Exploring Cooperation and Collaboration in Human AI Teaming (HAT) – EASA AI Concept paper V2.0

The aviation industry has always been at the forefront of technological innovation. With increasing computing power and the desire to reduce workload, increase efficiency and effectively manage unexpected events AI applications are moving from the background to the foreground, and will move from being technical systems supporting a human – human team to become a human-AI system team. The EU AI act (2021) (EU 2021) has the key provision of "Human Oversight" with the requirement of human control especially in high-risk situations. EASA with its AI roadmap is addressing the challenges and opportunities for AI in aviation, while keeping the human in the centre. The EASA concept paper: guidance for Level 1 & 2 machine learning applications (EASA 2024) proposes means of compliance in the area of human factors for AI in order to understand how the relationship between people and AI can be managed into an effective ‘team’. In the EASA concept paper cooperation and collaboration processes relate specifically to pre-defined AI-levels. Level 2 encompasses directed decision-making process with full authority of the end-user (Level 2A) and a supervised decision-making process (Level 2B). The EASA concept paper elaborates the different processes upon which the AI levels are based, encapsulating a full consideration of Human Factors within the context of AI system development. Within the scope of the EASA AI Concept paper, the move from Level 2A to Level 2B is defined by the move from cooperation to collaboration. EASA has introduced these terms to reflect differences in the extent of team working between AI and human task contributors. These terms, and their specific meaning in terms of human-AI teaming have not formerly been widely discussed or shared. This paper therefore seeks to redress the balance by describing the terms for an audience who will be central in ensuring the principles of the EASA Concept paper are implemented in AI based systems.