Architects discuss how artificial intelligence is expanding creative potential while raising questions about authorship, education and human intent in architecture.
Artificial intelligence promises to accelerate design processes and expand creative possibilities. But how can architects harness AI’s potential while maintaining the intentionality and critical thinking that defines great design?
That’s but one question posed recently to an HOK Up Next panel of experts featuring:
- Humbi Song, assistant professor and Emerging Architect Fellow at the University of Toronto.
- Paul Harrison, design principal for HOK’s Canadian practice and assistant professor of architecture at the University of Toronto.
- Loren Supp, principal-design for HOK’s Seattle studio and co-chair of HOK’s Design Board.
The conversation revealed both excitement and caution about AI’s role in shaping architecture’s future. View the full panel talk in the video below, then read the key takeaways.
The Promise and Peril of AI
The conversation began with the panelists broadly discussing what excites them about AI—and what keeps them up at night.
Extra set of hands: Supp discussed how AI can help get designers past the “blank sheet” problem at the beginning of the design process. “What excites me about AI is it really is like getting an extra set of hands,” he said. “It helps us create more ideas we can iterate from and explore, solving that blank-sheet-of-paper problem.”
Supercharged imagination: Song agreed that AI can supercharge the imagination, noting that designers are no longer limited to standard software workflows. “People are being very ingenious in how they integrate AI—beyond simple image generation,” she said. “Tool development has been especially exciting because it lets designers intentionally design their own design process.”
Loss of human thought/reflection: Harrison agreed that AI can be a powerful tool for designers but also warned about how it can erode critical thought and reflection. “Design is primarily driven by reflection. I need the time to sit back and think about something. The dangerous part of working with AI is when you use it to minimize the amount of reflection. For me the question is: How can we maximize the amount of reflection and thought that we’re giving our design?”
AI’s Impact on Design Workflows
The panelists emphasized that AI should enhance—not replace—designers’ individual creative processes.
Evolving design processes: Learning when and why to use AI is now a design skill in itself, noted Supp. “One of our fundamental skill sets is building our own design process,” he said. “AI doesn’t replace that—it supports different moments along the spectrum of the design process.”
Broader exploration with trade-offs: Harrison highlighted how computational approaches broaden the range of work designers can produce, though he acknowledged trade-offs. While parametric and AI tools enable rapid generation of thousands of iterations, they can shift focus away from tactile, reflective moments. “You lose some of the tactility of the design process,” he said, “but building a parametric model probably helps you learn something a little bit better than if you were to just draw it.”
AI as neutral third party: Song described how AI can function as a neutral third party in group settings. When multiple designers are ideating and AI generates imagery, “it almost becomes like a neutral third party that you can criticize without feeling like you’re offending the other designers’ sensibilities,” she noted.
Architecture Education in the Age of AI
The panelists explored how AI is changing architectural pedagogy and its impact on emerging talent, particularly around knowledge retention and the pressure to produce.
Automation changes what we teach: Harrison predicted that increasing automation will fundamentally alter what architects need to know. Tasks that once required intimate knowledge—like manually calculating stair dimensions—have already been automated in software like Revit. As this automation expands, “it puts a new kind of impetus on educators to teach the nuts-and-bolts things that we would presume would happen in practice but isn’t always the case,” he said.
The illusion of knowledge: Song observed what she calls “the illusion of knowledge” among students. “Students are producing so much work, really excellent work. But then there isn’t time to sit and reflect on what’s happening,” she said. When engaging in deeper conversation, she discovers “fundamental gaps in their understanding of what they’ve produced, because it’s been sort of co-written by AI.” To counter these trends, Song has incorporated more analog approaches—pencil and paper—in her classes.
Leveling the playing field: Supp noted that AI offers those coming out of architecture school a leg up in their career as native users of the technology. “It’s an interesting flip that’s happening. Someone much earlier in their career is able to establish the model by which the entire team is going to understand design progression throughout the life of the project. That’s unique.”
How Much AI is Too Much AI? Striking a Balance
The panelists concluded by emphasizing that human intention must remain at the core of architectural design.
Design for human experience: “We design for people. We design for the human experience,” Supp emphasized. “If your design process does not evaluate at its core level how what you’re doing is going to be perceived through time and space and physicality, it’s going to be a failure.”
Human intention matters: Harrison underscored why human authorship matters: “Humans are interested in what other humans do. We like to read into form as architects—to understand the ambitions of the architect through this close reading process. If a building is AI-generated, it’s a whole lot less interesting to me because it lacks intention.”
Intentional spot uses: Song advocated for measured, strategic AI use. “A lot of AI companies create hype language suggesting you can solve everything through AI,” she said. ” Personally, I’m very much advocating for a very slow pace, critical thinking, very intentional spot uses of AI. When you start to daisy-chain AI models together to just speed up the processes, it’s no longer creative. It’s not your work anymore.”
The consensus? AI offers powerful capabilities that can enhance architectural practice, but only when designers maintain control over their process, build in moments for reflection and ensure that human intention—not statistical averaging—drives the final design.