Every new technology gets the same doomsday conversation. I have been having it for over 30 years. in the 1990s builders wouldn’t give up their glossy paper brochures for a website. then came PV Solar and the pushback was the ﬁrst cost was too heavy a lift, and it felt too complicated to manage. VR made people dizzy, it was too expensive to create, and visualization computer programmers were far and few between.
With time, every one of those objections turned out to be a trivial ﬁx.
Now it’s AI. And here’s what it looks like from my perspective: I’m doing the same kind of work and providing the same services as when I had lots of employees — but the quality is better, I deliver it faster, and for less money.
I wrote up the full pattern — and why I’m not that concerned — in this article below.
Change is inevitable. Doing it well is a choice.

The Doomsday Conversation

Every few years, a client sits across the table from me, worried about the new technology that is about to change everything.

I know the conversation by heart, because I have been having it for over thirty years.

In the 1990s, it was the internet. Builders did not want to give up their glossy paper brochures. A website felt cheap next to a heavy, four-color brochure a buyer could hold — they were sure no one would ever choose a home from a screen. Then it was solar. I wrote my thesis on solar design in 1982, and for decades after, owners told me the same two things: the first cost was too heavy a lift, and the systems felt too complicated to manage. Then it was virtual reality and the metaverse. Put a headset on a client and walk them through a building that did not exist yet, and the pushback was simple: it made people dizzy.

And with time, every one of those objections turned out to be a trivial fix. Websites replaced the brochure rack. Solar got cheaper, and easier to own. The headsets stopped making people dizzy. The technology kept moving; the fear just kept changing its reason.

Now it is AI. The technology changes. The fear does not.

What actually happens

The pattern has been the same every time:

The technology arrives anyway. It does not wait for a vote. The only choice is whether you meet it early, on your own terms, or late, on someone else’s.
The people doing the work learn by doing. Not by debating it — by implementing it, making mistakes at small scale, and getting better.
The guardrails get built. This is the part people forget. Building codes were not written before there were buildings. They were written because of buildings — by practitioners, after experience, enforced by licensed professionals who carry responsibility for the result. AI will be no different. The standards will come from the work, and they will be carried by the professionals accountable for it.

I am not that concerned about AI. I have watched this cycle complete itself too many times.

The risks are real

Let me be clear: not being concerned is not the same as being naive. The risks are real, and I can see several of them from my own desk.

AI gets things wrong — confidently. It will hand you a plausible detail, a clean paragraph, a convincing number, and be wrong. In my world, a wrong answer isn’t a typo. It can be a failed inspection, a construction defect, or a life-safety problem.
It eats the bottom rungs of the ladder. If AI does the junior work, where do the next senior architects come from? A profession that stops training its young people is borrowing against its own future.
It tempts firms to cut the review instead of the production. The value was never the drafting hours. It is the judgment — and judgment can’t be delegated to a tool. My stamp means I am accountable. That does not change.
Ownership is unsettled. Whose work trained these models, and who owns what comes out of them? For a profession built on authorship, that is not a footnote.
And at the far end, the rogue scenario. People far deeper in this technology than I am worry about systems that eventually stop taking instructions. I don’t lose sleep over it — but dismissing it outright would be exactly the naivety I am arguing against. It is the strongest case there is for building the guardrails now, while these systems are still tools and we are still the ones holding them.

Those risks are exactly why implementation needs standards, review, and accountable professionals. A risk does not go away because you refuse to touch it. It just gets managed by somebody else, on their terms.

Here is what it looks like from my desk

I am doing the same work and providing the same services as when I had 33 employees. But now the quality is better, and I deliver it faster, and for less money.

That is not a prediction. It already happened.

AI did not replace the architect. It removed the layers between the client and the architect’s judgment. The drawings still get checked. The details still get detailed. The responsibility still sits exactly where it always did — with the licensed professional whose name is on the firm. What disappeared was the overhead: the production bottleneck, the turnaround measured in weeks, the cost structure that made good design a luxury.

Thirty years ago I told clients the internet would not replace the meeting — it would make the meeting matter more. I told them solar would not replace good design — it would reward it. I told them virtual walkthroughs would not replace buildings — they would help us build better ones.

Same conversation today. Same answer.

The only real question

The question is not whether AI is coming to your industry. It is. The question is whether you will implement it deliberately — with experienced people, clear standards, and accountability — or have it happen to you.

That is the work we do at MRAi Design Studio: helping firms and manufacturers implement AI the way architects implement anything that matters. Carefully. Practically. With the guardrails built in from the start.

Change is inevitable. Doing it well is a choice.