Thoughts on Artificial Intelligence and the Future of Architecture Practice

he Challenge Facing Architectural Practice

We live in a world of an uncertain future for the practice of architecture. The increasing sophistication of buildings and their attendant systems is driving the need for a new paradigm of architecture both in terms of theory and practice. Architects must be proficient at balancing complex, nonlinear relationships between diverse constraints (Landa, 1997). Each of these relationships must be resolved to create a built form. What makes practice difficult is that each relationship has multiple alternatives (Gane, 2012). Just selecting glazing requires balancing six different metrics against thermal, daylight, and glare objectives often from five different manufactures. All this for one, decision among hundreds. Not only are there hundreds of key decisions in the design of a building, but each decision interacts with all the others (Hazelrigg, 2003). Until the rise of Modernism, architects relied on methods as means of decision making (Gomez, 1983). These simple “rules” created “styles” of design that allowed for infinite creativity while preserving an underlying order (Plowrigtht, 2014). With the collapse of a coherent Modernist ideology, architects have been cut adrift into a sea of data without a globalizing theory to guide their decision making. Thus, both current practice and theory are inadequate to direct the design of buildings.

The Limits of Tacit Knowledge

The only way for architects to manage this complexity without ideology is by returning to method as a means of form generation and spatial configuration. However, while the number of constraints that buildings are subjected to have gown, our methods have not changed. Traditionally, architects have relied on knowledge gained from experience to make design decisions (Plowright, 2014). Design ability is the truest form of tacit knowledge. However, many decisions are actually being driven by explicit parameters such as cost, time, zoning, daylight, energy, glare, indoor air quality, etc. Tacit knowledge cannot balance the interactions between thousands of alternatives of explicit parameters. Architects must learn to tie the explicit parameters together with a computational method to manage them (Hazelrigg, 2003).  Only in this way can architects retain control over the design process.

The Conceptual Solution

While various architects have advanced systems based approaches to design, the underlying issue with these theories is that architects do not have a method or system to generate their designs (Noever, 1991). For example, the form of a fighter jet is not function of its systems but the interaction of the parameters necessary to achieve that function (Noever, 1991). Each system is in turned shaped by all the other systems under the overall design function of a fighter jet. It is not systems but the interaction of a system of parameters that should generate design. This embodies the truest expression of a Neo-futurist theory of architecture.

Practically speaking, this means that architects must begin to document methods around explicit decisions in the form of computational algorithms (Hensel, 2010). As each decision is coded into an algorithm with its inputs and outputs, a system of systems will form and begin to intelligently interpret the explicit design problem. Ultimately this will lead to an artificial intelligence capable of working hand in hand with designers to find insight and beauty while maintaining budgets and hitting ever more stringent performance targets (Ahuja, 2015). Artificial intelligence is the key to unlocking the architecture of the 21st of century – a Neo-futurist architecture of rationality and beauty.

Referenced Texts

Ahuja, Sandeep, Patrick Chopson, and John Haymaker. “Practical Energy and Cost Optimization Methods for Selecting Massing, Materials, and Technologies.” ARCC 2015 Conference (2015): n. pag. Web.

Gane, Victor, and John Haymaker. “Design Scenarios: Enabling Transparent Parametric Design Spaces.” Advanced Engineering Informatics 26.3 (2012): 618-40. Print.

Gómez, Alberto Pérez. Architecture and the Crisis of Modern Science. Cambridge, MA: MIT, 1983. Print.

Hazelrigg, G. A. (2003). ” Validation Of Engineering Design Alternative Selection Methods.” Engineering Optimization 35(2): 103-120.

Hensel, Michael, Achim Menges, and Michael Weinstock. Emergent Technologies and Design. Oxon: Routledge, 2010. Print.

Landa, Manuel De. A Thousand Years of Nonlinear History. New York: Zone, 1997. Print.

Noever, Peter, and Regina Haslinger. Architecture in Transition: Between Deconstruction and New Modernism. Munich: Prestel, 1991. Print.

Plowright, Philip D. Revealing Architectural Design: Methods, Frameworks and Tools. New York: Rutledge, 2014. Print.