Human Oversight (Article 14)

CONTENTS: 1. Introduction.2. Human oversight in the AIA.2.1.The Commission's proposal in April 2021.2.1.1.Human oversight, what for?To prevent or minimise risks.2.1.2.What kind of human oversight?Effective by design.2.1.3.How to achieve effective human oversight?2.1.4.The deployer's role in human oversight.2.2.Human oversight's path in the ordinary legislative procedure.2.2.1.Voices calling for a right to human intervention (and other safeguards) for decision-making based on high-risk systems.2.2.2.Humans, what humans?2.3.The final version: AIA's human oversight in a nutshell.3. Thinking aloud: what should we expect from human oversight in the AIA? 3.1.Can human beings fulfil the regulatory purpose of human oversight in the AIA? 3.2.Do humans need to be in the AI system's loop to ensure an effective human oversight under the AIA? 3.3.Beyond human oversight, is the AIA human-centred? 4. Concluding remarks.