Will Architecture/Engineering have its Kodak awakening?

I just saw this post by Daniel Schwarz , 15 minutes and the LLM Astra generate a building model in 15 Minutes based on a floorplan. So is this the first steps towards a Kodak moment in our industry?

In case you forgot, Kodak's business model/domination in film got killed with digital cameras. The irony was that their engineers invented the digital camera. Before I take my verdict, I want to share a list of things that I believe will get disrupted and which not:

• AI will replace/disrupt Normal Housing projects.
• Junior positions / conventional modeling and drafting
• Manual clash detection.
• AI controlled Robots will at some point replace the need for construction workers (will take some time. First the military robots are getting better, when the technology will get down in the civil sector).
• Code/compliance checking.
• Quantity takeoff, partly cost estimation.

AI will not replace:

• Luxury market
• Architects signing a building permit.
• Engineers signing off on a calculation.
• Requirements Engineering - translating wishes into needs and requirements, especially the human parts.
• For some time: construction work especially in renovation projects and in smaller scale.
• Coordination and orchestration of different trades, that's about relationships and negotiations so it will stay.

This is actually a very important aspect. AI can only know a transactional relationship, if you live like this as well, you don't lose a lot. If you see the merit of a relationship without a transactional benefit, well we will lose a lot. So my verdict: It's a compression issue. Fewer juniors, a problem for the industry and society when they can't learn on the job anymore. More seniors who will have to orchestrate more. Less value captured when you sell hours. So at worst it's a downward spiral, at best a self-correcting loop. Downward spiral: Fewer juniors train on the job, fewer seniors in the future to do the sign-off and the more tempting it will look to put AI in charge of decision taking. We end up handing over control. Self-correcting loop: the scarcer the reviewers become, the higher they price their service, the more interesting it becomes and there is money again for training juniors. The caveat is speed and time. Will we have time for it. As William Blake puts it in Auguries of Innocence:

The Babe is more than swadling Bands; Throughout all these Human Lands Tools were made, & born were hands, Every Farmer Understands.

Probably not by accident I had to think about him, he was born in a similar time of massive changes with massive social impact.

With the move from hand drawing, CAD, and now BIM, tools catch more value. But they need to change BIM modeling/authoring just for documentation, to produce plans, loses value. Solving construction problems, the detailing, the product development will increase. Again, a move towards the physical, taking into account constructability...

So the value capturing moved from the people doing the planning work, toward Software companies providing the tools, and now to AI providers. Here the big question will be liability again. Google was recently fined for wrong LLM summaries with the argument, that the summary, unlike a search result is a product by itself and a own claim. We will see how this progresses, sofar it was a Munich court and google goes into the next instance.

It's a mass market issue. In Switzerland many architects earn their money with housing projects. And this market can be easily disrupted already with today's automation technology, planning could look different. AI can only be an accelerator. What about me? I earn my money in creating systems where humans and machines can work together. I started out with purely automation driven initiatives like abstractBIM, continued to focus more and more on the underlying data layer that makes this collaboration possible and started to focus now more on defining and verifying the output that is created by humans or machines to take decisions. So till we hit the wall of not enough seniors to take meaningful decisions, I feel safe.

Rereading this, there is a linear extrapolation fallacy in this prediction. History was seldom linear, there was always a counter movement. So the stronger AI gets, the more people will start valuing the craft, the time spent to learn something. So not the quick fix, but the struggle.

Where does this leave us: In an interesting time full of chances and opportunities. At least that's my take.

This article was written by me. I used AI to discuss ideas, but all the formulations and the chain of thought are raw, pure Simon. So sorry for the roughness, I hope you enjoy.