AI BIM. Can ArchiCAD be ready for that?

[Podolsky]

Hi there!

I have decided to start a separate thread about Artificial Intelligence in AEC. I hope here we will able to collect as much information as what happens today in the AI world, show examples of using AI in BIM programs and programs driven by AI, and discuss what architects and other AEC professionals would like to expect from it.

I have a very strong opinion, that these technologies will come to us very, very soon (because AI already exists in many things we are using every day - like search engines or digital photo applications) and this might be a very interesting subject to review.

Please find below a scheme I have prepared, showing how AI for BIM might look like.

Building elements (as we know tools in ArchiCAD) are controlled by placement algorithms, that coming from building classification databases). For example - placing partition walls in the office with the right chosen sound insulation, fire ratings, correct corridors lengths, fire escapes, etc.

I/O engine responsible for Input / Output - but in architectural terms - automatic drawings generation and publishing, remote communication, including communication via e-mails, teamwork, IFC exchange. It is something like a secretary-robot, that supervising the BIM project.

Language analyzer transforming languages into commands, understandable by the system (software). It can understand human language and communicate with him as a chatbot, it’s also can understand different languages - like Python, JavaScript, AutoLisp, etc. Even read IFC (because IFC is a script).

Physical simulations help to improve correct element placement. Simulations shall be 100% on physics (more like physical engines in 3D animation software). Includes loads, earthquakes, heat distribution and loss, fire spread, wind load, radiosity, and photon tracing, similar to Monte Carlo.

Additional block called construction simulations helps to represent the construction process, including delivery, animation of cranes and installation process, construction timeline, and similar.

Each building element has two additional layers - assembly (if it’s a wall, then it might be studs, cover, insulation, and brackets) and behavior. Behaviour is connected to both physical and construction simulations.

[Podolsky]

I think it can.
Currently when we do projects in BIM - we are placing each element. Wall, another wall, pipe, bend, socket, slab, lintel, steel beam, bolt... We are building digital twins bit by bit how it would happen in real construction. But in real building there are thousands of elements, that needs to be put together. Plus regulations, standards etc. To do such a project on high quality - takes very high professionalism of designers team (that is not always possible to achieve) and very long period of time.
Usually there is no time for that and there are just few talented technicians for whole big office. By this reason companies might do one or two good looking BIM projects - to show it as great example, but usual routine projects are delivered somehow - just because there is not enough resources (not enough money to be paid, not enough time, not enough professional BIM modellers).

I know great example of such a "success": Leadenhall Building - project by Rogers in London City. It was built by construction company Lang O'Rourke. They have their own BIM team and using everything: Revit, ArchiCAD, NavisWorks etc. They "built" the building in computer 6 times, finding the best way to construct it. All sounds great, but ALL sub-contractors, who built the building, lost money. This is complete failure of BIM technology.

But various AI algorithms can handle it. Instead of "deconstruct" building into parts and collect it again and again, search for errors and clashes manually, marking up, review errors again and produce new errors - computer algorithms can operate by whole systems. For example in reality concrete slab and columns are one solid object (with rebars inside). Amount and density of rebars are depend on the overall structure. So why place then each column and add there rebars, when computer can handle the whole structure at once? Or services: why place each pipe and bend, when it's actually line from point A to point B. Computer can find the best route for services - similar way how today Google map is showing the best route for the car.

So, if start from right building classification - many things can be produced completely automatically. Let say we have high rise building. Structure - concrete slabs and columns. Size of underground parking lots, services voids, lifts, stairs, structural cores and corridors - all these controlled by building regulations. There might be several options for lighting and heating and cooling systems - but still all is more less standard.
Separation of floors to apartments of offices are also dependent on regulations and fire safety. The only freedom architects have - to choose right shape and facade system. Add some features - like specific entrance hall, view platform...
Or if it is traditional building - here everything even more standard, because traditional architecture principles have been developed for several hundreds years: standard walls, roof, slabs with some modern elements like steel lintels and concrete foundation.

Actually, before computer era architecture and construction engineering was more less "global". With manual BIM, I think, we went too deep into details and started loosing the ground and real meaning of architecture.

I might say even more - there is no another way as start to implement AI into BIM systems. Amount of data in BIM is increasing and people are not able anymore manage everything manually - it's turning into nightmare of IFC properties, model reviewing meetings and BIM related bureaucracy. CoBie in UK - great example. CoBie increases ArchiCAD time enormously - several megabyte IFC file turning into gigabyte, few gigabyte into ten of gigabyte, and nobody really understand why it's necessary to use CoBie (because it's new rules).

[matjashka]

What you're saying makes sense, for architects who spend their careers designing high-rise where - like you said - your freedom is to do massing and pick a facade system with some accents to break it up.

Your remarks about merging columns, beams and slabs into one intelligent system also make total sense in that environment. I spend good amount of my time on renovation work, often in historic buildings, and in some cases, you only know the existing condition after demolition subcontractor has been awarded the contract. It's a different approach than trying to BIM every penny spent on a 100 Million development (pick your favorite currency), hand the contractor a huge data set and proclaim another technological victory.

There are two major side effects of rapid computerization: first, it lulls us into thinking that computers make our lives easier, and second, computers do things that nobody really understands any more (the phone in your pocket). Ultimately, this will NOT be beneficial to Homo Sapiens. I'm against AI. Automaton is good, AI is very, very bad. As long as people write code, people can read and re-write code. The answer to humans doing a better job is not AI, it's better education and gaining more experience ("wisdom", as Solomon put it).

A story about design automation: over 15 years ago I became proficient with a program called Chief Architect. It could (roughly) auto-generate an entire stick-built roof and walls in less than a minute, together with a lumberyard shopping list, but my favorite function was that when I drew 4 walls, it would automatically generate a room with its own name, flooring, baseboard and crown moulding, even unique paint color inside that room. 5 clicks worth of work, if I remember correctly. I could still use that today...

[Podolsky]

I don't think we can really separate AI from automation. AI is automation. Just very advanced automation. In time of ArchiCAD 6.5 Graphisoft advertised GDL as "smart" object - this is also some sort of "artificial intelligence". Later they said that new stair generation is driven by AI.
What I do see in modern architectural and construction world - with all computerisation and new technologies there is still exist very big element of, I might call it, gambling. Financial speculations, that drives everyone mad. Developers, engineers, contractors, sub-contractors - all quite often are involved into big game with big financial risks, involving and avoiding the same time into contracts that can bankrupt them. And on top all this financial pyramid are lawyers, banks and insurance companies. And this is REALLY bad. So bad, that none of computers can replace them.
And I think this is real danger - corruption of HOMO SAPIENCE. From that all problems begins - 40% waste on construction sites, 10 time overpriced properties, generation of millennium that left without their own houses, finally obvious fact, that technologies in AEC are very outdated to compare with military, aerospace and cars industry.
I'm quite sure, that if it would be investments (as part of fight with CO emission for example) on government level into AEC software development - we would not have anymore all these nonsenses with ArchiCAD-Revit-IFC etc.

As I understand, idea of BIM was to make life of architects and constructors easier. But currently it is too heavy.

[jl_lt]

Hi Mr Podolsky, you state that you want AI to solve our problems which occur, acording to you, because "nobody checks anything". But then, in another post you say that once AI begins to solve our problems it needs to be checked by humans so what do you want? either we get out of the way and let this theorical AI you describe do its job or just let the computer be an assistant as it currently is.

As Mr matjashka said, I think most if not all of the problems you describe dont have anything to do with software like Archicad or Revit, and the AI that would be required to solve them (an AI capable of taking desicions and setting goals independent of human input) is currently non existant nor anywhere close to exist (yes, i saw your examples). And i dont know if i want anything like it to exist.

Some statements:

-As far as i know, current comercial software is as good as the input it receives from the user. correct me if im wrong, but i doubt it.
-Construction is an inherently human affair. It will exist in one way or another as long as humans exist. Thus, as it is an human affair also restricted by time and budgets, it is also prone to errors. Deal with it. The goal is to minimize them as much as possible.
-Incompentence cannot always be the cause of errors. A lot of the time it has to do with people wanting to chew on more that they can take. Architects trying to design beyond their technical capabilities, Contractors trying to give the best price just to find themselves way overhead, clients trying to develope more than they can pay for, and so on.. So, unless an AI can tell you "hey humans, the building you are trying to built is not the optimal choice for all parties involved" and we actually LISTEN to it, no ammount of computing power will ever prevent us from attempting stupid things. And that is the beauty of construction and architecture.
-If given sufficient time and money, most if not all errors could theorically be corrected before construction begins, but it is a situation that in most cases is not possible even in super high budget projects. The AI you mention could help with this, but...(see next point)
-As of today, Archicad cannot handle in-place external reference editing nor it has an automatic dimension tool that is truly automatic, and, oh yeah, it sometimes moves things out of position ; do you really think this or any other software are near solving the problems you describe through AI?
-Too many parties involved with their own agenda, most of which dont give a d-mn about architecture. Recently we got invited to a competition for some cheap apartment building. Just reading the BEP dizzied me. Do we really need 3 meetings a week with God knows how many people, 5 client representatives, BIM specialist, consultants and the architects, for a building that i actually can draw in 2d cad by myself in less than a month if left unbothered? Again, software nor AI can solve any of this, unless i can substitute all the parties involved (except the architects of course) with intelligent AI.
-Also, how about integrating the contractors DURING the design phase instead of AFTER, so they can give their input during the design and BIM process?
-Most of us dont have the minerals to stay out of situations that we dont like. So, architects agree to laughable fees, Contractors agree to even more laughable contracts and so forth, "to keep the client happy", which almost always end in the situations you describe.
-Instead of AI, sometimes the "client" just needs a kick in the rear.

[Bricklyne Clarence]

jl_lt wrote:
Hi Mr Podolsky, you state that you want AI to solve our problems which occur, acording to you, because "nobody checks anything". But then, in another post you say that once AI begins to solve our problems,
you say it needs to be checked by humans so what do you want? either we get out of the way and let this theorical AI you describe do its job or just let the computer be an assistant as it currently is.

As Mr matjashka said, I think most if not all of the problems you describe dont have anything to do with software like Archicad or Revit, and the AI that would be required to solve them (an AI capable of taking desicions and setting goals independent of human input) is currently non existant nor anywhere close to exist (yes, i saw your examples). And i dont know if i want anything like it to exist.

Some statements:

-As far as i know, current comercial software is as good as the input it receives from the user. correct me if im wrong, but i doubt it.
-Construction is an inherently human affair. It will exist in one way or another as long as humans exist. Thus, as it is an human affair also restricted by time and budgets, it is also prone to errors. Deal with it. The goal is to minimize them as much as possible.
-Incompentence cannot always be the cause of errors. A lot of the time it has to do with people wanting to chew on more that they can take. Architects trying to design beyond their technical capabilities, Contractors trying to give the best price, clients trying to develope more than they can pay for, and so on.. So, unless an AI can tell you "hey humans, the building you are trying to built is not the optimal choice for all parties involved" and we actually LISTEN to it, no ammount of computing power will ever prevent us from attempting stupid things. And that is the beauty of construction and architecture.
-If given sufficient time and money, most errors can theorically be corrected before construction begins, but it is a situation that in most cases is not possible even in super high budget projects. The AI you mention could help with this, but...(see next point)
-As of today, Archicad cannot handle in-place external reference editing nor it has an automatic dimension tool that is truly automatic, oh yeah and sometimes things move out of position ; do you really think they or any other software are near solving the problems you describe through AI?
-Too many parties involved with their own agenda, most of which dont give a d-mn about architecture. Recently we got invited to a competition for some cheap apartment building. Just reading the BEP dizzied me. Do we really need 3 meetings a week with God knows how many people, 5 client representatives, BIM specialist, consultants and the architects, for a building that i actually can draw in 2d cad, alone, in less than a month? Again, software nor AI can solve any of this, unless i can substitute all the parties involved (except the architects of course) with intelligent AI.
-Also, how about integrating the contractor DURING the design phase instead of AFTER, so they can give their input during the design and BIM process?
-Most of us dont have the minerals to stay out of situations that we dont like. So, architects agree to laughable fees, Contractors agree to even more laughable contracts and so forth, "to keep the client happy", which almost always end in the situations you describe.
-Instead of AI, sometimes the "client" just needs a kick in the rear.

This is an excellent post and response.
I couldn't have said it better myself.
Any of it.
You've boiled down architecture and construction down to a "t".

Thank you for that.

[Podolsky]

I'm quite clear in what I want from AI. First at all AI driven projects will not produce such ridiculous amount of errors as human beings do. So, final result will be checked by humans - but the process will be much faster. Very often people who involved into project just waste a lot of time on the meetings - trying to pretend they are working hard, but in reality with their empty talks for hours solving problems, that possible to solve in 10 minutes. In this situation AI can help too. Because if in the project driven by AI no problems occur - nothing to solve then on the meetings.
The goal of AI in architecture and construction is to achieve zero waste on construction site, increase speed and safety, and, finally, reduce the price of real estate. Nobody is dreaming to pay millions for apartments today - when real price is 10 times less. Population is growing on the earth - people need where to live. There was government programs after Second World War to give affordable houses to people in various countries - thanks to that we were born and got education. But today humanity again is falling into age of feudalism - when human society is divided into landlords and peasants (with smartphones this time). Young people cannot live in Facebook and Instagram only. They also need physical space where to be.
We architects, are trying to be on the side of landlords - because they give job to us. And actually we do not care about global problems in our societies - like homelessness. We want to be rich and work for rich, isn't it?
But, if all AEC industry is involved today to build one huge Babylon tower, that will fall one day? I think it's obvious, that real estate market is bubble, that can burst one day. I even did hear an opinion, that war in Syria is started because of separation of humans by city areas - rich and poor.

Okey, back to AI. Big and famous companies as Zaha Hadid Architects and Foster + Partners are using some sort of AI in their projects already - for example for automatic drawings generation of complicated facade elements, made from glass. There is a guy in Internet, who did all that for them - programming on C in Rhino, when program was able then to produce hundreds of drawings (with dimensions and annotation - completely automatically) for manufacturing. He is talking about different things in his posts.



When I asked: Can ArchiCAD be ready for AI, I mean - will ArchiCAD have possibility to be connected to AI to work completely automatically without human input. Let say it's running on server, interaction with human is happening via chat and email. Currently I'm not sure technically it is possible - let say to run ArchiCAD without UI and communicate with the program using terminal (as many modern software do).

[Podolsky]

I also would like to say, that humanity must expect digital social revolution in the future. When computer technologies will completely change our lives. And, actually, it already changing, as industrial revolution, electricity, medicine, phone and cars changes our lives. We even understand time now different than 300 years ago.
But there is always stubbornness and resistance of people, who want to change nothing. Like people, who was against of trains. Or who preferred riding horse instead of using cars. When it happens - we facing sometimes ridiculously stupid conversations: for example when I was student, teacher of furniture design told me, that computers are not precise and because of that they cannot be used in furniture design and manufacturing. Nothing more idiotic I did not hear all my entire life.

I've got a feeling, that today BIM technology is not really changing AEC industry (how it was expected), but AEC industry - because they are not willing to adopt new horizons of technology - changing BIM. And very often to the level when it becoming useless and simply time consuming.

[Bricklyne Clarence]

Podolsky wrote:
......

When I asked: Can ArchiCAD be ready for AI, I mean - will ArchiCAD have possibility to be connected to AI to work completely automatically without human input. Let say it's running on server, interaction with human is happening via chat and email. Currently I'm not sure technically it is possible - let say to run ArchiCAD without UI and communicate with the program using terminal (as many modern software do).

Why would anyone want this?

Design without human input. At that point is it even "design"?

Architecture isn't like other fields you seem to be comparing it to that can be automated for the most part to remove the human aspect of it.
Nor should we want it to be.

You can't program or write into an algorithm, the capacity for creativity, sensitivity for aesthetics, emotional intelligence and cultural and socio-logical nuances of design.
These are things that are uniquely.....human.
And which you can quantify, digitize or break down into "ones" and "zeroes".
An innate part of how we as a species perceive and interact with our world.

We've been trying as a species for centuries now to somehow quantize beauty and aesthetics (everyone remembers the "Vitruvian man"?) - , possibly with the aim of having rules that can allow it to be mass-produced, but the parameters for creativity and aesthetics change with every passing generation and society, to fit them in a particular time.
Each generation has their own parameters for what qualifies as beautiful and aesthetically pleasing - either based on the past or based on what they see of themselves as a people.

Again, these are things you can't break down into an algorithm and then use to replace that aspect of design in the process of design - especially in Architecture.

Why would I want the program telling me (or anyone else) how my building should look a certain way or fit into its context or otherwise conform to requirements it needs to?

Just to save time? Money? Really?

ArchiCAD is a tool, at the end of the day. And I hope it always remains so.
A very good tool that HELPS me do what I need to do, but still a tool nonetheless.
Not a design "partner"....
Not even a "staff member"....
Certainly not a replacement.
But a tool.

[Podolsky]

Disagree. To design a building - not necessary to open ArchiCAD (AutoCAD, Revit, Vectorworks) and start to draw. You can take a paper and start draw with a pen. And many architects still do that.
After you open web-site - some sort of chat interface with a bot and typing:
I need a building, located there and there, this amount of floors, that kind of structure, modern (or traditional). This way - communicating with robot and answering to its questions - you designing a building. Results are showing on the screen as previews.
Absolutely the same way, how some architects designing without touching computers - just instead of robot they have drafters.
So, robot is creating BIM file and at any moment person can connect to the file using BIM program and continue (if he wants) to work in BIM environment.

So, why not? Today the similar way we are using a lot of things. For example - when we take a photo with iPhone. We actually instead could say: "No, no! We don't need AI in photo processing! We need RAW file and will process everything manually in PhotoShop."

So why it's bad to have automatic function, that can convert PointCloud scan into BIM model completely automatically? Or help to make proposals for planning application and build later complete model with all necessary construction documentation?

I just want to show one small example: this is Jerusalem robot. Written on AutoLISP and running in AutoCAD small program, that calculates areas in planning applications in municipality of Jerusalem. They have there quite strict rules about building areas - and robot is making these schemes and calculations. You're uploading a file on the web-site and getting respond via email. Very simple tool - but at least all planning applications have same graphical style of colour-coded area schemes.
To build the same robot on ArchiCAD impossible - because it does not have terminal commands and cannot process the files without manual input.

[Anonymous]

@Podolsky
Thanks for your thoughts and links on AI. Very interesting indeed. I see you are really enthusiastic about it. But I think your expectations may be too high. I 100% agree that AI is definitely a solution for those repetitive tasks on the design process. i.e.: Pen thicknesses for 2d representation. Pen thickness is a technik that came from the old drawing boards to give the sense of depth in a 2d drawing. Foreground elements with thick lines and background ones with thin lines. It would really help in the documentation process. But in the other hand, I think you are not correct on the assumption that Architecture is "Just" a matter of solving problems. It is way more complex than that. It involves human interaction and experiences that only humans can have. For example: A machine will never know what is the loss of a Family Member. Or what is to be discriminated by the color of its skin... gender... economic or political status... etc... But in the other hand there is a current lack of affordable housing in the world that Architects will not be able to solve in the short or medium time. And AI could give some help on this. Perhaps Architecture will become much more like what Fashion/Cloth industry. Sometimes one just need regular pair of jeans and not a tailored suit. But on someother wedding he will sure want the finest tailored suit. Anyway... The answer is always in the middle path.