Moving Beyond the Hype: Advancing Common Principles for Responsible AI in the Built Environment

In a short time, artificial intelligence (AI) has come to dominate news headlines, and its applications have proliferated across industries, including a strong and rising interest in AI technologies, research and uses across the many sectors of the built environment. From architectural design, to building information modeling and market valuation, there is an enormous range of current and potential applications across different architecture, engineering, construction and building operations (AECO) industry practices. However, the interest may sometimes be the result of rushed reactions and undue excitement around the technology. Our aim is to bring industry stakeholders together to coalesce around a set of common principles and well-informed practices that enable AI to promote health and well-being in the built environment.

Cutting through the hype while managing the risks
AI has the unfortunate honor of having its own type of hype. “AI hype” refers to the exaggerations that distort our expectations and understandings of AI (LaGrandeur, 2023). These exaggerations result, in part, from grand launches of new AI tools. As an example, we can look to the promotion of Google’s AI model Gemini, whose demo video in December 2023 showcased impressive capabilities, practically holding a full conversation by analyzing images, hand gestures and voice in real time. However, the demo had been heavily edited in post-production to make the tool’s ability to generate them seem faster and more seamless than its actual current capabilities permit. The demo’s description on YouTube reads “for the purposes of this demo, latency has been reduced and Gemini outputs have been shortened for brevity.” This is important information, as well as Google’s explainer on how the video was made using still images. But in a world where our collective imaginations are captivated by the potentially bewildering capabilities of AI, efforts must be made to adequately manage people’s expectations by being transparent and accurate when discussing specific AI tools.

AI hype is also caused by a sense of urgency among CEOs. One study involving 3,000 CEOs across 30 countries found that 75% of them believe that having the most advanced generative AI tools comes with competitive advantage (Capone, 2023). The increasing accessibility of many “generative AI” tools (tools such as Midjourney, ChatGPT and Adobe Firefly, which generate content after receiving short prompts) has meant that the wider public have began using them; and people have discovered potential efficiencies to be gained in the workplace. This has led to a sense of urgency to address the use of AI for many organisations. On the one hand, organisations acknowledge and celebrate that there may be efficiencies gained. On the other hand, organisations are increasingly aware that there are risks involved, such as the improper disclosure of proprietary data through the use of publicly accessible software (Kherroubi Garcia, 2023). With this, some companies have opted to prohibit the use of AI (Ray, 2023), whilst others have opted to create their own generative AI tools (Lu, 2023). Whilst the first option may mean missing out and being left in the dust in increasingly competitive business environments, the latter requires the sort of budget and expertise that are unavailable to many organizations, especially smaller businesses.

But this piece is not about falling behind by missing the AI bandwagon. Quite the opposite. We are here to suggest that the AECO industry reach its full potential by collaborating to thoughtfully and slowly foster responsible AI cultures and behaviours. We want AI to help accelerate better buildings and support our respective missions, promoting health and wellbeing in built environments, and making it easier and more efficient to rapidly advance sustainable practices.

Remaining true to ourselves to advance thoughtful, responsible use
By designing, building, developing and creating the spaces where people spend approximately ninety percent of their lives (EPA, 2023), the AECO industry has the potential to significantly enhance people’s health and wellbeing. This is the premise for the continued development of and research underlying the WELL Building Standard (WELL Standard), and what over 74,000 commercial and residential locations have prioritized by pursuing WELL certification or other WELL Standard achievements. The WELL Standard is a set of evidence based strategies across ten concepts (air, water, nourishment, light, movement, thermal comfort, sound, materials, mind and community) that aim to advance human health and well-being in buildings, communities and organizations. The strategies in the WELL Standard are based on research and principles of being equitable, global, evidence based, technically robust, customer focused and resilient (IWBI, 2024) But this work and the mission to enhance human health and well-being through the built environment requires time and care. As exciting as AI technologies are, and as tempting as it may be to lean into the AI hype, we cannot rush into their development, deployment and adoption. Doing so would risk contravening on the values that the WELL Standard espouses.

Applications of AI technologies have already raised concerns in the built environment. One notable example began in October 2018, when tenants of an apartment complex in New York City were notified by the Division of Housing and Community Renewal (a government agency) of their landlord’s interest in installing facial recognition technology to open the front door of the building. The landlord did not consult directly with tenants, who were concerned about the technology’s intrusiveness. Over 300 residents filed a complaint in January 2019, and the landlord eventually back-pedaled the project (Gagne, 2019). The case has since encouraged a state-wide movement against facial recognition technologies in buildings (DeGeurin, 2023). This case demonstrates the risk to people’s well-being, and potential for loss of trust where use of data is not made clear; where AI is not thoughtfully deployed in the built environment.

Specific to the above case, facial recognition software has exposed substantial risk of bias, leading to serious ramifications if not mitigated. For example, in law enforcement contexts, false arrests have been conducted due to the technology’s error rates among different races and demographic groups, with highest error rates among people of color (Johnson & Johnson, 2023). As a result, regulators and lawmakers are exploring ways to regulate this technology (Witley and Vittorio, 2023), with provisions already set in the EU AI Act (European Parliament, 2024). Conversely, it is clear that integrating AI tools into the built environment can significantly improve our health. Consider that one of the CDC’s recommendations for responding to the COVID-19 pandemic was to improve ventilation in buildings (CDC, 2023). Further, using technology to monitor and remediate indoor environmental quality can enhance human wellbeing (IWBI, 2023). With this, heating, ventilation and air conditioning (HVAC) systems and building management systems that collect and respond to data are extremely important in mitigating against health risks. Crucially, HVAC systems have already been developed with AI applications (AVNET Silica, 2023).

The above examples help refine the challenge of adopting AI in the built environment: how do we embrace AI whilst promoting people-first places?

AI Guiding Principles for Healthy Built Environments
At IWBI, we focus on the impact of buildings on the well-being of their occupants and how organizations can improve their policies and protocols to promote people first places. So, where more organizations are embedding AI in their practices, how do we embrace AI whilst continuing to promote people-first places? We have been investigating this question for some time now, and we have identified the process for reaching an answer.

By embracing the value of the built environment for human well-being, we believe that diverse industry actors and stakeholders across the built environment must collaborate in establishing practical guiding principles for the adoption of AI in the work we all do. Analogizing to how IWBI develops the WELL Standard through its governance process, we know the importance of gathering feedback from a diverse, balanced group of stakeholders to develop strategies that can be practically deployed and lead to the greatest impact. This project will be ambitious. The diversity of practices within the AECO industry on the one hand, and the diversity of applications and techniques involved in AI on the other, lend themselves to an endless list of complex interrelations and questions.

At IWBI, in recognition of the potential for AI to benefit the built environment, we started to focus on educating ourselves about the technology and its potential risks and aimed to contribute to a culture of curiousity and innovation by sharing relevant research with our colleages. We are working with Kairoi, the AI ethics and research governance consultancy, to conduct desk-based research, and deliver internal workshops with our very own AI champions. We are already laying some of the groundwork for relevant, industry-wide guidelines. However, we must collaborate with industry leaders for this endeavour to be meaningful and impactful. It is for this reason that the next step is involving industry experts. We will do so at IWBI’s flagship convening, the WELL Conference in May 2024, during the roundtable called “How can we embrace AI while promoting people first places?” and by encouraging informal conversations about the responsible use of technology to further our people-first mission.

The role for AI in the built environment can be the next big thing for the promotion of people’s well-being. We sincerely hope AECO industry actors and stakeholders will embrace the responsible AI revolution in the built environment.

Header image: Anton Grabolle / Better Images of AI / AI Architecture / CC-BY 4.0