By Rebecca Edmunds, AIA, interviewing Phil Bernstein, FAIA
My Q+A with Phil Bernstein occurred after he spoke at the Practice Management Luncheon at A’22. His talk, “New value propositions for practice: Models and strategies for innovation,” has been a topic of conversation in the PMKC ever since.
Rebecca Edmunds: What is the most significant challenge to the traditional practice dynamic today?
Phil Bernstein: Architectural practice must define its value proposition for the future; how do we as architects best deliver the value of all we do for clients and convert that value to align with, and take advantage of, the technological developments happening now? Architects do important work but typically get paid as part of the building industry’s commoditized value conversion system. Defining our value proposition—our contributions to the creation of the built environment—is pertinent for two reasons. One is because the definition of the public health and safety is shifting under our feet. It has been a question of life safety and codes, but it’s transforming to include climate change, social justice, and managing life safety across processes, such as the supply chain, as opposed to simply deciding whether the required fire door opens in the right direction.
Second, AI is coming at the profession rapidly with computational platforms powered by artificial intelligence that can think and work autonomously—so much of the low-level transactional work of architecture will become automated, such as drawing coordination, schedules, specification generation, document management, and area calculations. Certain low-level design tasks, like solving routine planning problems, are likely to also be driven by AI. These activities comprise much of what we do as architects. So, what is the value of an architect when machines perform those tasks?
RE: Having come from technical research and development in a Fortune 500 company to architecture, where projects had an overarching budget aimed and aspirational value, I always wondered why the work of architecture is so focused on hourly billing.
PB: In my talk, I argued that we as a profession must get into the outcomes business. Instead of selling time for money, we ought to sell our value as professionals who make things happen. To do that, we must become much better at the analysis and predictive capabilities that make those results possible. Today’s technologies—modeling, analysis platforms, and artificial intelligence—are about predicting outcomes. Not abstract ideas such as, is this thing beautiful or is the economy going to grow? We would do well by setting measurable objectives about important performance characteristics--things clients really care about--building operations, staffing, energy consumption, and even hospital infection rates or the flow of customers through a retail shop, and connecting those outcomes to our business models. Doing so would accelerate our value proposition as architects, rather than continuing to ponder why no one understands what we do or thinks architecture is important—that question doesn't increase our value to clients.
RE: What is the greatest obstacle to overcome to get architects to understand this?
PB: Tremendous intransigence and entropy exists in our profession and in the systems in which we operate as designers. The players in these systems—clients, architects and builders—are in loosely coupled relationships that tend toward disorder, meaning chaos enters the system easily. And architects tend to be resistant to disruptive ideas. Scott Simpson, who ran Hugh Stubbins’s office, always said, ‘architects like to enforce change on everybody except themselves.’
Our methods and tools change, and some experimentation happens around the fringes in alternative practice models like architect design/build or architect-as-developer, but the basic ways we organize the work and our structures of compensation remain very similar to the constructs of the early 20th century; meaning our working systems haven’t radically changed in over 100 years.
No systemic change is possible independently since architects don’t have enough leverage to change things in these systems on our own. So we must re- examine our relationship to the basic players in the process—our clients, and collaborators like engineers and builders—in terms of our value. We’re good at conceptualizing changes and leading teams to make change happen, but we must have a willingness to experiment with how we work. It seems like there is never time to be thoughtful about change: when the economy is good, everyone is too busy to think about innovating. When it is bad, everyone is too busy trying to keep their businesses afloat.
Our industry’s leadership structure relies on six collateral organizations so no one place exists for professionals to find insight or innovation on tackling the challenges technology presents us in a consolidated way, and this makes change even more difficult to identify and sustain. A small practitioner in the Midwest doesn't understand the relationship between her problems and those of a large New York-based firm. I’ve advocated for the industry to form collaborative, cross-discilinary platforms, where people can work collectively on the technological and other challenges we face.
RE: Are there models for firms seeking to embrace change? Are any firms approaching practice differently?
PB: Despite some practice experimentation, nothing has coalesced into a persistent and replicable model. Several books outline alternative architectural practice that attempt different models, but no substantial research has been conducted. Are they sustainable? How have they worked? Are they repeatable?
Architecture practices are often personality dependent, so are difficult to replicate. If new models emerge, they're likely to come from either multidisciplinary giants, like the AECOMS and the Stantecs, or in sustained relationships between large-scale practices, with their clients and builders. Those are good opportunities where digital technology can be applied in interesting ways, and there is the wherewithal to experiment. Traditional practices are not terribly profitable, which inhibits experimentation and innovation. Thus, new ideas are often tested among larger entities and collaborators that have more resources.
RE: What is the most urgent message you'd like to deliver to practitioners about the evolving profession?
PB: We must be very clear about the value proposition of architects in an era where computational automation will make significant inroads in our work. Using a BIM tool to create a model is one thing; having AI create one is entirely different. If we're not completely clear in defining what we add to the process, then automation, which is likely to generate a lot of not-so-good design, is going to erode the edges of practice. For example, consider the effort invested in routine project production tasks in our hourly based, time-for-money business model; these tasks will be easily automated and under attack by AI adopters.
The profession’s history of computational automation is that we gain the efficiency, but we give the value back to clients. For example, everybody uses BIM, but nobody's fees got higher. We used to charge for CAD time but eventually had to give it away. The technology we face today is potentially a point where the system brings more than incremental efficiency. In that scenario, we better have a new value proposition.
RE: In your excellent book, Machine Learning: Architecture in the Age of Artificial Intelligence, you state that ‘intelligent machine automation presents the best chance for architects to change the time-for-money dynamic’ by taking ownership of ‘the numerative and performance-based and the environmental and social implications of design’. However, this requires the profession to collectively create a shared data bank of project work to train AI systems, so we can get the greatest benefit from AI and move into a new model of the value architects bring to clients and communities. Is any entity working on creating this collective base of data? What steps can we as architects take to advance this critical aspect of having an, for lack of a better term, “ownership stake” in AI before another player, such as the construction side of the building industry, runs ahead of us?
PB: My book contains two simple arguments that revolve around the single premise that architects should proactively guide the future of AI technologies lest we be swept up in a destiny over which we have no control. Therefore, we should (1) work with the industry to build the data collectives that are needed to properly train these systems to do useful things, and (2) speak with a single voice about the development of their capabilities. As I mentioned early, these sorts of collective activities are tough to accomplish, but they are, to some degree, existential. AI/ML technology is accelerating so rapidly that most of the output examples in my book—which has only been on the shelves for a little more than a year at this point—are already obsolete, but my core recommendations survive. The potential opportunities and infringements of AI for architecture are closing in fast, and the Institute is now starting to engage in the conversation I’m recommending. Rather than characterize the problem, much like early BIM discussions, as “who wins the technology war” between designers and builders, perhaps we would benefit from a cooperative approach that is mutually helpful across the delivery chain. That’s where I’d like to see our industry leaders take us. And I’m happy to help.
Phil Bernstein, FAIA, is an Associate Dean and Professor Adjunct at the Yale School of Architecture, where he has taught since 1988. He is a former vice president at Autodesk, where he was responsible for the company’s Building Information Modeling strategy that included the development of the Revit platform. Prior to Autodesk Phil was an associate principal at Pelli Clarke + Partners where he managed many of the firm’s most complex commissions including projects for the Mayo Clinic, Washington National Airport, and Goldman Sachs. He is the author of Architecture Design Data: Practice Competency in the Era of Computation published by Birkhauser in 2018, and Machine Learning: Architecture in the Age of Artificial Intelligence by the RIBA Publishing in 2022. He is a Senior Fellow of the Design Futures Council and former Chair of the AIA Contracts Documents Committee.
(Return to the cover of the August 2023 PM Digest)