The Science in Sermons project is a large grant-funded research effort that uses AI and 📝Text Analysis or NLP to understand how science is portrayed by religious organizations. It draws on a 📝Database of Sermons of over 150,000 sermons with the appropriate metadata, analyzed with 📝CulturePulse text-analysis technology. The work blends evolutionary science, computer science, and religious studies into a cohesive interdisciplinary approach, rooting questions from religious studies in evolutionary theory and applying big-data NLP methods to study them. The framing treats religious groups as complex adaptive systems that change, and are changed by, their environment, and asks how and why religious schemas shift under environmental pressures, particularly perceived threats such as terrorism and economic hardship, as well as natural disasters. Three lines of analysis carry the hypotheses. Using word embeddings, the conceptual distance between key religious terms and scientific terms is expected to be greater in contexts of higher socio-economic insecurity and perceived threat. Using semantic network analysis, the structure of conceptual networks is expected to change under perceived threat, with science-related terms becoming more central in times of socio-economic stress. Using topic modeling, churches that mention science in positively-sentimented topics are expected to cluster where religious identities are less primary, while regions where religious identity is primary show more negative sentiment around science. As of this writing the proposal sits in review at the John Templeton Foundation, passed to their board in 2021 and awaiting final approval.