Which Biases and Reasoning Pitfalls Do Explanations Trigger? Decomposing Communication Processes in Human–AI Interaction

Innovative applications of artificial intelligence in zoonotic disease management

“…User-centered design and trust-building are essential for successful adoption. Adequate training of end users on the capabilities and limitations of AI is also crucial to preventing the risk of its misuse and the tendency to excessively rely on it [ 132 ]. Finally, the specialized skills required to develop, evaluate, and continually update AI systems pose implementation challenges [ 133 ].…”

Deconstructing Human‐AI Collaboration: Agency, Interaction, and Adaptation

“…Others focused on articulating why designing these interaction paradigms is so difficult [YSRZ20] and what actually is expected from AI collaborators in various domains (e.g., medicine) [CWS*19, MAKA21,ADJ*21]. El‐Assady et al [EAM22] explore the details of the communication processes and propose dynamic models of explanation intent and receiver perception. In an attempt to model this complex interaction, researchers have even looked towards human‐human communication for inspiration [WCM*20, MX17].…”

“…Interdisciplinary research has looked at how human‐AI communication can learn from human‐human interaction [WCM*20]. In particular, El‐Assady et al [EAM22] proposed a communication dynamics model that makes a clear distinction between perspectives (i.e., sender or receiver). This framing of communication is also incorporated into the design of our initial design space.…”