Exploring the evolving relationship between architects and AI with Gary Barley

As one of Women in BIM’s newly appointed global AI Ambassadors, Gary Barley, Technical Director (Digital) at Pascall+Watson is helping shape the conversation around how Artificial Intelligence can be adopted responsibly, creatively and effectively across architecture and the wider built environment.

Through the AI Ambassador programme, Women in BIM is bringing together recognised industry leaders to support Members as they navigate one of the most significant technological shifts facing the profession. By sharing practical experience, research and real-world insight, the Ambassadors will help bridge the gap between emerging AI technologies and their everyday application across the AEC sector.

Drawing on his experience as Technical Director (Digital) at Pascall+Watson, Gary brings a unique perspective on integrating AI into live architectural practice while maintaining creativity, design intent and professional responsibility. In this interview, he explores how architects can embrace AI with confidence, where the greatest opportunities lie and why human expertise will remain central to the future of design.

Within architectural practice, AI is often framed as either a creative accelerator or a threat to authorship. From your experience, where does it genuinely enhance architectural creativity rather than constrain it?

AI genuinely enhances creativity when it serves as a partner to human intent, not a replacement. To maintain authorship, designers must begin with a strong, self directed vision, whether rooted in a hand sketch, a physical model, a collection of images or rigorous spatial guidelines.

In early stage design, AI acts as a powerful conceptual accelerator. Based on initial sketches/input, it rapidly generates formal variations for the architect to critique uncovering unexpected combinations of ideas that might never have emerged organically. This can lead us toward richer, more diverse design directions.

Beyond creativity, using AI to automate routine repetitive tasks, can allow architects to redirect more time and energy toward high value creative problem solving.

Ultimately, AI acts as a powerful amplifier for architects who already possess strong conceptual foundations and critical judgment. However, the risk remains that it could constrain the creative development of those still learning to cultivate their own design voice.

There is growing concern that AI could homogenise design outcomes. Do you see evidence of this in practice, or is AI actually expanding the range of architectural possibilities?

Homogenisation is a risk, but it is not inevitable. To avoid default outcomes, the modern designer must treat AI as just one tool within a broader, hybrid workflow.

Initiate workflows with good quality text prompting and hand-drawn sketches or specific 3D model views. This forces the AI to adhere to a unique composition rather than a model’s default layout. Furthermore, progressive practices are training proprietary AI models on their own unique style guides and material palettes to ensure distinct, branded outputs.

Finally, human curation and traditional craft, such as post production refinement, remain essential. AI only homogenises design when we let it do all the thinking. By pairing unique source inputs and custom-trained models with human oversight, we break the repetitive loop and actually expand architectural possibilities.

You’ve been directly involved in embedding AI tools into live design workflows. What has been the most unexpected benefit or challenge that has emerged from real world use?

The greatest benefit has been the “wow” moment when staff realise AI’s practical value. It is incredible to see a designer watch their sketch come to life, create the minutes of a meeting from a recorded transcription, or a technical team member instantly resolve errors in their code.

Conversely, the greatest challenge lies in education and adoption. At Pascall+Watson, we have a vast spectrum of roles and digital skillsets. While early adopters are already pushing boundaries, others are still finding their footing.

Because AI’s applications are so diverse, a one size fits all training program simply doesn’t work. My ongoing challenge is designing tailored educational pathways that resonate with different learning styles, ensuring every team member, regardless of their starting point, can confidently integrate AI into their daily workflows.

Many architects remain cautious about AI adoption. Is this caution a rational response to risks around authorship, ethics and liability, or is it more of a cultural lag behind technological capability?

Caution is a rational response to risk, not a cultural lag. Architects are legally responsible for all decisions and documents, prioritising public health, safety, and welfare. Their designs must balance visual inspiration and accessibility with structural integrity, sustainability, and regulatory compliance.

A lot of AI solutions operate in a “black box” and cannot not explain the logic or reasoning behind their answers, while professional liability insurers do not cover “algorithmic error.” Until legal frameworks establish clear boundaries of liability for AI-assisted design, architects must remain cautious about using AI for final deliverables.

The architects who survive the AI transition will not be those who blindly adopt it, nor those who reject it, but those who rationally integrate it as a highly sophisticated tool.

How should architectural practices think about “design intent” in an AI-augmented workflow? Does AI change who is designing, or does it simply extend the architect’s capability?

In practice, we see AI acting as both a capability extender and an author, depending on its deployment.

For research, documentation support, and iterating defined concepts, AI is an incredible tool. It extends what an architect can achieve by handling complex optimisation.

The shift in ‘who is designing’ occurs when we use AI to blindly generate initial concepts, spatial organisation, or aesthetic direction, before thinking through the problem ourselves. When the AI generates the concept and the architect merely selects and refines it, the creative process shifts from creation to curation.

While this curation led workflow can produce visually striking results, it is fundamentally different. If we use a “blank page” text prompting approach and rely on AI to bypass the struggle of the initial ideation, we risk losing the very design intent that defines the profession.

Looking ahead, will AI become a competitive differentiator between practices or will it become a baseline expectation similar to BIM and what will determine which path the industry takes?

While both began as competitive edges, their adoption differs. BIM was shouted from the rooftops to signal collaboration, whereas AI has been a quieter, almost secretive assistant boosting internal productivity.

Today, BIM is a baseline expectation; clients define the deliverables, and if you can’t provide them, you can’t bid. AI is not yet a client requirement, but it enables more competitive bidding by driving internal efficiencies.

Ultimately, AI will follow both paths. It will become a baseline tool, but true competitive differentiation will lie in how it is integrated. Many firms will treat AI as a tick-box exercise, just as they did with BIM, without transforming their practice. The true differentiator won’t be the technology itself, but the organisational discipline to use it well.

Hear more from Gary at the Women in BIM Conference London on Thursday 15 October. Register today to join Gary’s AI in Action Workshop.