Contemporary architecture, as well as design, has revolutionised the approach to form creation, prioritising increasingly efficient and, above all, adaptive modelling. It is essential for designers to identify the requirements that the product to be realised must meet for its own production. This methodology, driven by technological advancements in the field of artificial intelligence (AI) and the use of advanced algorithms, enables the exploration and generation of optimised products from various perspectives based on chosen criteria; the resulting forms integrate functionality and aesthetics. This research proposes the use of tools within the parametric modelling environment Grasshopper, such as Galapagos and Octopus, which employ AI algorithms to enhance the design process and optimise forms for additive manufacturing. Experimentation with these plugins allows leveraging evolutionary algorithms to explore a wide range of design solutions, enabling designers to efficiently optimise complex forms. In this context, AI facilitates tackling multi-objective optimisation problems, improving parameters such as structural strength, material usage, and minimisation of printing times. This approach not only enhances the efficiency of the design process but also opens up new possibilities for innovation in design by integrating the advanced computational capabilities of AI with the creative potential of parametric design.
Artificial Creativity. Design Evolution in the Age of AI
The advent of artificial intelligence (AI), in combination with the ubiquity of digital sensors, computer networks, and automation that has characterised the last decade, is transforming the socio-economic environment and defining a probable new industrial era. This evolution also inevitably involves the world of Design by redefining the designer’s role, the object of design, and user relationships. Many of today’s artefacts, whether an iPhone application, a car, or a building, are increasingly connected to the designer who conceived them, thanks to a continuous flow of data detailing many aspects of the user experience. This same information can be used to train AI neural networks capable of autonomously generating specific solutions without human intermediation. An artificial intelligence engine can thus anticipate users’ needs and behaviours, proposing solutions that are improved and customised according to the particular use that distinguishes each customer. This paradigm shift has important implications for the role of the designer. Through the analysis of pioneering case studies, this paper analyses the possible developments, delving into the repercussions for design theory and practice concerning the theoretical framework used today to interpret the discipline of Design.
Advanced Digitization of Cultural Heritage via NeRF
Digitization processes, increasingly employed in various ways for the accessibility and preservation of cultural heritage, continuously engage with the diverse range of materials that define it. Objects like reflective ceramics and transparent glass have complex optical properties, which pose a unique challenge for digital acquisition methods such as photogrammetry and laser scanning, as well as for the modelling process. These materials often generate specular reflections or refract light, which interfere with the conventional algorithms used in the acquisition processes, leading to incomplete or inaccurate 3D models. This study explores the potential of Neural Radiance Fields (NeRF), an innovative 3D reconstruction technique based on deep learning, to overcome these limitations. By using volumetric encoding of scenes and simulating complex light interactions, NeRF captures phenomena like reflections and refractions with consistent realism. The exploration is carried out through a stress test on ceramic and glass materials, using both NeRF technology and traditional systems like digital photogrammetry. The comparison of the results highlights the advantages and disadvantages of both technologies, while emphasizing the current need for their complementarity, with workflows still largely hybrid.
AI-Enhanced UAV Photogrammetry Point Clouds for Cultural Heritage Assessment
Advancements in low-altitude remote sensing and image analysis have revolutionized the digitisation of real-world objects, initially represented as point clouds. Over the past decade, drone-based surveying has gained traction; however, noise during data capture and 3D reconstruction remains a critical challenge, affecting the accuracy and usability of UAV images in real-world applications. This study presents a novel approach to enhance cultural heritage assessment using deep learning models based on the Autoencoder Gaussian Mixture Clustering Model and the Autoencoder K-means Clustering Model. These models improve the accuracy of point clouds generated from UAV images for better condition assessment and survey. The research focuses on photogrammetric accuracy parameters for clustering point clouds. The new approach chooses K-means for finding global patterns due to its robust accuracy, and the Gaussian Mixture Clustering Model for local changes and inspection applications. Moreover, these models investigate the accuracy of point cloud clusters generated from K-means and the Gaussian Mixture Clustering Model. To validate the new approach, the study evaluates variations in key parameters under diverse built-environment conditions using the Temple of Neptune point cloud in Paestum, Italy. The results demonstrate significant improvements in point cloud 3D reconstruction, leading to more accurate surveys and assessments. The proposed method surpasses traditional techniques, offering enhanced applicability for point-cloud filtering. Furthermore, comparisons of clustering results highlight the algorithm, establishing its promising potential for advancing UAV-based surveying and inspection practices.
Between Impossible and Probable. Architectural Recognition Through Qualitative Evaluation of Artificial Intelligence Response
The paper investigates evaluation methodologies for AI-based virtual reconstruction of historical architecture, focusing on the uncertainty inherent in generative predictions. The research critically compares traditional analog reconstruction workflows with AI-driven reconstruction processes based on GAN networks, proposing a conceptual framework called the “Uncertainty Tree” to describe the decision-making chain involved in reconstruction. The study develops a qualitative evaluation methodology intended to complement quantitative pixel-based assessments by introducing prediction thresholds ranging from impossible to highly probable reconstructions. Through experiments on datasets of Greek Doric temples and Mudejar Romanesque churches, the research argues that specialized datasets and expert qualitative evaluation significantly improve the plausibility and reliability of AI-generated architectural reconstructions. The paper ultimately proposes a new epistemological framework for evaluating AI predictions in heritage reconstruction, emphasizing human supervision, dataset specificity, and iterative qualitative assessment.
Hybrid Construction of Knowledge Graph and Deep Learning Experiments for Notre-Dame De Paris’ Data
After the fire that destroyed part of the cathedral Notre-Dame de Paris, a working group specialized in digital data coordinated a scientific project that would allow the management of all the digital data produced by scientific research activi-ties along with restoration operations. The ERC advanced grant “nDame Heritage” project combines digital humanities with computer science and artificial intelli-gence to create a collaborative knowledge system that analyzes multiple views from different experts on the same cultural heritage objects. In this work, we designed a hybrid artificial intelligence workflow based on both knowledge graphs and deep learning models for semantic segmentation of 2D images. We show that this hybrid approach can help experts to process, integrate, and enrich Notre-Dame’s data.
Artificial Intelligence for Space Weather Prediction
This paper reviews the application of artificial intelligence methods for space weather prediction, with a focus on solar flare forecasting. It contrasts physics-based and data-driven approaches, highlighting how machine learning models trained on large datasets of solar magnetograms and flare records can provide probabilistic predictions. While AI methods represent the current state-of-the-art, challenges remain in terms of data imbalance, feature selection, and model interpretability.
Preliminary Study on Architectural Skin Design Method Driven by Neural Style Transfer
This paper explores the application of neural style transfer as an AI-assisted method for architectural skin design, aiming to enhance formal diversity and support conceptual design processes. The research investigates how convolutional neural networks can extract and recombine content and style features from different visual sources to generate alternative façade design proposals.
The methodology is based on the neural style transfer approach introduced by Gatys et al., implemented using a pre-trained VGG-19 network. As illustrated in the workflow diagram (Fig. 2, p. 744), the process defines content and style loss functions to iteratively optimize an output image that combines structural features from a content image with stylistic attributes from a reference image. The study applies this method to multiple sets of architectural images, including traditional Chinese buildings, modernist architecture, and urban skylines, combined with stylistic references such as Notre Dame, landscape painting, and science-fiction imagery. Results demonstrate that style transfer can generate diverse and visually suggestive façade configurations, supporting architects in the early design phase by providing rapid exploratory variations. However, the generated outputs remain conceptual and require further interpretation and development, highlighting the role of AI as a tool for inspiration rather than a deterministic design system.
The Importance of GAN Networks in Graphic and Creative Learning Processes Associated with Architecture
This paper examines the role of Generative Adversarial Networks (GANs) and deep learning techniques in graphic production and creative processes related to architecture. The study discusses how AI systems trained on large labeled image datasets can recognize, classify, and generate visual content, highlighting the interaction between generative and discriminative networks as a mechanism for iterative learning and image synthesis.
Through a critical review of current applications—such as DALL·E, MidJourney, AICAN, and The Next Rembrandt—the research analyzes the potential of GAN-based systems to produce architectural representations, artistic images, and design proposals. It also explores their integration into architectural workflows, including parametric design, generative design, simulation tools, and BIM-based environments. Results indicate that AI can support creative exploration, automate repetitive processes, and generate alternative design scenarios, while remaining dependent on training data and constrained by limited contextual understanding. The study concludes that GAN-based AI represents a powerful assistive tool rather than a replacement for architects, with future developments likely to enhance hybrid human–machine design processes.
Automatic Virtual Reconstruction of Historic Buildings Through Deep Learning. A Critical Analysis of a Paradigm Shift
This paper critically examines the emerging use of artificial intelligence for the automatic virtual reconstruction of historic buildings, comparing it with traditional heritage reconstruction methodologies. The authors trace the historical evolution of reconstruction practices—from manual drawing, archaeological interpretation, CAD modelling, photogrammetry, and BIM—to current AI-based approaches founded on Deep Learning. The study focuses on Generative Adversarial Networks (GANs) trained to infer missing architectural parts from ruined structures, using synthetic datasets of Greek temples represented in multiple ruin states and reconstructed versions. A complementary Natural Language Processing workflow is also tested to improve segmented image generation and automate parts of the training process. Results suggest that AI can identify hidden formal and constructive patterns, generate multiple predictive alternatives, and support specialists in evaluating reconstruction hypotheses. At the same time, the paper warns that neural networks simplify architectural diversity into generalized stylistic rules and still depend on carefully designed datasets. The authors conclude that traditional scholarly expertise and AI prediction systems will likely coexist, marking a paradigm shift in how virtual reconstruction may be practiced in the near future.