{"id":7861,"date":"2025-09-10T18:21:08","date_gmt":"2025-09-10T18:21:08","guid":{"rendered":"https:\/\/www.newsbeep.com\/il\/7861\/"},"modified":"2025-09-10T18:21:08","modified_gmt":"2025-09-10T18:21:08","slug":"ai-cant-draw-a-damn-floor-plan-with-any-degree-of-coherence-common-edge","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/il\/7861\/","title":{"rendered":"AI \u201cCan\u2019t Draw a Damn Floor Plan With Any Degree of Coherence\u201d \u2013 Common Edge"},"content":{"rendered":"

Recently I began interviewing people for a piece I\u2019m writing about \u201cArtificial Intelligence and the Future of Architecture,\u201d a ludicrously broad topic that will at some point require me to home in on a particular aspect of this rapidly changing phenomenon. Before undertaking that process, I spoke with some experts, starting with Phil Bernstein<\/a>, an architect, educator, and longtime technologist. Bernstein is deputy dean and professor at the Yale School of Architecture, where he teaches courses in professional practice, project delivery, and technology. He previously served as a vice president at Autodesk, where he was responsible for setting the company\u2019s AEC vision and strategy for technology. He writes extensively on issues of architectural practice and technology, and his books include Architecture | Design | Data \u2014 Practice Competency in the Era of Computation<\/a> (Birkhauser, 2018) and Machine Learning: Architecture in the Era of Artificial Intelligence<\/a> (2nd ed., RIBA, 2025). Our short talk covered a lot of ground: the integration of AI into schools, its obvious shortcomings, and where AI positions the profession. <\/p>\n

\n\t\t\t\t\t\t\t\t\t\t\tPB: Phil Bernstein
MCP: Martin C. Pedersem\t\t\t\t\t\t\t\t\t\t<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

You\u2019re actively involved in the education of architects, all of them digital natives. How is AI being taught and integrated into the curriculum?<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

I was just watching a video with a chart that showed how long it took different technologies to get to 100 million users: the telephone, Facebook, and DeepSeek. It was 100 years for the phone, four years for Facebook, two months for Deepseek. Things are moving quickly, almost too quickly, which means you don\u2019t have a lot of time to plan and test pedagogy.\u00a0<\/p>\n

We are trying to do three things here. One, make sure that students understand the philosophical, legal, and disciplinary implications of using these kinds of technologies. I\u2019ll be giving a talk to our incoming students as part of their orientation about the relationship between generative technology, architectural intellectual property, precedent, and academic integrity. And why you\u2019re here to learn: not how to teach algorithms to do things, but to do them yourself. That\u2019s one dimension.\u00a0<\/p>\n

The second dimension is, we\u2019re big believers in making as much technology as we can support and afford available to the students. So we\u2019ve been working with the central campus to provide access to larger platforms, and to make things as available and understandable as we possibly can.<\/p>\n

Thirdly, in the classroom, individual studio instructors are taking their own stance on how they want to see the tools used. We taught a studio last year where the students tried to delegate a lot of their design responsibility to algorithms, just to see how it went, right?\u00a0<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

Control. You lose a lot of design autonomy when you delegate to an algorithm. We\u2019ve also been teaching a class called \u201cScales of Intelligence<\/a>,\u201d which tries to look at this problem from a theory, history, and technological evolution perspective, delving into the implications for practice and design. So it\u2019s a mixed bag of stuff, very much a moving target, because the technology evolves literally during the course of a semester.\u00a0<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

I am a luddite, and even I can see it improve in real time.<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

It\u2019s getting more interesting, minute to minute, very shifting ground. I was on the Yale Provost\u2019s AI Task Force<\/a>, which was the faculty working group formed a year ago that tried to figure out what we\u2019re doing as a university. Everybody was in the same boat, it\u2019s just some of the boats were tiny, paper boats floating in the bathtub, and some of them were battleships\u2014like the medical school, with more than 50 AI pilots. We\u2019re trying to keep up with that. I don\u2019t know how good a job we\u2019re doing now.\u00a0<\/p>\n

\u00a0<\/p>\n

\t\t\t\t\t\"machine\t\t\t\t\t<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

What\u2019s your sense in talking to people in the architecture world? How are they incorporating AI into their firms?<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

It\u2019s difficult to generalize, because there are a lot of variables: your willingness to experiment, a firm\u2019s internal capabilities, the availability of data, and degree of sophistication. I\u2019ve been arguing that because this technology is expensive and requires a lot of data and investment to figure it out, the real innovation will happen in the big firms.\u00a0<\/p>\n

Everybody\u2019s creating marketing collateral, generating renderings, all that stuff. The diffusion models and large language models, the two things that are widely available\u2014everybody is screwing around with that. The question is, where\u2019s the innovation? And it\u2019s a little early to tell.<\/p>\n

The other thing you\u2019ve got to remember is the basic principle of technology adoption in the architectural world, which is: When you figure out a technological advantage, you don\u2019t broadcast it; you keep your advantage to yourself for as long as you can, until somebody else catches up. A recent example: It\u2019s not like there were firms out there helping each other adopt building information modeling.<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

I guess it\u2019s impossible to project where all this goes in three or five years?<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

I don\u2019t know. The reigning thesis\u2014I\u2019m simplifying this\u2014is that you can build knowledge from which you can reason inferentially by memorizing all the data in the world and breaking it into a giant probability matrix. I don\u2019t happen to think that thesis is correct. It\u2019s the Connectionists vs. the Symbolic Logic<\/a> people. I believe that you\u2019re going to need both of these things. But all the money right now is down on the Connectionists, the Sam Altman theory of the world. Some of these things are very useful, but they\u2019re not 100% reliable. And in our world, as architects, reliability is kind of important.<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

Again, we can\u2019t predict the pace of this, but it\u2019s going to fundamentally change the role of the architect. How do you see that evolving as these tools get more powerful?<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

Why do you say that? There\u2019s a conclusion in your statement.\u00a0<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

I guess, because I\u2019ve talked to a few people. They seem to be using AI now for everything but design. You can do research much faster using AI.\u00a0<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

That\u2019s true, but you better check it. <\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

I agree, but isn\u2019t there inevitably a point when the tools become sophisticated enough where they can design buildings?<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

So, therefore \u2026 what?\u00a0<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

Where does that leave human architects?<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

I don\u2019t know that it\u2019s inevitable that machines could design entire buildings well \u2026<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

It would seem to me that we would be moving toward that.<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

The essence of my argument is: there are many places where AI is very useful. Where it begins to collapse is when it\u2019s operating in a multivalent environment, trying to integrate multiple streams of both data and logic.<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

Which would be virtually any architecture project.<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

Exactly. Certain streams may become more optimized. For instance: If I were a structural engineer right now, I\u2019d be worried, because structural engineering has very clear, robust means of representation, clear rules of measurement. The bulk of the work can be routinized. So they\u2019re massively exposed. But these diffusion models right now can\u2019t draw a damn floor plan with any degree of coherence. A floor plan is an abstraction of a much more complicated phenomenon. It\u2019s going to be a while before these systems are able to do the most important things that architects do, which is make judgments, exercise experience, make tradeoffs, and take responsibility for what they do.<\/p>\n

\u00a0<\/p>\n

\t\t\t\t\t\"Phil-Bernstein\t\t\t\t\t<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

Where do you fall on the AI-as-job-obliterator, AI-as-job-creator debate?\u00a0<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

For purposes of this discussion, let\u2019s stipulate that artificial general intelligence that can do anything isn\u2019t in the foreseeable future, because once that happens, the whole economic proposition of the world collapses. When that happens, we\u2019re in a completely different world. And that won\u2019t just be a problem for architects. So, if that\u2019s not going to happen any time soon, then you have two sets of questions. Question one: In the near term, does AI provide productivity gains in a way that reduces the need for staff in an architect\u2019s office?<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

That may be the question I\u2019m asking \u2026<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

OK, in the near term, maybe we won\u2019t need as many marketing people. You won\u2019t need any rendering people, although you probably didn\u2019t have those in the first place. But let me give you an example from an adjacent discipline that\u2019s come up recently. It turns out that one thing that these AIs are supposed to be really good at is writing computer code. Because computer code is highly rational. You can test it and see if it works. There\u2019s boatloads of it on the internet as training data in well organized locations, very consistently accessible\u2014which is not true of architectural data, by the way.\u00a0<\/p>\n

It turns out that many software engineering companies who had decided to replace their programmers with AIs are now hiring them back because the code-generating AIs are not reliable enough to write good code. And then you intersect that with the problem that was described in a presentation I saw a couple of months ago by our director of undergraduate studies in computer science, [Theodore Kim,] who said that so many students are using AI to generate code that they don\u2019t understand how to debug the code once it\u2019s written. He got a call from the head of software engineering for EA, who said, \u201cI can\u2019t hire your graduates because they don\u2019t know how to debug.\u201d And if it\u2019s true here, I guarantee you, it\u2019s true everywhere across the country. So you have a skill loss.\u00a0<\/p>\n

Then there\u2019s what I would call the issue of the luddites. The [original] Luddites didn\u2019t object to the weaving machines, per se, but they objected to the fact that while they were waiting for a job in the loom factory, they didn\u2019t have any work. Because there was this gap between when humans get replaced by technology and when there are new jobs for them doing other things: you lost your job plowing that cornfield with a horse because there\u2019s a tractor now, but you didn\u2019t get a job in the tractor factory, somebody else did. These are all issues that have to be thought about. <\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

It seems like a lot of architects are dismissive because of what AI can\u2019t do now, but that seems silly to me, because I\u2019m seeing AI enabling things like transcriptions now.\u00a0<\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

But transcriptions are so easy. I do not disagree that, over time, these algorithms will get more capable doing some of the things that architects do. But if we get to the point where they\u2019re good enough to literally replace architects, we\u2019re going to be facing a much larger social problem.\u00a0<\/p>\n

There\u2019s also a market problem here that you need to be aware of. These things are fantastically expensive to build, and architects are not good technology customers. We\u2019re cheap and steal a lot of software\u2014not good customers for multibillion-dollar investments. Maybe, over time, someone builds something that\u2019s sophisticated enough, multimodal enough, that can operate with language, video, three-dimensional reasoning, analytical models, cost estimates, all those things that architects need. But I\u2019m not concerned that that\u2019s going to happen in the foreseeable future. It\u2019s too hard a problem, unless somebody comes up with a way to train these things on much skinnier data sets.\u00a0<\/p>\n

That\u2019s the other problem: all of our data is disaggregated, spread all over the place. Nobody wants to share it, because it involves risk. When the med school has 33,000 patients enrolled in a trial, they\u2019re getting lots of highly curated, accurate data that they can use to train their AIs. Where\u2019s our accurate data? I can take every Revit model that Skidmore, Owings & Merrill has ever produced in the history of their firm, and it\u2019s not nearly enough data to train an AI. Not nearly enough.<\/p>\n

\t\t\t\t\t\tMCP: <\/p>\n

And what do you think AI does to the traditional business model of architecture, which has been under some pressure even before this? <\/p>\n

\t\t\t\t\t\tPB: <\/p>\n

That\u2019s always been under pressure. It depends on what we as a profession decide. I\u2019ve written extensively about this. We have two options. The first option is a race to the bottom: Who can use AI to cut their fees as much as possible? Option number two, value: How do we use AI to do a better job and charge more money? That\u2019s not a technology question, it\u2019s a business strategy question. So if I\u2019ve built an AI that is so good that I can promise a client that x is going to happen or y is going to happen, I should charge for that: \u201cI\u2019m absolutely positive that this building is going to produce 23% less carbon than it would have had I not designed it. Here\u2019s a third party that can validate this. Write me a check.\u201d\u00a0<\/p>\n

Featured image courtesy of Easy-Peasy.AI.\u00a0<\/p>\n","protected":false},"excerpt":{"rendered":"Recently I began interviewing people for a piece I\u2019m writing about \u201cArtificial Intelligence and the Future of Architecture,\u201d…\n","protected":false},"author":2,"featured_media":7862,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[31],"tags":[437,434,435,436,438,146,85,46],"class_list":["post-7861","post","type-post","status-publish","format-standard","has-post-thumbnail","category-arts-and-design","tag-arts","tag-arts-and-design","tag-artsanddesign","tag-artsdesign","tag-design","tag-entertainment","tag-il","tag-israel"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/7861","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/comments?post=7861"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/7861\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media\/7862"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media?parent=7861"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/categories?post=7861"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/tags?post=7861"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}