing legislation and standards.
Week Four: EU AI act accountability framework – Fundamental rights impact assessment and conformity assessment. In addition to familiar types of documentation, the AI act creates new requirements arounds AI risk assessment, the assessment of risks to fundamental rights and freedoms, algorithmic transparency and accountability, human oversight procedures, post-market monitoring measures and the requirement to register high-risk AI systems with the EU. This session will explore these new types of compliance obligations.
Week Five: EU AI Act accountability framework – Technical documentation. This session explores the technical documentation requirements required for accountability of AI systems under the Act, including advice on how non-technical professionals can understand and manage this. This included guidance on algorithmic auditing – the processes through which we can assess the accuracy, bias or reliability of an AI system.
Week Six: Stakeholder Communication – This final session deals with some of the practical measures of navigating compliance with the AI Act, including appropriate distribution of internal roles and responsibilities, considering liability, potential fines and enforcement for non-compliance, potential skills and training, and how to think about handling complaints or incidents.