Forum discussion

AI + AEC climate action

Fellow Sages,

I was at an interesting lecture last night. Phil Bernstein (formerly of Autodesk, now at Yale) was speaking (in Lawrence, Kansas, at the U of Kansas) about AI and design/AEC ... you may know that he authored the 2022 book, Machine Learning, about AI and the AEC field. 

Phil has a very interesting posture about AI ... he's both an enthusiast and a skeptic. Regardless of one's views, AI is here and coming, so the AEC / built environment community is dealing with it, working with it, and will be moreso of both in the coming years. 

I was taken with one key point he made, in response to a question posed by BNIM's Laura Lesniewski about using AI for climate action in design (she asked it well, so I'm glad she beat me to the punch): 

Phil suggested that "we" (the AEC / built environment community) need to let tech know what we need and want in the way of AI supported tools. Moreso, he noted that vendors (like Autodesk, where he worked for years) are compelled by coherent input from key sectors. He also noted that the AEC sector is not well known for such coherent input ... but that perhaps a coherent whitepaper request, asking for certain types of tools (informed by understanding of what might be possible, etc.), and signed by, say, hundreds of firms (ideally cross disciplines) could be compelling. 

This is something that we might be able to spark / find funding for / organize / incite ... and it seems to me that this is important. It is not my personal expertise, but I suspect that we could engage with a few groups (such as the Tech and Practice KC at AIA if they are still active ... the subgroups at LFRT ... and some of the tech savvy sustainability leaders at firms who are already on this edge) to unfurl some kind of approach. The whitepaper might have to be funded, in order to be specific enough to be meaningful. 

Thoughts? Perhaps someone is already working on this. To be honest, I was surprised that Phil was not; he would be a good advisor to the project if he were willing. I know him a little, and chatted with him after the talk. 

Kira

0

You rely on LEEDuser. Can we rely on you?

LEEDuser is supported by our premium members, not by advertisers.

Go premium for $15.95  »

Tue, 02/20/2024 - 20:53

OK, the AIA Strategic Council tackled this issue last year (affect of AI on AIA) and they very quickly found Phil Bernstein.  He is amazing, totally full of information on the topic.   His book "Machine Learning" can be found on Amazon here: https://www.amazon.com/Machine-Learning-Architecture-Artificial-Intelligence/dp/1914124014/ref=sr_1_1?crid=3GT34C0535SJ7&keywords=machine+learning+phil+bernstein&qid=1707085632&sprefix=machine+learning+phil%2Caps%2C235&sr=8-1 I have the book but have not read it yet.  Many of the AI study group members did read it and had high praise for the book, so if you are looking for an intro to the topic, I think it is a good place to start.  Phil is a professor at the Yale A-school, was an executive for years at Deltek working on Autocad and Revit. When we had a meeting last year with Phil Bernstein, he had some very clear thoughts on this:
  • Either we work on this problem or we get it worked on for us.
  • The big issue is Data.  We need to collect data, manage it, keep it up to date and Architect's just don't do that.
    • Lawyers collect case history and keep it in managed file systems
    • Doctors gather info.
    • Architects haven't historically managed their data in a similar way.
  • Can the industry declare in a coherent way what we need AI to do, we could help the tech companies get somewhere fast.
  • He really stressed with us that they (the tech companies) are going to tackle this.  if we tell them what we want, they will factor that in.  if we give them zero suggestions, they will just do what they want to do, and they aren't generally informed on what we need to do.
The SC Design Group studied this topic for us and when you  have a group of people that self identify as "designers" they do the prettiest presentations.  What they came up with was amazing - an animated video.  They posted questions to ChatGPT and that program wrote the script.  The then used Mid-Journey to create avatars and environments, then they used Murf.AI, D-ID and Canva-Pro to create a video reading their ChatGPT script and using their Avatars. I soooo love what they came up with, it is super entertaining. (the guy in front of the Cape May-Lewes ferry, his birthday is April 22, as such Earth Day's date). I tried to upload the file, but alas, that didn't work.  You can grab it from my dropbox here:  https://www.dropbox.com/scl/fi/ra7f78iu6zeygzwq9ohad/Q-A-with-ChatGPT-FINAL.mp4?rlkey=rrb89v1k8ncvaicrwdncobm9x&dl=0  Chatham House rules apply here - do NOT share this beyond this network please. They also came up with suggestions about how to handle this. All of this has been given to the AIA Board and we are now in that wait-and-see period. Their recommendations are relatively standard:  Education and Training, Advocate for what we want, and create valuable partnerships with allied organizations along with some sort of steering committee. But the two most interesting ones have to do with creating a data repository that would basically be a data exchange platform.  ChatGPT learned how to communicate using all of your facebook/tik tok/instagram postings.  DrawGPT can use drawings from Dodge reports to learn how to draw.  Once that happens, we will be cooked. And the last thing was to have the AIA consider redefining HSW.  Phil Bernstein covered this well in a talk he gave to the group - i have a recording of that if folks are interested, it is worth a watch.  He specifically questioned whether we shouldn't add: Fire, Climate Change and Embodied Carbon, Modern Slavery, and Pandemics.  if a resolution is brought before the AIA at their annual meeting (kinda like the the 2019 Resolution for Urgent and Sustained Climate Action - attached) and membership votes to pass it, well then the Board has to take action. That is a much better tactic than submitting reports to the board. Justsayin We can also write a letter, I think that is a good tactic.  Just wanted all to know we have a team of folks at AIA who are in year two of working on this. Let me know if I can provide more info or clarity.

Tue, 02/20/2024 - 21:07

Gives Caeser what belongs to Caeser...So this is from our company's GPT version 4 The question:  How can AI assists regarding Climate Action in design?
_______________ Question how can AI assist us for climate action in design Answer AI can assist in climate action in design in several ways: 1. Energy-Efficient Design: AI can analyze various design parameters and suggest the most energy-efficient design options. This can be particularly useful in architecture and urban planning, where AI can help design buildings and cities that minimize energy use. 2. Material Optimization: AI can help in selecting materials that are both sustainable and suitable for the design. It can analyze the lifecycle of different materials and suggest those with the least environmental impact. 3. Predictive Analysis: AI can use predictive analysis to forecast the environmental impact of a design over its entire lifecycle. This can help designers make more informed decisions and create designs that are more sustainable. 4. Simulation and Modeling: AI can simulate different design scenarios and their impact on the environment. This can help in testing and refining designs before they are implemented. 5. Waste Reduction: AI can help in optimizing design processes to minimize waste. For example, in manufacturing, AI can optimize production processes to reduce waste and increase efficiency. 6. Carbon Footprint Tracking: AI can also help in tracking the carbon footprint of a design process, providing valuable data that can be used to make the process more sustainable. 7. Climate Resilient Design: AI can help in designing infrastructure and systems that are resilient to climate change impacts. For example, it can help in designing buildings that can withstand extreme weather events. 8. Education and Awareness: AI can also play a role in educating designers and the public about the importance of sustainable design and the impact of design decisions on the climate.In all these ways, AI can be a powerful tool in promoting climate action in design.  

Tue, 02/20/2024 - 21:47

Thx Luke. Good stuff there, and still meaning to formally introduce myself to this group, but your question prompted another one... I do like that it added the last paragraph. Question how can AI exacerbate climate change? ChatGPT AI has the potential to exacerbate climate change through several mechanisms: Increased Energy Consumption: Training and running AI models often require significant computational resources, which can result in increased energy consumption. This energy consumption, if not sourced from renewable sources, can contribute to greenhouse gas emissions and thus exacerbate climate change. Resource Intensive Hardware: Building and maintaining the hardware infrastructure needed for AI, such as data centers and high-performance computing systems, can have a significant environmental impact. The production and disposal of these hardware components contribute to resource depletion and pollution. Supply Chain Impact: The production of AI technologies involves mining and processing of raw materials, which can lead to habitat destruction, pollution, and other environmental impacts. Additionally, the disposal of electronic waste generated by obsolete AI hardware can contribute to environmental degradation if not managed properly. Misallocation of Resources: While AI has the potential to optimize resource allocation and improve efficiency in various sectors, there is also a risk that it may prioritize short-term economic gains over long-term environmental sustainability. For example, AI-driven optimization algorithms might prioritize cost savings without considering the environmental consequences of resource depletion or pollution. Unintended Consequences: AI systems are complex and can produce unintended consequences. For instance, AI-powered automation may lead to increased consumption and production, offsetting any gains in efficiency and contributing to higher levels of resource consumption and greenhouse gas emissions. Biases and Inequities: AI systems can perpetuate biases and inequities, which can have indirect impacts on climate change mitigation and adaptation efforts. For example, if AI systems are trained on biased data or designed without considering the needs of marginalized communities, they may exacerbate existing disparities in access to resources and opportunities for climate action. To address these risks and harness the potential of AI for climate change mitigation and adaptation, it's essential to implement responsible AI practices, prioritize sustainability in AI development and deployment, and ensure that AI technologies are used in ways that support rather than hinder environmental goals.

Tue, 02/20/2024 - 21:55

I start falling in love with this group....

Tue, 02/20/2024 - 23:44

Luke, I agree - What a great and thought-provoking thread - thank you Kira! Regarding Kira's question about a "coherent white-paper", I'm intrigued by the question of how integrated the working group might want or need to be... Using the Green Commons, it could be fairly stratightforward to get some volunteers that (if properly distributed) could represent an "AEC perspective", which I'm sure Autodesk, and others would be interested in. But is that perspective broad enough to really address the built environment issues that AI can impact, and that lie before us? In addition to AEC, I see potentially huge implications for Owners, Developers, Plans Examiners, Facility Operations, etc. Another way to design the engagement would be to to ask:
"Who are the key stakeholders for the built-environment supply chain, cradle to grave?"
If we did that, we could potentially develop a set of issues/synergies/pitfalls/recommendations based on all of the stages. Such a big "summit" (for lack of a better word) would definitely take some planning, funding, and commitment to meaningful follow up to produce the kind of "coherent" document that would garner lots of signatures and real interest from the Tech/AI sector, but it does seem possible. With this group's connections + some others from the Green Commons, I think we definitely know most of the right folks (including Phil) to try to initiate something like this, but even so, such an undertaking wouldn't be for the faint of heart... Speaking (broadly) for our group, we could selfishly suggest such an event be held adjacent (or in conjuction with) Greenbuild, but a number of the stakeholder groups that I listed above aren't automatic attendees at Greenbuild, so it might need it's own time and place... Thoughts? Clark

Fri, 02/23/2024 - 15:59

Love where this thread has gone ... 

Responding to Anne: I appreciate that AIA SC tackled this with vigor. And that in order to get action, we'd need a Resolution (or just take it out of AIA for work, then back to them when we want some of their muscle). The terrific work done on several climate sprints at SC by Mary Ann Lazarus and others did what those often do ... sent to board and studied or tabled or killed. Specifically, the Carbon Literacy work has been in the board's court for ... multiple years. Maybe best to have a different org handle that, since AIA is single discipline. 

More generally: I'd love to build on this good AI work and start exploring these things, but quickly: The need for data is REAL. The AI developers are not waiting for our industry to think about this for a year or five years ... and things will evolve. It would be idea, I think, to try to organize some defined needs and work on pointing developers to the right types of data for those, and continue the discussion. 

I know this sounds like we should design and build the airplane while in flight, but that may still be better than just being so slow that our role winds up being 100% reactive. I think a both/and could happen here. It is also like codes ... we should get some architects and engineers (and others in our space) to the tables, if we can. 

Sat, 02/24/2024 - 22:25

This is a wonderful thread, and as someone grappling w applying ML/AIA to product data and workflow, I am obsessed - and also can only pop on now for a moment to say that I've come across a number of startups recently thru mentoring and assisting with due diligence for investors - that are using ML/AIA in interesting ways (one is a tool to support and 'mentor' younger designers as they create facade assemblies etc (with a focus on thermal bridging and embodied carbon) - a bunch of others that I can't recall now. For those of you who know Saif from AIA, he is now at Noblis - they do a ton for the Fed Gov including the creation and maintenance of SF Tool for the GSA (side question, do designers ever use that tool?!) - anyway, Noblis has its own ML/AI department (?) and would be a good addition to any white paper or study as they are at the intersection of the design community and our biggest property owner...

Mon, 02/26/2024 - 16:18

On Friday of this week at the AIA Leadership Summit, one of the 1-hr sessions focuses on AI – I won’t be there, but it would be great to hear from anyone who is. Here is the description: AI/AIA: Surviving or Thriving AI is the next great disruptor. Just as CAD revolutionized the craft of hand drafting, AI is likewise a catalyst for dramatic and shifting methods of doing business. As association management professionals and business managers, we are continually challenged to streamline our operations as a means to be better stewards of resources - financial and human. Learn how to begin an iterative process that gradually on-boards various AI tools to help you build websites, automate administrative tasks, develop compelling content, personalize member communications, implement innovative fundraising strategies, optimize volunteer engagement, develop better resource allocation and run data analytics for decision making. Learn through a series of success stories and share a few of your own. As a session finale, participants will imagine the architectural profession in the age of AI as well as techniques to assist our collective through the ushering in of a new design landscape. Learning objectives • Understand the role of AI as a disruptor in various industries, including association management and business operations. • Learn how to identify areas in your organization that can benefit from AI tools and technologies. • Explore the iterative process of integrating AI tools into your operations to streamline tasks, improve efficiency, and optimize resource allocation. • Gain insights from real-life success stories and share experiences to learn practical strategies for leveraging AI. Speaker(s): Amber Lombardo, Executive Director, AIA Memphis, Memphis, TN Jean Carroon FAIA, LEED Fellow Principal 420 Boylston Street, Boston, MA 02116 Direct 617.850.6651 | Main:617.262.2760 Jean.carroon@goodyclancy.com [Goody Clancy logo] From: Barbr

Add new comment

To post a comment, you need to register for a LEEDuser Basic membership (free) or login to your existing profile.