Artificial intelligence is already proving its prowess as a designer. The digital world was the first to embrace AI’s ability to consider hundreds of variables and generate designs based on what it thinks we would like best. Artists have been using AI to develop concepts and even create their pieces. But increasingly AI’s influence is creeping into the physical world too. Zaha Hadid Architects has even started an in-house team that uses AI to help them design workplaces. They have successfully been able to use AI to recommend changes to offices based on data from occupancy sensors. AI has reportedly helped them rearrange furniture and relocate amenities like coffee spots to more popular locations.
Can AI potentially design entire offices in the future by improving the understanding of workers’ needs? To explore this possibility, it’s essential to grasp what AI is and how it functions in the realm of design. Contrary to the misconception that AI software can learn like a human, these advanced programs require training, which relies heavily on vast amounts of data. The specific data used significantly influences the outcomes generated by AI. To envision AI designing an office, it would necessitate access to extensive data not only concerning the usage patterns of a particular office but also a wealth of diverse examples showcasing various office designs and their utilization.
AI faces limitations in interpreting subjective ideas, including notions like “good,” “beautiful,” or, in the context of office design, “productive.” To enable AI to optimize for these concepts, they must be precisely defined. The close relationship between an office and the company occupying it complicates these definitions. To genuinely design an office that enhances workers’ job performance, AI would require access to data outlining individual tasks and their broader impact on the company. This necessitates access to highly detailed company information, which could either be sensitive or unavailable.
Another reason that AI will not likely be the next great office designer is the way that AI is most commonly used in design. Since it is hard for AI to choose which design is “best” it is often used as a tool for what is called generative design. This process uses AI to spit out hundreds or even thousands of different designs. This way designers can look at a wide variety of design options while seeing how they would impact the qualities of a final product. Offices, like anything that is being designed, can be optimized for a variety of different outcomes. AI might be able to spit out different ideas that show how each might perform but in the end a human will need to be able to make the final decision based on which characteristics are the most beneficial.
The effectiveness of AI is contingent upon the quality of the training data it receives, and transforming such data into a practical format for office-related applications poses significant challenges, if not insurmountable ones. While it’s feasible to track people’s physical presence using sensors, gauging their emotional and subjective experiences is an entirely different matter. To truly understand both the behavior and preferences of each office user, qualitative surveys and interviews become indispensable. Once these preferences have been ascertained, AI can assist designers in incorporating a comprehensive array of employee preferences into their decision-making process.
AutoDesk Studio employed this approach while planning their new Toronto headquarters. By inputting the feedback from surveys conducted with their 250 employees into a generative AI software, they could sift through the outcomes to identify solutions that satisfied the most criteria. AutoDesk’s discovery was that instead of designing generic spaces suitable for all teams, the AI proposed dividing workspaces into distinctive neighborhoods tailored for varied use. Creative teams, for example, were provided with more collaborative areas and whiteboard spaces, whereas computational teams were situated around a collaboration area equipped with private single-person work pods.
Artificial intelligence is a new technology, one that has not yet reached its full potential. There seems to be a ways to go until it will be able to synthesize usage data and employee preferences well enough to turn them into a complete office design. “While AI excels at data-driven insights, its ability to generate original creative design ideas remains a work in progress,” explained Jaymie Gelino, chief operating officer at JLL’s Project and Development Services. “Translating AI-generated designs into architecturally accurate plans requires refinement.”
The beauty of artificial intelligence lies in its capacity for continuous self-improvement. If architects and workplace designers can effectively train AI to comprehend what works and what doesn’t for different companies and employees, we may witness its integration as a standard practice in office design. However, even if AI reaches a stage where it can determine the “optimal” office design based on data, human oversight will likely remain essential. Similar to any tool, AI is susceptible to errors if left unattended. Instances of AI “hallucinations,” as seen in applications like ChatGPT, resulting in incorrect responses to even straightforward queries, serve as a cautionary example. While AI has demonstrated its design capabilities, it still has a considerable journey ahead before assuming the primary role in office design projects.