Track 2.3: AI-Driven Decision-Making for Equitable and Inclusive Cities

Changing work and life patterns is an opportunity for both big and smaller cities to become more liveable by following an integrated and balanced approach to the built environment, innovation culture, public services and socio-cultural life and evoking a strong sense of active citizenship. To tackle complex challenges of the 21st century, a deep rethinking of policy and implementation is required so that meaningful interventions and public services can be developed. Public Interest Design (PID) is a methodological framework that can be used to make cities, communities, and neighbourhoods more liveable. To answer those challenges our PID capacity building program would like to supports city administrations in the Baltic Sea Region in becoming more responsive and agile by encouraging the use of design methods in public services. This is also in line with the principles of shaping a resilient city, especially in the field of public spaces, the revitalization of which after the pandemic has become crucial for the proper functioning of cities. PID capacity building programme evaluation framework was elaborated in a participatory process. The evaluation objectives were co-defined with representatives of the partner cities. Evaluation of the PID capacity building programme covers (1) PID ecosystem change within the partner and following cities as well as the implementation of the PID charter, (2) documentation of the PID use cases implementation and (3) testing and implementation of PID training programme. As cities evolve to meet the needs of modern society, the role of architects and urban engineers is shifting towards more participatory, technology-driven, and adaptive approaches. This presentation introduce and evaluates training programmes designed to equip professionals with the skills needed to co-create liveable urban spaces using digital solutions, agile methodologies, such as public interest design. As part of the Livability project and its training program, a web-based digital platform was developed to facilitate continuous learning and provide on-demand access to practical tools and solutions. Created using the Unreal Engine framework, the platform enables participants to revisit and apply relevant content as needed throughout their professional practice. This study explores innovative methods of education and training, emphasizing problem- and project-based learning, work-based education, and employer engagement. A key focus is the integration of above-mentioned digital technologies such as interactive platforms connected to external applications in training and their impact on effectiveness of the training process. Presentation introduces Training Programme and its evaluation introduced in frame of the Liveability Interreg BSR project, showcasing how Public Interest Design approach facilitate engagement between public local governments, academia and civil society. BSR was chosen because the governance models being implemented there are advanced and could become good examples for other cities in different geographical contexts. It also highlights the importance of lifelong education and the role of academia- public sector collaboration in preparing professionals for the challenges of urban transformation. Findings suggest that agile design thinking methodologies—characterized by iterative, user-centred approaches—can support developing responsive urban environments. Alternative learning strategies, including blended learning and digital tools, are discussed as effective means to foster liveability and resilience. The training program improves the capacity of local leaders and civic workers by providing them with knowledge, practical skills, and tools to carry out participatory processes that contribute to the transformation of urban spaces. It supports the integration of co-creation practices and introduces a more agile working culture that embraces experimentation - key factor in fostering creativity and enabling the implementation of social innovation. The training program, by peer-to-peer practical sessions, helps to design and implement real-world urban regeneration projects conducted in Baltic Sea cities such as Gdynia, Kiel, Riga, and Pori. In conclusion, this presentation underscores the need for continuous upskilling in planners, emphasizing digital literacy, user-centred approaches, stakeholder collaboration, and interdisciplinary learning. By embracing new technologies and alternative education models, training programmes can better prepare professionals to create more liveable, inclusive, and resilient cities. 

Background In the context of digital urban development, contemporary urban planning encounters a key dilemma: although functional mixed-use development is widely promoted to enhance urban vitality, its influence on residents' emotional well - being lacks effective quantification. Traditional planning evaluation indicators, mainly focusing on physical space and economic benefits, are unable to comprehensively reflect the psychological responses of urban residents to spatial environments, resulting in sub - optimal urban designs. This problem is evident in Nanjing's central urban area, where historical blocks face development stagnation due to insufficient functional mixing, while CBDs suffer from residents' psychological stress caused by excessive mixing.​ This research innovatively integrates smart city facilities and social media data, applying the Bertopic-CRF big data model to establish a data - driven analysis framework. By extracting real - time emotional information from social media and correlating it with spatial elements, the framework analyzes the interaction between urban space and human psychology. Research Objectives This research aims to: 1.Quantify nonlinear relationships between POI functional mix and public emotional responses; 2.Identify optimal mix thresholds balancing efficiency and well-being; 3.Decipher functional synergies that amplify positive sentiment; 4.Develop transferable "emotion-calibration" tools for urban design. The central question asks: How do spatiotemporal patterns of functional mix dynamically shape collective emotional landscapes in high-density urban cores? Methods We established a Spatiotemporal Coupling Diagnostic Model (STCDM) integrating: Emotion data:Weibo posts (2015-2024) processed via BERT-CRF hybrid model. POI mix dynamics: Spatiotemporal entropy cubes (500m grids × quarterly) for 19 functional categories. Coupling analysis combined: ① Bivariate Local Moran's I detecting spatial clustering of mix-sentiment relationships; ② Coupling Coordination Degree (CCD) quantifying mix-emotion synergy; ③ Panel Vector Autoregression revealing temporal interactions. Three stratified prototypes (High/Mid/Low-mix zones) were analyzed. Theoretical Contributions The research reveals that the mix-sentiment relationship is not linear but characterized by dynamic synergy and time-lagged effects. Moderate functional mix enhances emotional well-being, while excessive mixing beyond a certain threshold diminishes these benefits. This challenges conventional density-centric planning paradigms and proposes an "Emotional-Entropic Resonance" framework, emphasizing the dynamic alignment between psychological experience and spatial configurations. Additionally, the study identifies "psycho-functional hysteresis" in historic districts, where emotional adaptation lags behind physical renewal—a critical insight for heritage conservation strategies. For planning practice, the study provides actionable strategies: ①Coupling-Optimized Zoning: Identifying optimal mix-sentiment synergy ranges to guide functional layouts and avoid over-mixing. ②Sentiment-Responsive Governance: Establishing real-time monitoring systems to trigger interventions when emotional synergy declines (e.g., tactical urbanism in high-mix areas). ③Functional Synergy Amplification: Prioritizing emotion-enhancing combinations (e.g., green-commercial integration) to elevate overall spatial well-being. By integrating smart city infrastructure and social media data analytics, this research enriches emotional geography theories with empirical evidence.

Traditional urban planning relies on predictive modelling, which is often inadequate for complex, nonlinear urban systems. This research addresses the shift from prediction to inference in planning, focusing on how interventions impact future urban scenarios. Leveraging Artificial Intelligence (AI), specifically Reinforcement Learning (RL), it aims to enhance scenario planning and performance evaluation for sustainable regeneration within high-density built environments, moving beyond traditional prediction methods to address the crucial challenges of existing urban renewal. Despite the widespread application of scenario planning and performance evaluation approaches in practice, the potential of AI in these methodologies has yet to be fully realised. Previous scenario planning studies mainly focus on land use and spatial simulations related to urban growth, with scant attention paid to the built environment. A fundamental debate in scenario planning concerns whether future scenarios should explore plausible futures beyond expectations or prescribe desirable futures. This dichotomy gives rise to exploratory and normative scenario planning approaches. This paper presents a framework seeking to integrate the foci of both approaches. Specifically, within the context of urban regeneration, we define multiple target benefit orientations to explore future scenarios driven by different priorities. Subsequently, the framework identifies integrated benefit scenarios – essentially physical urban form proposals – while explicitly accounting for the economic costs of regeneration. Incorporating these costs is crucial, transforming scenario planning from a cost-disregarding hypothetical exercise into a feasible, implementable framework. The proposed methodological framework adapts Chakraborty and McMillan’s (2015) scenario planning guidelines, comprising four stages. Scenario Modelling (SM): Three sustainability-oriented target scenarios are defined: Social Equity (S1), Economic Growth (S2), and Environmental Health (S3). The environment of RL is set as the current spatial configuration. Agents (roads, blocks, buildings) perceive environmental stimuli. Agents can execute six distinct actions: two at the block level (Land Use Change - A1; Vehicular-to-Pedestrian Conversion - A2) and four at the building level (FAR Transfer - A3; FAR Increase - A4; FAR Decrease - A5; Building Function Conversion - A6). Performance Evaluation (PE): Twelve indicators (5 for S1, 4 for S2, 3 for S3) corresponding to the three scenarios serve as reward functions, guiding the RL process. Actions from each iteration are synchronised with Grasshopper (a Rhino-based visual programming plugin) to generate the final physical-form plans for each scenario. Retrospective Tracing (RT): The SHAP interpretability algorithm analyses the marginal contribution of each action to its respective scenario, identifying potential action weights. Spatial Planning (SP): The algorithm is re-executed, incorporating the derived action weights. Crucially, the total economic cost of all regeneration actions per iteration is introduced as a loss function. This cost encompasses demolition, new construction, associated infrastructure costs (linked to building area), green space development costs, and opportunity costs. The algorithm identifies the most cost-efficient physical form solution, optimising integrated multiple benefits. The framework was validated using the Nanjing Hexi Central District (18.17 km², comprising 128 blocks and 2,125 buildings). Results indicate that the optimised solution, derived through scenario simulation via RL, achieved improvements over the baseline of 25% in Social Equity (S1), 16% in Economic Growth (S2), and 28% in Environmental Health (S3). SHAP analysis revealed FAR Transfer (A3) as the most impactful action for enhancing Social Equity and Environmental Health in this context, with SHAP values of 56% and 51% respectively. Incorporating weighted actions and the economic cost loss function as constraints, the algorithm generated the integrated optimal solution. A comparative analysis of this solution against the officially published urban regeneration plan (regarding economic cost, social, economic, and environmental benefits) is currently underway. This research offers significant advancements for AI-driven planning practice. Firstly, addressing a critical gap in recent literature, it specifically targets the sustainable regeneration of existing high-density built environments – a paramount challenge for mature cities – rather than solely simulating urban growth. Secondly, the conceptual framework provides a structured tool for planners to compare the long-term implications of various high-density regeneration strategies. Thirdly, explicitly incorporating regeneration costs as a loss function helps avoid "idealised yet unimplementable" planning pitfalls. Future applications can adjust action or reward weights based on specific regeneration contexts (e.g., prioritising land use change in industrial innovation zones, or restricting demolition actions in historic districts) to generate contextually appropriate physical form solutions.

Urban renewal has become a dominant mode of development in contemporary Chinese cities. As vital carriers of spatial character and cultural identity, historic areas now serve as critical nodes within renewal strategies. However, planning for the conservation of these areas involves managing complex, multi-dimensional information systems. These systems must adapt to evolving urban conditions and respond to the competing demands of cultural preservation and modern urban life. Planners need to make refined design decisions based on a large amount of dynamic data and the demands of diverse groups in real-time. Traditional methods relying on manual data collection and subjective judgment are increasingly inadequate for such dynamic and data-intensive planning tasks, while existing heritage management systems, such as GIS or HBIM, often serve a specific professional problem. To address this gap, there is an urgent need for a more intelligent, data-driven planning tool. This study presents the development and application of a digital platform specifically designed to support intelligent information management and analytical decision-making in the conservation planning of historic urban areas coordinated by multiple people. The platform integrates a cloud-based multi-level database with a modular desktop application to form a cohesive and scalable planning tool. The database, built using PostgreSQL and hosted on Alibaba Cloud’s lightweight server infrastructure, assigns each historic area a dedicated data environment. These databases store multi-temporal records of urban morphology, land use, building condition, regulatory documents, historical information, and stakeholder input, reflecting the evolving development trajectories of each area. The terminal tool includes multiple functional modules that can be accessed simultaneously by the government, professionals, resident representatives and other various users. These modules support tasks such as real-time data entry and updating, inter-user information sharing, and data-driven morphological and non-morphological analysis. Importantly, the platform is capable of extracting and visualising key indicators from large and complex datasets to support evidence-based planning. This enables planners to assess both current conditions and the potential impacts of various conservation strategies with greater precision. This intelligent platform has been tested in the planning work of multiple historic areas in China, such as Gu Nan Street, demonstrating its practical applicability. On the one hand, the information management function of the platform has changed the way of information communication between diverse groups and the planning work, realising the consideration of multiple demands in planning decisions. On the other hand, in real-world projects, the multi-objective analysis of the platform has effectively revealed the cumulative effects of long-term external interventions and internal transformations on the urban morphology and socio-spatial dynamics of historic areas, and can effectively complete the comparison and selection of subsequent planning schemes. This helps decision makers to identify the spatial characteristics and remaining defects of the block as the core basis for the formulation of a more livable design. At the same time, it will also promote the planning work to comprehensively consider the social and cultural information in the block, so as to maintain the authenticity of cultural heritage in the process of urban development. By addressing the limitations of traditional conservation planning—particularly in managing fragmented and dynamic data—this platform demonstrates how digital technologies can contribute to more scientific, integrated, and responsive urban planning. Beyond enhancing efficiency and accuracy, the system provides a methodological and technical foundation for realising multi-objective, human-centred development goals in historic urban environments. As such, it offers valuable insights into how emerging technologies can help cities meet future challenges while safeguarding their cultural legacies.

The rise of Artificial Intelligence and digital platforms is reshaping urban economies, governance, and daily life. As cities transition into always-on, hyper-connected environments, traditional planning paradigms are being challenged by new forms of mobility, employment, and interaction. In cities like Bengaluru, India - a major tech and innovation hub, and Berlin, Germany - a leader in urban experimentation and digital governance, they illustrate divergent yet converging pathways toward a digital urban future. This paper explores how AI and digital technologies are enabling 24-hour urban economies while raising critical questions about equity, governance, and spatial justice. ISOCARP 2025 emphasises the importance of creating resilient, forward-looking cities that embrace digital transformation. Directly responding to the congress theme of 'Urban Economy and the Digital Age: 24-Hour City and AI', this paper critically examines how emerging technologies and AI-infused governance are shaping urban economies and planning strategies across different geographies. Drawing on personal insights from Bengaluru and Berlin, the paper situates digital transformation within a global–local continuum, showcasing how these two vastly different urban contexts are grappling with similar disruptions and opportunities. The relevance of this topic is underscored by themes of urban equity, community engagement and economic diversification — all of which are crucial pillars of ISOCARP’s vision for inclusive urban futures. The paper proposes a strategic planning framework for integrating AI-driven innovation into urban economic development, emphasising the creation of inclusive, 24-hour cities that are both resilient and responsive. Building on comparative insights from Bengaluru and Berlin, it focuses on three key dimensions: 1. Digital Infrastructure and Governance: Examines how smart governance tools like real-time data platforms, digital twin models, and predictive analytics can enhance participatory planning and real-time service delivery. Case studies highlight Berlin’s participatory e-governance models and Bengaluru’s growing use of AI in traffic and utility management. 2. New Urban Economies: Moving beyond growth led by the real estate sector, the paper explores policy instruments like land value capture, tax incentives for tech-driven SMEs, and urban innovation zones, which promote economic diversification. The potential of these tools to deliver community benefit and long-term prosperity, especially in underserved districts, is assessed. 3. Public-Private-People Partnerships (4P): By emphasizing bottom-up knowledge and local innovation ecosystems, the paper advocates for governance models that centre public engagement and cross-sector collaboration. It proposes a district-scale pilot framework to test AI-enabled urban solutions co-designed by residents, startups, and government agencies. The analysis concludes with a roadmap for planners and policymakers to integrate ESG principles, local data ecosystems, and AI capabilities into district- and city-level strategies. This approach aims to foster inclusive economic growth, enhance urban resilience, and realize the potential of 24-hour cities in an increasingly digitized world.

Cities in the digital age are undergoing profound restructuring as artificial intelligence (AI) adoption converges with the rise of 24-hour urban economies. This paper examines Riyadh’s transformation within the framework of Saudi Vision 2030, focusing on its ambition to emerge as an AI-enabled global city with round-the-clock economic activity. Drawing on comparative benchmarking across the Global Cities Index (GCI), the Global Financial Centres Index (GFCI), and the Globalization and World Cities (GaWC) classification, the study situates Riyadh against leading peers including London, New York, Dubai, Abu Dhabi, and Seoul. Results reveal a dual trajectory: Riyadh has achieved significant momentum in global rankings and secured unprecedented AI investment commitments—nearly US$27 billion between 2024 and 2025—yet continues to lag behind regional competitors in human capital, livability, and institutional capacity. Regulatory reforms permitting 24-hour commerce and infrastructure projects such as the Riyadh Metro provide important enablers, but governance mechanisms remain underdeveloped. The paper contributes conceptually by integrating literatures on night-time economies, AI urbanism, and global benchmarking, and empirically by presenting Riyadh as a distinctive Global South case. A phased policy roadmap is proposed to consolidate progress, strengthen governance, and position Riyadh as a leader in AI-enabled, culturally specific 24-hour urbanism.