HOME   |
GLIMPSES
Unfolding Artificial Intelligence with SASI Studio
- Stella Manferdini
- GLIMPSES
Keywords:
Artificial intelligence is having a significant impact on architecture and product design, introducing novel possibilities and innovative solutions. However, it must be recognised that there are different forms of artificial intelligence and that the latest trends in images generated by neural networks are only the tip of a large iceberg. Indeed, the construction industry works on projects that require different skills, from creative to technical-organizational. Such a complex theme has been unfolded to us by Mattia Santi and Francesca Silvi, Co-Founders and Directors at SASI Studio, a London based architecture, design and technology studio at the forefront of A&D innovation.
Floating Mansion, SASI Studio
Image copyright: @SASI Studio
Museum, SASI Studio
Image copyright: @SASI Studio
How is artificial intellicence redesigning the engineering field with respect to architecture and product design?
In recent years, new algorithmic and BIM design tools have been introduced, from simple 3D modeling to the most advanced simulation-based and shape research systems. The whole of these technologies is giving designers the possibility to manipulate complex geometries and to system different kinds of data that were previously problematic to manage manually. For several years now we have been using genetic algorithms and optimization tools based on technologies that allow us to understand trends and optimize projects in this regard. Particularly in architecture AI can be used to analyze large amounts of data and identify patterns and trends more quickly and efficiently than a human design team would. In fact, these neuronal networks are trained on a number of datasets, ideas and references that go far beyond human capacity.
Alongside these tools there is always a need for manual integration. In fact, such technologies are used in the early stages of formal research. Successively, the “traditional” design part intervenes to transform ideas and sketches into architectural models. Thus, there is still a big gap between what artificial intelligence can do and the actual delivery of a project. Then, technologies little known as Neural Radiance Field, which allows the transformation of two-dimensional images into three-dimensional ones, can greatly facilitate the work of designers in the acquisition of information from physical reality.
The discourse on artificial intelligence has a greater depth than that of a simple text-to-image translation. We can say that it is part of a larger process that has brought a real revolution in our industry and has allowed us to create curved buildings or structures endowed with behaviors difficult to rationalize. Finally, it has given us the possibility to transfer data through digital machines such as 3D printers, robotics and other tools that allow us to translate information in physical matter. There are, however, several aspects in which it must be made even more specific in order to facilitate design development.
Tower, SASI Studio
Image copyright: @SASI Studio
Pavilion, SASI Studio
Image copyright: @SASI Studio
The subject of artificial intellicence is not recent. Wanting to collocate it from a temporal point of view, when did it enter the world of architecture?
Around 2007 the first algorithmic design tools were introduced. Grasshopper in particular was one of the very first visual programming languages that allowed us to achieve simulations of genetics and structural optimization. At the time that visual programming was introduced, also certain automatisms made their entrance in this sector, albeit not to the extent that they are now. Therefore, not through the use of neural networks but rather through algorithms able to manage complex systems with minimal human intervention. Besides, speaking of neural networks, the first experiments date back to the beginning of the century, although not applied to the world of architecture.
Which artificial intelligence platforms do you use the most and what are their features?
In the field of design and architecture, Midjourney, Stable Diffusion and DallE are the most widely used platforms. We prefer to work with systems based on Generative Adversarial Neural Networks (GUNs) because we believe that in it the learning process is closer to the way humans learn to draw an object.
How does these new intelligent systems coexist with green architecture models?
AI can help in the context of green architecture to create simulations, access and manage material data. In addition, it can help architects make more informed decisions about the placement of buildings, the layout of spaces and the use of materials, thus improving energy efficiency and project sustainability. Especially in the initial stages it is a powerful tool to understand how to set the entire project in an eco-sustainable way both within the site where it must be positioned and oriented and in terms of the spaces that make up the building itself.
Sculptural stairs, SASI Studio
Image copyright: @SASI Studio
Sofa, SASI Studio
Image copyright: @SASI Studio
Speaking of metaverse and virtual words, what is your perspective on these new frontiers? How do they support your work?
Metaverse is a natural evolution of the two-dimensional internet as we know it and today it is still in its infancy. The hype on this theme began about two years ago when Mark Zuckerberg announced the change of the name of Facebook into Meta. But the idea of metaverse already existed before that day. Technologies of spatial computing, virtual reality, augmented reality and mixed reality will make us go from users interacting with the digital world via a screen to users that within the digital world interact with their physical body. Indeed, the metaverse can allow the creation of a digital layer that overlaps the physical reality offering new possibilities of interaction between the physical and the virtual world.
The emergence of blockchain technology also allowed defining the concepts of ownership of a digital object, making the context of the metaverse more accessible. Today one can thus possess a virtual terrain. Interest in the topic continues to grow and more and more platforms such as Sandbox and Decentraland are emerging, driving the volume of investment in this industry. We can consider the metaverse as the system of all these realities, no one being the metaverse per se – just as every website contributes to the internet but does not identify with the latter tout court.
Which of your projects best exemplify the benefits of using these technologies?
In many of the projects we have developed, we use algorithms-based technologies for shape research and rationalization. Projects such as Vaults Tower and Islay Wharf Tower rely on these technologies both for formal research and for the study of combinations for the façade’s modularity. We tend to use different types of technologies depending on the design phase. For example, artificial intelligence in the strict sense is used in the initial phase of idea research. In fact we use AI as an evolved Pinterest that allows us to interact with different ideas and models. In the rationalization phases we use genetic algorithms or multi objective optimization. Finally, we use virtual reality in project visualization to simulate the appearance and performance of a building before it is built.
Read other articles about...
Rewriting the architecture of social spaces with Rizoma Architetture
The social role of architecture as a place and a means of encounter, a forge of creativity and avant-garden, is investigated by the Bologna-based studio Rizoma Architetture, so as to turn such notions into concrete and innovative projects.
The new frontiers of the workspace with deamicisarchitetti
Since 2005, the year in which Giacomo De Amicis founded deamicisarchitetti, the Italian studio gives rise to projects that combine memories and contemporaneity. Its specific language, which is composed of advanced criteria of environmental sustainability, comfort, technology and attention to the context, applies to the most diverse areas.