Living with risk, then and now: A dual review of Cam Grey's Living with Risk in the Late Roman World and of current AI-assisted book reviewing.

This AI-assisted review article offers a dual review: a book review of Living with Risk in the Late Roman World by Cam Grey, and a critical review of the current potential of large language models (LLMs), specifically ChatGPT's DeepResearch mode, to assist in thoughtful and scholarly book reviewing within risk science. Grey's book presents an innovative reconstruction of how communities in the late Roman Empire perceived and adapted to chronic environmental and societal risks, emphasizing spatial variability, cultural interpretation, and the normalization of uncertainty. Drawing on commentary from a human reviewer and a parallel AI-assisted analysis, we compare the distinct strengths and limitations of each approach. The human review provides deep contextual judgment, skepticism, and sensitivity to narrative bias, while the AI-generated review offers thematic organization, broad literature synthesis, and analytical clarity. Our findings suggest that AI-assisted tools, when used alongside expert human insight, can significantly facilitate and enrich the scholarly review process. We argue that such hybrid methods hold promise for accelerating critical synthesis and expanding the scope of reflective inquiry in risk analysis, especially as the field increasingly engages with historical, cultural, and interdisciplinary perspectives.