ARTICLE 20

You Can’t Compete With AI. So Stop Trying.

The future of architecture isn't human versus machine. It's what happens when we combine the two.

Is design dead?

Is art dead?

Is music dead?

If artificial intelligence can generate an image in seconds, compose music, analyze thousands of pages of information, write code, create design variations, and increasingly interact directly with the software architects use every day, what exactly are we supposed to do?

Compete with it?

Good luck.

You cannot compete with a machine that can potentially access more information than any individual human being could consume in a lifetime.

I don't care how good of a designer you think you are.

And I'm not saying that because I don't believe in design or because I haven't spent years doing it. I've worked on and designed hundreds, if not thousands, of projects over my career, including work that has received recognition and awards.

I love design.

But there is also a point where confidence turns into ego.

You are not personally going to read every architecture book ever written, study every building ever constructed, examine every architect who ever practiced, understand every material assembly, memorize every building code, know every engineering discipline, absorb every artistic movement and retain all of it simultaneously.

A machine can increasingly work across enormous amounts of that information at a speed no human being can match.

Pretending otherwise doesn't make us more creative.

It just makes us stubborn.

It would be like jogging on the treadmill at 7.5 miles per hour a few mornings a week and deciding you're ready to line up against the fastest runner in the world.

I respect the confidence.

But I wouldn't put money on you.

So I think we are asking the wrong question.

The question isn't:

How do humans compete with AI?

The question is:

What happens when we combine the power of the machine with the power of the human mind?

That is where things get interesting.

AI Is Not Your Brain. It Is an Extension of It.

There seems to be a perception that using artificial intelligence means pushing a button and letting a computer think for you.

That hasn't been my experience at all.

I see it much more as an extension of my brain.

You give it information. You challenge it. You question it. You correct it. You give it context. You tell it what matters and what doesn't. You push ideas through it, pull them apart, reconstruct them, and sometimes realize the first idea was terrible.

The quality of the output still depends enormously on what you put into it.

I've seen people interact with AI systems they've barely given any context to, and the thing responds like a valley girl from Clueless.

That's not necessarily the technology failing.

You haven't taught it anything about how you think.

Everyone will eventually have access to versions of these tools.

That doesn't mean everyone will get the same result.

Having AI may eventually be like having Google. The competitive advantage isn't simply having access to it.

It's knowing what to ask.

Knowing what information matters.

Knowing when the answer is wrong.

Knowing when to push further.

Knowing when to ignore it completely.

The real advantage becomes:

experience + context + curiosity + judgment + taste + verification.

When used that way, AI becomes something much more powerful than an answer machine.

It becomes a way of extending thought.

And for me, that has been incredibly personal.

Dyslexia Changed the Way I See This Technology

I've struggled with dyslexia my entire life.

Writing has never been the easiest way for me to communicate.

I know what I want to say. I can usually see the entire idea in my head. I'm extremely visual and tend to think about problems from the big picture inward.

But getting those ideas out grammatically, organizing them into a clean sequence, spelling everything correctly, and turning those thoughts into writing can take me two or three times as long as it might take someone else.

I can do it.

It just costs me more mental energy.

There have been plenty of times when I knew exactly what I wanted to communicate but dreaded having to sit there for an hour writing an email that someone else could probably write in fifteen minutes.

So yes, if I can explain exactly what I want to say and use a tool to organize those thoughts into an email in thirty seconds, I'm going to use it.

There's no shame in the game.

Congratulations, you're the best email writer in the office.

Here's your trophy.

Meanwhile, I would rather spend that hour solving the actual problem the email was about.

What exactly are we protecting by refusing to use the tool?

The hour?

The spelling mistakes?

The frustration?

If the ideas are still mine, the opinion is still mine, and I'm directing what gets communicated, what have I actually lost?

For someone like me, AI can be an equalizer.

People who were naturally strong writers have always had an advantage because they could communicate their ideas more easily.

Now my ideas can flow closer to the speed at which I actually think.

That matters.

And it raises a larger question.

Maybe intelligence was never supposed to mean who could memorize the most information or write the cleanest paragraph.

Maybe those were simply abilities that our previous tools rewarded.

If machines increasingly handle retrieval, organization and repetitive production, then human intelligence may become even more about synthesis:

Seeing connections.

Questioning assumptions.

Recognizing patterns.

Understanding people.

Identifying nonsense.

And knowing what actually matters.

A Circular Saw Is Still a Tool

Imagine you're cutting a board.

You can take a hand saw and spend fifteen minutes cutting through it.

Or you can grab a circular saw and finish the same cut in seconds.

Nobody walks onto a construction site and says:

"That doesn't count. You cheated. You used a power tool."

We understand that tools increase human capability.

The person still has to know where the board needs to be cut.

They still have to measure it.

They still need to understand the assembly.

They still need judgment.

They still need to know when they're about to cut the wrong damn board.

The circular saw simply removes unnecessary effort from the process.

AI is beginning to do the same thing with knowledge work.

And architecture contains an enormous amount of unnecessary friction.

Not Every Task Deserves to Be Protected

This may be uncomfortable, but a lot of professional work is made up of mediocre tasks.

Not mediocre because the people doing them are mediocre.

The tasks themselves are.

Moving information from one place to another.

Formatting schedules.

Writing routine emails.

Renaming views.

Organizing sheets.

Producing meeting notes.

Searching through hundreds of pages for one code requirement.

Copying information between programs.

Adjusting repetitive model elements.

For years, being fast at those things had value because somebody had to do them.

But we shouldn't confuse being good at a repetitive task with that task having some sacred creative value.

If a machine can do in thirty seconds what used to take me thirty minutes, I don't feel threatened by that.

Take it.

Please.

I didn't become an architect because I had a lifelong dream of renaming sixty-seven Revit views.

Efficiency is not cheating.

It is the entire reason we invent tools.

Why Am I Spending an Hour Looking for One Sentence?

Think about building codes.

Building codes contain hundreds and hundreds of pages of interconnected requirements, references, exceptions, tables, definitions and cross-references.

Architects aren't walking encyclopedias.

When we need an answer, we look it up.

Sometimes you know exactly where to look.

Other times you're sent on a wild-goose chase through five different sections to eventually find one sentence telling you that a particular occupancy requires a particular separation.

If technology can help me identify the relevant provisions in three minutes instead of an hour, why wouldn't I use it?

The responsibility hasn't disappeared.

I still need to verify the actual code section.

I still need to understand whether it applies.

I still need to exercise professional judgment.

But I'm spending my time thinking about the requirement instead of searching for the requirement.

Those are two very different activities.

And architecture needs far more of the first one.

AI Can Also Be Spectacularly Wrong

This is where the conversation needs some balance.

AI can be wrong.

Really wrong.

And sometimes it can deliver that wrong answer with the confidence of a guy at a bar explaining something he learned five minutes ago.

That is dangerous.

Especially in architecture.

Building codes.

Life safety.

Structures.

Specifications.

Contracts.

Accessibility.

Building science.

A convincing answer is not necessarily a correct answer.

That is exactly why expertise still matters.

If you don't understand the subject well enough to question the machine, AI can simply help you be wrong faster.

That isn't intelligence.

That's accelerated stupidity.

The professional still has to verify the information, understand the context and take responsibility for the decision.

I don't see AI replacing that responsibility.

I see it changing the amount of information we can process before making the decision.

This Is Already Entering the Tools We Use

This isn't some theoretical future that might happen decades from now.

We're already seeing natural-language assistance, automation, generative design systems, model analysis and AI-supported workflows move into the software architects and engineers use.

Tasks that once required a series of manual operations are increasingly becoming things we can simply describe.

Think about where that goes.

Today:

"Create and organize these views."

Tomorrow:

"Review this model and show me potential accessibility conflicts."

Then:

"Compare the architectural and structural models and identify the major coordination issues."

Eventually:

"Here are the project constraints, zoning requirements, budget, site orientation, structural system, energy targets and owner priorities. Show me several viable approaches and explain the tradeoffs."

The architect does not necessarily disappear.

The architect's job moves further toward judgment, direction, synthesis and decision-making.

That is a very different profession from clicking buttons all day.

And frankly, I think that's a better profession.

The Most Interesting Part Is Design

I've started experimenting with AI not simply as something that produces images but as a design partner.

That's an important distinction.

I don't want a machine to tell me what the building should be.

I want to use the machine to challenge what I think the building should be.

What happens if this moves?

What happens if the geometry follows this pattern?

What happens if we reverse the circulation?

What precedent exists for this?

What am I missing?

What are the weaknesses in this concept?

Give me the argument against what I just proposed.

What happens environmentally?

Structurally?

Experientially?

Economically?

Those questions still originate with the designer.

AI simply expands how quickly you can investigate them.

And this is where I think designers need to set their egos aside.

I don't care how naturally talented you are.

I don't care how many beautiful buildings you've designed.

I don't care how many followers you have or how many design awards are sitting on the shelf.

No individual designer can personally hold the collective design knowledge of the entire world in their head.

That doesn't mean the machine is automatically a better designer.

It means it can give a good designer access to an extraordinary field of information and possibilities.

There is a difference.

The goal isn't to ask AI:

"Design me a cool building."

The goal is to combine what it can process with what you understand.

That's the hybrid.

If Everyone Can Produce, What Becomes Valuable?

This may be one of the biggest changes coming to architecture.

For a long time, part of an architect's value came from our ability to produce information other people couldn't easily produce.

Drawings.

Renderings.

Models.

Details.

Specifications.

Calculations.

Presentations.

Those things required specialized software, specialized knowledge and enormous amounts of time.

What happens when producing them becomes dramatically easier?

The value moves.

If everyone can generate an image, generating an image isn't enough.

If everyone can produce twenty design options, producing options isn't enough.

If everyone can draft faster, drafting speed isn't enough.

The question becomes:

Which option should we build?

Why?

What is appropriate for this site?

What actually meets the client's needs?

What can be constructed?

What will age well?

What will leak?

What will cost twice what the client can afford?

What looks incredible in a rendering but makes absolutely no sense in reality?

That is judgment.

And judgment comes from experience.

AI doesn't make everyone an expert.

If anything, it may make it easier to expose who actually understands what they're doing.

Someone with real experience can look at an AI-generated solution and say:

No.

That wall assembly is wrong.

That won't drain.

That isn't accessible.

That detail isn't constructible.

That doesn't address the client's actual problem.

Try again.

The machine can generate possibilities.

Expertise determines which ones survive.

We Have Seen This Movie Before

There was a time when many people viewed the internet as a novelty.

Then the world connected.

Knowledge that once required a trip to a library became available from a desk and eventually from something we carry in our pockets.

People could communicate across continents instantly.

Entire industries emerged.

Education changed.

Commerce changed.

Media changed.

Work changed.

Communities formed between people who otherwise never would have met.

And no, it wasn't all positive.

We also somehow managed to turn dating into something resembling ordering a shirt on Amazon:

Scroll through the options.

Check the pictures.

Read the reviews.

Maybe return it in thirty days.

But that's a conversation for another article.

The point is that transformative technologies are rarely understood when they first arrive.

We judge the early version while imagining the future will look exactly like it.

It won't.

Architecture Has Another Problem: We Hoard Knowledge

This is actually one of the areas where I think AI could change architecture for the better.

Throughout my career I've worked around different sides of architecture, engineering and construction.

And something has always bothered me.

We hide an enormous amount of knowledge from one another.

Every company has its details.

Its specifications.

Its standards.

Its methods.

Its internal library.

And I understand why.

Companies spent years developing those resources. They represent intellectual property and experience.

But there is a downside.

As an industry, we repeatedly solve the same problems independently.

Go research something as basic as where a vapor retarder belongs in a particular wall or slab assembly and you can quickly enter a world of contradictory information.

Put the insulation here.

No, put it there.

Put the vapor retarder above this.

No, put it below that.

Vent this.

Don't vent this.

I've talked before about the problem of protecting exposed exterior foundation insulation.

It sounds like an incredibly basic building-envelope detail until you actually start researching how everyone handles it.

Then suddenly everyone has a slightly different answer.

Why?

Some of that is climate.

Some is project-specific.

But some is simply fragmented knowledge.

Architectural and engineering firms frequently treat their details like they contain the secrets of the universe.

Having worked with drawings from many different companies, I can tell you something:

We're not all inventing completely new buildings every Monday morning.

Details migrate.

Someone works at Firm A.

They learn something.

They move to Firm B.

That knowledge comes with them.

Someone changes the detail.

Someone else changes it again.

Sometimes we improve it.

Sometimes we copy the same bad idea for twenty years.

The profession would benefit enormously from more collective examination of what actually works.

Bring the information to the table.

Compare it.

Debate it.

Test it.

Improve it.

I Learned This Lesson in High School

This reminds me of a kid I sat next to during an exam in high school.

I was trying to copy off him.

He was guarding that test like he possessed classified government information.

Wouldn't let me see anything.

He ended up getting around a 30.

I got an 85.

So apparently he did not possess the answers to the universe after all.

Architecture occasionally reminds me of that.

There absolutely are things that deserve intellectual-property protection.

But we should also recognize the difference between proprietary innovation and basic industry knowledge that could make buildings better if more professionals were willing to examine it openly.

Other technical professions have long traditions of publishing research, criticizing it, reproducing results and developing broader bodies of shared evidence.

Architecture does some of this through universities, building-science research, professional organizations and technical publications.

But enormous amounts of practical knowledge remain scattered between individual firms.

AI has the potential to connect more of those fragments.

And that's exciting.

What About Jobs?

This is the uncomfortable part.

Will some jobs change?

Absolutely.

Will some tasks disappear?

Almost certainly.

Could some positions that are primarily built around repetitive production be affected?

I think it would be naive to pretend otherwise.

Efficiency has consequences.

It always has.

CAD changed drafting rooms.

Digital photography changed photography.

The internet changed newspapers, retail, travel agencies, advertising and countless other industries.

Technology doesn't normally ask whether we're ready before it changes the economics of a profession.

The answer can't simply be:

"I don't like this, therefore I'm not going to use it."

The better question is:

What becomes more valuable when this task becomes cheap?

That's where I want to position myself.

The More Artificial Things Become, the More Human Things May Matter

There is another side of this that I don't think receives enough attention.

If production becomes easier, human relationships may become more valuable.

A homeowner doesn't simply need someone capable of generating a floor plan.

They need someone who can sit across the table and understand why the house they've lived in for fifteen years suddenly doesn't work anymore.

They need someone to notice that two spouses are describing the same problem in completely different ways.

Someone needs to walk the site.

Stand in the space.

Understand the neighborhood.

Talk to the contractor.

Deal with the building official.

Explain the tradeoffs.

Tell the client when an idea isn't worth the money.

And sometimes recognize that what the client is asking for isn't actually what they need.

A machine can process information.

Understanding a person is something different.

The more production becomes automated, the more valuable the genuinely human parts of architecture may become:

Trust.

Communication.

Empathy.

Leadership.

Taste.

Judgment.

And accountability.

Those aren't secondary skills.

They may become some of the profession's most important skills.

But What About Creativity?

If AI can generate a building, an image, a song or a painting, what happens to human creativity?

I think creativity becomes more important, not less.

Because generating something isn't the same as knowing whether it's good.

Having 10,000 possibilities actually creates another problem:

Which one matters?

Which one fits the place?

Which one fits the client?

Which one can actually be built?

Which one feels right?

Which one solves the problem instead of merely looking impressive?

Which one should never leave the computer screen?

Information is not judgment.

Computation is not wisdom.

Possibility is not intention.

Those remain deeply human responsibilities.

The designer increasingly becomes the person directing an enormous field of possibilities rather than manually producing every possibility individually.

That requires taste.

Experience.

Curiosity.

Empathy.

Technical understanding.

And perhaps most importantly:

Knowing which questions are worth asking.

So Is Using AI Cheating?

Is using Revit cheating because someone once drafted the same building by hand?

Is using CAD cheating because someone once used a parallel bar?

Is using a calculator cheating because someone else did calculations with a slide rule?

Is using Google cheating because somebody once spent an afternoon digging through a library?

Is using a circular saw cheating because somebody else used a hand saw?

At some point this becomes ridiculous.

Technology removes friction.

What matters is what we do with the time and capability it gives back to us.

There are legitimate questions about artificial intelligence.

Copyright.

Privacy.

Bias.

Accuracy.

Authorship.

Liability.

Job displacement.

The concentration of information and power.

Those deserve serious discussion.

But pretending the technology is going away because it makes us uncomfortable isn't a strategy.

The Hybrid Designer

I think the future belongs to what I would call the hybrid designer.

Not the machine.

And not the person stubbornly pretending the machine doesn't exist.

The hybrid.

Human imagination combined with computational scale.

Experience combined with rapid access to information.

Intuition combined with analysis.

Creativity combined with iteration.

The ability to zoom out and understand the human problem while using machines to process things no individual human could process alone.

AI does not eliminate the value of knowing things.

It increases the value of knowing what matters.

It doesn't eliminate expertise.

It changes where expertise becomes valuable.

It doesn't eliminate creativity.

It gives creative people a considerably larger canvas.

And it doesn't make everyone equally capable.

Give ten people the same camera and you don't get ten Ansel Adamses.

Give ten people the same guitar and you don't get ten Jimi Hendrixes.

Give ten architects the same AI system and you aren't going to get ten equally good buildings.

The tool is not the mind using it.

Buckle Up

Architecture is going to change.

I don't know exactly how far it goes or exactly what the profession looks like twenty years from now.

Nobody does.

But I am fairly confident about one thing:

Trying to beat the machine at being a machine is probably a losing strategy.

I don't want to spend my career competing with software to see who can organize a sheet faster.

Let the software win.

Congratulations.

It is officially the world's greatest sheet organizer.

I want to spend that time solving the building.

Understanding the person who will live there.

Questioning assumptions.

Exploring another possibility.

Learning something.

Designing something better.

The machine can process.

The human can determine what is worth processing.

The machine can generate.

The human can decide what deserves to exist.

Put the two together and we may be entering one of the most interesting periods architecture has seen in a very long time.

So no, I don't think designers should be afraid of AI.

I think we should learn how to drive it.

Because the circular saw is already sitting on the table.

You can keep using the hand saw if you want.

I'm plugging mine in.

Michael Demetrios Frudakis, AIA, NCARB
Principal Architect