As a BIM Manager I have a big iron ball around my leg with “Windows” carved on it as Autodesk only supports the Microsoft OS for working in Revit. So it was the beginning of May 2024, just a week after Apple released the latest iPad Pro Ad when I picked up on the backlash from the depiction of the industrial press crushes cameras, violins, paint tubes, clay, arcade consoles, notebooks, and every tactile creative tool imaginable.
Many people who have a working relationship with tools and culture felt their heart sink looking at the short video. Instead of thinking about all the tools that “fit” inside the iPad, it landed as a metaphor for creativity itself being compressed, flattened, and sanitized into a single digital inter(sur)face.
The Efficiency Trap
One of the promises of BIM is to free architects from coordination chaos. Then AI is at the door with automated repetition in its back pocket. Both have delivered on efficiency — but efficiency isn’t neutral. Each system encodes assumptions about what counts as good practice: accuracy, speed, and cost control over ambiguity, experimentation, or story.
Design Intent cannot become another field in a parameter list, as this instantly creates a grey zone of authorship. The process shifts from why we design to how we optimise. Architecture becomes data-rich, the favorite spin word of software salespersons, but quite possibly idea-poor.
Expression / Automation
In Filterworld, Chayka warns that algorithms flatten culture by rewarding what’s already popular. Every tool carries an inherent logic that shapes the outcome of its use — when you hold a hammer, everything looks like a nail. And today, the algorithmic logic behind our software is more pervasive than ever.
Most BIM authoring tools reward repeatable logic: families, templates, scripts. Add AI tools into the mix — trained on vast datasets of existing buildings — and you end up generating forms that simply echo what has already been built.
The result? Homogenised, coordinated, and eerily predictable environments. The more our digital models align, the more our neighbourhoods and cities start to resemble a dystopian sci-fi set: polished, optimised, and strangely interchangeable.
Information Without Intent
Intent in architecture has always been messy. It’s intuition, conviction, and critique wrapped into form. But digital workflows make it dangerously easy to skip that stage.
An AI-assisted plan might satisfy every metric and still feel hollow because the “why” has been automated out of the process. The more we optimise, the less we express.
What’s left are perfectly acceptable buildings that don’t seem to stand for anything.
The Filtered City
Chayka’s Filterworld describes a cultural landscape where everything feels polished, familiar, and ultimately interchangeable. Parts of our built environment are heading the same way. You only have to compare the City of London with Canary Wharf at the Docklands to see two different eras of design logic — one messy, layered, accidental; the other smooth, optimised, and algorithmically beige.
The City grew over centuries: awkward junctions, odd lots, medieval street patterns stitched together with steel and glass. It’s chaotic, contradictory, and full of intent — the product of countless human decisions colliding over time. Canary Wharf, by contrast, is the architectural equivalent of a clean spreadsheet: gridded plots, coordinated façades, interchangeable lobbies, value-engineered public spaces. It performs well on paper but often feels like it was generated from the same brief repeated block after block.
As BIM and AI-driven workflows spread, more places risk drifting toward the Canary Wharf model: efficient, compliant, and instantly forgettable. Like a digital feed tuned by metrics, the contemporary city becomes a loop of familiar forms.
Reclaiming the Process
If algorithms flatten design, it’s because we’ve quietly allowed them to decide what matters. Reclaiming architecture doesn’t mean rejecting digital tools; it means reclaiming ownership of the design process.
AI is most useful as a provocateur rather than a producer. Its value lies in raising questions, testing assumptions, and exposing blind spots — not in delivering finished answers that bypass judgment.
BIM, despite its reputation for neutrality, is never neutral. It encodes priorities and workflows, and those can be challenged. When cultural cues, conceptual moves, and narrative intent are embedded into the model, it stops being a coordination container and starts becoming part of the design conversation.
The same applies to data. Systems trained only on global precedents inevitably reproduce global sameness. Broaden the inputs with local stories, ordinary buildings, and contextual quirks, and the outputs begin to shift.
And friction must remain a feature, not a flaw. If the model always agrees with you, if everything resolves too easily, then design has already ended and parameter management has taken its place.
Reclaiming ownership of the process means deciding where optimisation stops and intent begins. Technology can support that decision — but it should never be allowed to make it for us.