And while the amount of automation is forever increasing, there’s still certain tasks that are extremely difficult to automate. The problem is, those difficult to automate processes are often hilariously easy for humans to do. Robots are great at doing the exact same thing over and over, but throw any variability into the mix, and it’s very difficult to make them able to adapt to it.
Yes, some processes can be fully automated. No, we can not fully automate the assembly of a whole car. It’s still too complex.
One important caveat to this is that this isn’t really a technical limitation, but more of a financial one. We absolutely could build a ridiculous Rube Goldberg machine of a factory that took raw materials in on end and spat out fully assembled cars on the other. And have it so automated that the place operates with the lights off.
However, it’s far cheaper to build a mostly automated factory, rather than a completely automated one. It’s a financial limitation, not a technical one.
Keep this in mind. A building is a memory of the tools used to create it. And as those tools change, the buildings change.
Yes, the way we currently build site-built wood-framed construction would be a poor fit for automation. But other building forms may be more amenable to automation. The form of the building itself may change to fit automated construction techniques. The most obvious example of this is 3D printing. And yes, those buildings have many issues. But we not only consider 3D printing.
For example, imagine a building that isn’t so much made from materials cut to length as is made from small blocks or modules assembled together. It’s easier for me to imagine automating ICF construction than stick-built construction. Or maybe in the future, every home will be built with an at-grade or basement slab or basement suitably flat to serve as a working platform for construction robots. Maybe homes are built around a large central vertical shaft, the shaft that once held a large articulated construction robot. (I’m imagining a large robot riding on vertical rails in a vertical shaft. The robot has a long articulated arm that can reach into individual rooms and perform work functions.) Once automation is good enough, you start designing the building for the automation, rather than the automation the building.
Imagine if I could build homes for half the cost of conventional homes. However, it comes with the caveat that these homes look very different from conventional ones. They look different because their design was optimized for automated construction. Would they sell? If they still functioned as homes. If they still kept you warm, dry, and comfortable. Unless the homes were just completely crazy, I think those homes would sell just fine. People have learned to live in all sorts of things. Pioneers built houses on the prairie out of sod. If automated construction requires substantial modification to building design, but those buildings are substantially cheaper or faster to construct? The architecture will bend to the automation.
Tell me you’ve never designed new cancer treatments, programmed a video game, or decoded a species genome w/o telling me.
If you think construction is too complicated for AI and robotics to ever be able to handle all that shows is how very little you know about the industries AI has already had a large impact on and how little you think of the people working in those fields.
Don’t mistake my criticism as exclusion of other industries. At no point did I bring up any other industry and its relation to AI so let’s leave the straw man out of the argument. Furthermore you conflate sci-fi and hype with reality.
Specs exist, drawings exist, but work always deviates for one reason or another. It’s one thing to look at lines on a page and another to do it and be there. Little decisions to make changes to suit a situation or field conditions have to be made every day.
I’m not arguing a straw man. AI has already made strong headway in the areas I listed. Things that two decades ago people would not have expected to happen.
I’m not conflating sci-fi and hype. You’re conflating what you see around you right now with what will be around you decades from now. You understand the difficulties of your own line of work but don’t understand the similar difficulties in many other professions and fields, and so even though AI has already begun to overcome those very similar issues in actual practice in the real world right now you still believe it’s a thing that can never happen.
You describing the detailed procedures necessary to construct a house doesn’t make your point. It does the opposite. It’s showing you know your own little world and not much outside of it. There are already buildings being 3D printed using AI to adjust for real world discrepancies not shown on the schematics. It’s a thing already happening. Yes, those aren’t easily movable devices that can build an entire row of homes. That’s not a thing yet. But it will be, and I’d be very surprised if it took more than 20 years to happen.
I have no interest in arguing with you about it, because you’re wrong. Categorically and completely. Feel free to set a calendar reminder from two decades from now so you can rub in my face how it turned out that I was so smug and wrong and you were right all along. It won’t happen.
No shit dude. This is a meme about AI and construction. The entire argument this meme presents is the ability of AI (or rather lack of) to replace construction.
So coming in with a “AI has made in-roads with x other field” is a complete unrelated bullshit you’ve brought in to strengthen your defense of AI. That’s like if I said Invermectin doesn’t do shit for COVID, and you come in with “well it works great on horses”. Regardless of the legitimacy you feel your statement has, it’s off topic. And it still doesn’t change the fact that AI isn’t replacing shit when it comes to actual trades.
People said the same thing about bowling alleys, because setting up pins was a job and now a machine does it. Have you seen that machine? It’s a huge contraption that needs a guy who works on it full time. Imagining robots on a construction site in the forseeing future is insanity. Even a robot carrying around concrete bags is nuts.
You need to use your imagination more. A robotic construction site will not look anything like a human one. Bags of readymix concrete have the form they do because they are intended to be carried by a human being. But the goal isn’t to get a bag from point A to point B. The goal is to get the concrete from point A to point B. Instead of a humanoid robot carrying a bag of readymix, imagine an automated hopper on the end of a crane, with a computer vision and reasoning system that allows it to precisely place it where it needs to go. And instead of that concrete being formed by a humanoid robot holding a trowel in its hands, it will be mounted on a robotic arm mounted on a glorified roomba.
When we automated car factories, we didn’t do so by making millions of mechanical men. Construction will be no different. Individual humans are not masters of every construction trade. Why should we demand the robots be masters of all trades? We can simply purpose-build robots for particular construction tasks. Builders wealthy enough may own a fleet of them. Smaller operations will rent them as needed. Just replace the cost of hiring subcontractors with the cost of renting a particular trade robot.
This is true in 2026. I bet at one point, auto workers had the same kind of thoughts about how robots could never build a whole car.
The robots still can’t.
And while the amount of automation is forever increasing, there’s still certain tasks that are extremely difficult to automate. The problem is, those difficult to automate processes are often hilariously easy for humans to do. Robots are great at doing the exact same thing over and over, but throw any variability into the mix, and it’s very difficult to make them able to adapt to it.
They fired everyone at Xinyi auto glass, automated the whole factory. See the Netflix movie American Factory for more info.
Glass is not a car. That’s a few components on a car.
Yes, some processes can be fully automated. No, we can not fully automate the assembly of a whole car. It’s still too complex.
One important caveat to this is that this isn’t really a technical limitation, but more of a financial one. We absolutely could build a ridiculous Rube Goldberg machine of a factory that took raw materials in on end and spat out fully assembled cars on the other. And have it so automated that the place operates with the lights off.
However, it’s far cheaper to build a mostly automated factory, rather than a completely automated one. It’s a financial limitation, not a technical one.
Cool story, but this is my day job. I very aware of what the industry trends are.
Tell me you’ve never worked construction w/o telling me.
Keep this in mind. A building is a memory of the tools used to create it. And as those tools change, the buildings change.
Yes, the way we currently build site-built wood-framed construction would be a poor fit for automation. But other building forms may be more amenable to automation. The form of the building itself may change to fit automated construction techniques. The most obvious example of this is 3D printing. And yes, those buildings have many issues. But we not only consider 3D printing.
For example, imagine a building that isn’t so much made from materials cut to length as is made from small blocks or modules assembled together. It’s easier for me to imagine automating ICF construction than stick-built construction. Or maybe in the future, every home will be built with an at-grade or basement slab or basement suitably flat to serve as a working platform for construction robots. Maybe homes are built around a large central vertical shaft, the shaft that once held a large articulated construction robot. (I’m imagining a large robot riding on vertical rails in a vertical shaft. The robot has a long articulated arm that can reach into individual rooms and perform work functions.) Once automation is good enough, you start designing the building for the automation, rather than the automation the building.
Imagine if I could build homes for half the cost of conventional homes. However, it comes with the caveat that these homes look very different from conventional ones. They look different because their design was optimized for automated construction. Would they sell? If they still functioned as homes. If they still kept you warm, dry, and comfortable. Unless the homes were just completely crazy, I think those homes would sell just fine. People have learned to live in all sorts of things. Pioneers built houses on the prairie out of sod. If automated construction requires substantial modification to building design, but those buildings are substantially cheaper or faster to construct? The architecture will bend to the automation.
Tell me you’ve never designed new cancer treatments, programmed a video game, or decoded a species genome w/o telling me.
If you think construction is too complicated for AI and robotics to ever be able to handle all that shows is how very little you know about the industries AI has already had a large impact on and how little you think of the people working in those fields.
Don’t mistake my criticism as exclusion of other industries. At no point did I bring up any other industry and its relation to AI so let’s leave the straw man out of the argument. Furthermore you conflate sci-fi and hype with reality.
Specs exist, drawings exist, but work always deviates for one reason or another. It’s one thing to look at lines on a page and another to do it and be there. Little decisions to make changes to suit a situation or field conditions have to be made every day.
I’m not arguing a straw man. AI has already made strong headway in the areas I listed. Things that two decades ago people would not have expected to happen.
I’m not conflating sci-fi and hype. You’re conflating what you see around you right now with what will be around you decades from now. You understand the difficulties of your own line of work but don’t understand the similar difficulties in many other professions and fields, and so even though AI has already begun to overcome those very similar issues in actual practice in the real world right now you still believe it’s a thing that can never happen.
You describing the detailed procedures necessary to construct a house doesn’t make your point. It does the opposite. It’s showing you know your own little world and not much outside of it. There are already buildings being 3D printed using AI to adjust for real world discrepancies not shown on the schematics. It’s a thing already happening. Yes, those aren’t easily movable devices that can build an entire row of homes. That’s not a thing yet. But it will be, and I’d be very surprised if it took more than 20 years to happen.
I have no interest in arguing with you about it, because you’re wrong. Categorically and completely. Feel free to set a calendar reminder from two decades from now so you can rub in my face how it turned out that I was so smug and wrong and you were right all along. It won’t happen.
No shit dude. This is a meme about AI and construction. The entire argument this meme presents is the ability of AI (or rather lack of) to replace construction.
So coming in with a “AI has made in-roads with x other field” is a complete unrelated bullshit you’ve brought in to strengthen your defense of AI. That’s like if I said Invermectin doesn’t do shit for COVID, and you come in with “well it works great on horses”. Regardless of the legitimacy you
feelyour statement has, it’s off topic. And it still doesn’t change the fact that AI isn’t replacing shit when it comes to actual trades.People said the same thing about bowling alleys, because setting up pins was a job and now a machine does it. Have you seen that machine? It’s a huge contraption that needs a guy who works on it full time. Imagining robots on a construction site in the forseeing future is insanity. Even a robot carrying around concrete bags is nuts.
You need to use your imagination more. A robotic construction site will not look anything like a human one. Bags of readymix concrete have the form they do because they are intended to be carried by a human being. But the goal isn’t to get a bag from point A to point B. The goal is to get the concrete from point A to point B. Instead of a humanoid robot carrying a bag of readymix, imagine an automated hopper on the end of a crane, with a computer vision and reasoning system that allows it to precisely place it where it needs to go. And instead of that concrete being formed by a humanoid robot holding a trowel in its hands, it will be mounted on a robotic arm mounted on a glorified roomba.
When we automated car factories, we didn’t do so by making millions of mechanical men. Construction will be no different. Individual humans are not masters of every construction trade. Why should we demand the robots be masters of all trades? We can simply purpose-build robots for particular construction tasks. Builders wealthy enough may own a fleet of them. Smaller operations will rent them as needed. Just replace the cost of hiring subcontractors with the cost of renting a particular trade robot.