What do you think of using AI models to automate some building design? This is a company that were sent to me and I'm not really sure what to think.
Is this possible? How valuable would it be if so to use? I'm usually skeptical of things like this but the team seems good, they are from the research team at Autodesk.
Some of the features look interesting
Works with Revit - Integrates seamlessly with industry-standard CAD workflows. Use your existing part families, cost and labor data.
Ready to build - Designs are NEC and Customer Specification compliant, use real parts, are fully constructible and optimized for prefabrication.
Everybody is rushing into the AI space right now because it's the "new thing." But I think it's important when exploring or implementing any new technology to keep an eye on what we're actually trying to accomplish with it. These things are just tools. Sophisticated and powerful tools, but still just tools. How they get used, and by whom, determines the quality of what they produce.
Central to figuring that out is understanding the key differences in what humans are good at vs what computers (even elaborate LLM systems) are good at. Humans are very good at goal-direction, value judgments, and making decisions which are dependent on a spectrum of good-bad choices and fit. Computers (even LLMs) are good at defined, linear processes and dealing with right-wrong answers. To be successful using computers as tools, you have to use the right tool for the job.
So, applying that to the example you provided, the expert system can do a lot of heavy lifting automatically generating system options which fit defined right-wrong criteria (e.g. "can this HVAC system cool-heat these spaces at this unit size?" or "will these ducts and pipes all fit through this chase without conflict?").
What these tools can do is replace human judgment about "best choice" or "best fit." Nor can they really do goal-directed creativity in any kind of useful way not requiring a human to curate the results carefully.
csarchitect
Jan 24, 24 2:49 pm
Agreed with the sentiment here! AI overall is very over-hyped. But I do see some value in having junior folks be able to even have the expert system do a lot of heavy lifting. Of course, the value of human ingenuity and creativity cannot be replaced and I hope we all agree with that.
JLC-1
Jan 24, 24 3:15 pm
you know what pisses me off? after working 35 years on autocad, the product is worse than ever, same with all the software introduced in the last 3 decades - too many "visionaries" doing things that don't improve the existing products or the user experience. Stick with what is good and works and improve upon it, don't chase unicorns for the sake of novelty.
I am seeing more AI architecture companies like https://www.augmenta.ai/
What do you think of using AI models to automate some building design? This is a company that were sent to me and I'm not really sure what to think.
Is this possible? How valuable would it be if so to use? I'm usually skeptical of things like this but the team seems good, they are from the research team at Autodesk.
Some of the features look interesting
Links
https://betakit.com/autodesk-a...
https://aecmag.com/bim/towards...
Everybody is rushing into the AI space right now because it's the "new thing." But I think it's important when exploring or implementing any new technology to keep an eye on what we're actually trying to accomplish with it. These things are just tools. Sophisticated and powerful tools, but still just tools. How they get used, and by whom, determines the quality of what they produce.
Central to figuring that out is understanding the key differences in what humans are good at vs what computers (even elaborate LLM systems) are good at. Humans are very good at goal-direction, value judgments, and making decisions which are dependent on a spectrum of good-bad choices and fit. Computers (even LLMs) are good at defined, linear processes and dealing with right-wrong answers. To be successful using computers as tools, you have to use the right tool for the job.
So, applying that to the example you provided, the expert system can do a lot of heavy lifting automatically generating system options which fit defined right-wrong criteria (e.g. "can this HVAC system cool-heat these spaces at this unit size?" or "will these ducts and pipes all fit through this chase without conflict?").
What these tools can do is replace human judgment about "best choice" or "best fit." Nor can they really do goal-directed creativity in any kind of useful way not requiring a human to curate the results carefully.
Agreed with the sentiment here! AI overall is very over-hyped. But I do see some value in having junior folks be able to even have the expert system do a lot of heavy lifting. Of course, the value of human ingenuity and creativity cannot be replaced and I hope we all agree with that.
you know what pisses me off? after working 35 years on autocad, the product is worse than ever, same with all the software introduced in the last 3 decades - too many "visionaries" doing things that don't improve the existing products or the user experience. Stick with what is good and works and improve upon it, don't chase unicorns for the sake of novelty.