Artificial Intelligence and Collaborative Information Behaviour in Emergency Response: A Review

Emergency response requires professional responders, public authorities, humanitarian organisations, volunteers and affected communities to seek, share, interpret and use information collaboratively. Artificial intelligence (AI) is increasingly used to support these activities, yet existing research has largely focused on algorithmic performance rather than its contribution to collaborative information behaviour. This structured narrative review examined research at the intersection of AI, collaborative information behaviour and emergency response. Selected PRISMA 2020 reporting principles were used to improve transparency in documenting the search and selection process. Eight structured searches using OpenAlex as the sole academic index, covering publications from 2000 to 2025, identified 600 records; 535 remained after duplicate removal, and 58 full-text publications were assessed for eligibility. After eligibility reassessment, 37 publications were retained in the review. The studies were analysed according to emergency setting, actors, data sources, AI techniques, information functions, evaluation approaches and reported limitations. The reviewed publications showed that AI could support information filtering, needs extraction, damage assessment and situation summarisation. However, none of the 37 publications was classified as providing direct evidence of improved operational collaboration between emergency-response organisations. Five recurring considerations were identified: provenance, uncertainty communication, interoperability, human validation and representational inclusion. The concept of AI as an epistemic intermediary provides a perspective for examining how AI-supported information is interpreted and used by human and organisational actors.