AI-Enhanced Stakeholder Engagement for Climate Adaptation: Evidence from Lithuania

Abstract

Civil engineering faces the dual challenge of decarbonisation and resilience under increasing threat of climate change. While artificial intelligence (AI), machine learning, and digital twins are increasingly applied to optimise design, material reuse, and hazard modelling, most systems remain techno-centric and overlook the human dimensions of adaptation. This article addresses this gap by combining a nationally representative survey of Lithuanian residents (n = 1,013, 2023) with the design of an AI-enabled platform for civil engineering adaptation. The survey captured six domains (hazard experiences, adaptation behaviours, motivational drivers, preparedness levels, institutional linkages, and climate attitudes) providing a behavioural evidence base that reveals how climate concerns and motivations translate into action. The results highlight differentiated motivational pathways, moderate levels of preparedness, uneven institutional communication, and four distinct citizen profiles with specific adaptation probabilities. Building on these insights, the article proposes the Citizen-informed AI for Climate Adaptation (CiA-CA) framework, which systematically maps citizen evidence onto AI system design variables. The framework informs the development of the Lithuanian Construction Materials Reuse Optimization (LSEPO) platform, created under the Civil Engineering Research Centre (CIMC), by integrating hazard-prioritised digital twins, recommender systems with motivational weighting, clustering for personalisation, and preparedness-aware interfaces. Conceptually, CiA-CA advances the integration of behavioural adaptation evidence with socio-technical AI design. Empirically, it provides one of the first nationally representative datasets on climate adaptation behaviours in the Baltic region. Practically, it offers a blueprint for municipalities and industry partners in Lithuania to embed citizen evidence into AI-enabled platforms, with potential transferability to similar European contexts.