Can data help designers anticipate human factors over time as buildings learn?

Building sensors and data-collecting personal devices are providing designers with a new layer of input on how spaces are used. Harnessing this data to achieve more human centered outcomes depends on developing new metrics of architectural usability; a process that is likely to be driven by social research, not new tech.

In the newly overlapping domains of physical and digital space, designers have gained new input sources to inform and then reshape the hybridized world of interfaces vying for our attention. The prospects of blending UX methods and design approaches aimed at creating a common space between the domains of screens and physical space led me to a talk at the Center for Architecture in New York. Entitled “AI and Sensemaking: Human-Centered Design in the Age of Abundant Data“, the panelists ranged from data scientists, multi-media creators and workplace consultants. Presentations were given on the impact that data abundance is having on human centered design practices. Informed or at least propelled by a torrent of new user data, the case studies included examinations of metrics for measuring occupant satisfaction and interactive data visualizations. I arrived wondering, what could a data driven design practice bring to human centered design? Across the board, the practices described were searching for identifiable markers of success in achieving an optimized human centered design metric. The challenges are compounded by the nature of working within an expanding and hybridized field. My sense was that the potential of the methods overshadowed the work examples themselves. So I decided to reframe my thinking around a more preliminary question, how might data driven design tools help us develop more human centered environments?

Occupant data collection coupled with AI powered generative design can utilize high level goals and constraints to provide designers with thousands of possibilities, drawn up instantly. When used to optimize buildings for energy or structural efficiency, the results are definitive. Will the same hold true when aiming for softer outcomes, such as greater occupant emotional wellness, ergonomics and social cohesion? Can data help designers anticipate human factors over time as buildings learn? This new abundance of both data and data driven compilations promotes the quick conclusion that as users become measurable in more ways designers will gain new purchase to bend the interfaces of spatial and informational experiences to better suit user needs. New data streams do not however provide a workaround for the established sequential cadence of invention. Iteration, a core methodology of human centered design, implies better communication. It is a deeply conversational practice, rooted in the social sciences and it seems unlikely to be accelerated or improved by more raw data in itself. Data driven design practices, poised to drink from new torrents of building occupancy feedback, should not make the mistake of believing that more data inputs will provide any back doors to sensemaking.

One of the speakers at the talk, an organizational sociologist, described how the collaborative spaces of many new workplaces are touted as bona fide products of data-driven design, and therefore considered highly innovative. These spaces are seldom used as anticipated and many go unused entirely. Overconfidence in data driven design compounds one of my favorite issues in architecture: a core belief in the ability of prediction. Adding new input streams and design methodologies makes it clear that the objectives of achieving new levels of usability is difficult without involving the the social sciences to build meaningful criteria for and around the objectives of designing better usability for users. When developing interfaces and measures of success for our hybridized built-digital environments, design firms should proceed by first taking a user experience view of technology.

Another reason that more user data doesn’t equal better spaces of interaction is that the systems we’re looking at are dynamic and evolving. They are built for a future that doesn’t exist, even if the designers who build them don’t acknowledge that. As Stewart Brand put it in How Buildings Learn “All buildings are predictions. All predictions are wrong”. In thinking about a framework for constructively harnessing user data, I’ve modified Brand’s famous diagram to include a new inner layer of Occupant Data. This newly visible and innermost layer moves faster than all the others.

My addition to Stewart Brand’s building layers diagram showing the occupant data layer

Melissa Marsh, founder of the workplace consultancy PLASTARC and speaker at the talk frames the situation optimistically. She puts forward that the optimization of architecture for human factors (physical, social, cognitive and neurological), will be embraced enthusiastically, as we have seen with modeling for ecological and environmental factors in recent years. In the search for coherent practice around engaging user data in the design of physical space that truly is human centered, I find it useful to borrow from the 10 usability heuristics developed digital interfaces. The usability studies performed by the Nielson Norman Group, although focused on screen based applications, provide the type of bench-marking needed to guide us forward into practical agreements about the use and real utility of occupant data in human centered design.

For a look into how user data is poised to reshape another market, coliving, here is a post summarizing an article co-written with my colleague and building technology consultant, Tim Lehotsky. The guiding principals listed at the end of that post provide some indication of where we are headed in our research.