Small-Firm Architects and AI in 2025
AI Gets My Attention
AI became a ‘thing’ in 2023 when ChatGPT became accessible to everyone. Around that time, I became curious and started paying attention to what was happening. Over the past 18 months, I have learned some things that I thought would be worth sharing.
It is kind of amazing
No one truly knows why it works (also amazing)
The worry about AI taking over the world looks more and more foolish
It really will be as big a deal as the Industrial Revolution
How would you characterize your interest in AI?
Take a second to let me know your level of interest in AI
What To Do About AI
The good news is that there is no need to hurry and start using AI today. The bad news if you aren’t a tech-lover is that it is time to start finding out about AI.
The State of AI
Most of what you have heard about AI is about large language models [LLM]. This initial version of AI is quite limited in what it can do. An LLM takes a prompt or question and starts expanding on that by predicting what word is likely to come next. It has billions upon billions of examples to draw from. If it always used the MOST likely word in building its response, it would be like an encyclopedia of facts. So a tiny bit of randomness is introduced about which of the most likely words to choose. The response often becomes creative.
It can also be crazy.
The very latest versions of AI add a problem-solving routine that allows it to actually do higher level math and ‘reason’. So it will definitely get interesting.
A really good article about how AI works is the one linked below. It is more of a monograph than an article but you will be glad you stuck it out.
What Is ChatGPT Doing … and Why Does It Work?
How AI Will Impact Architects
When CAD came along, it took almost 20 years before most firms were using it exclusively. AI will take a while too as individual aspects of the practice get their apps. I wouldn’t be surprised it it takes 20 years before the practice of architecture is completly transformed.
The integration of AI into architecture has plenty of applications that are already being planned that will transform the way buildings get designed. Here are some current use cases in various stages of development:
Generative Design: AI algorithms can explore numerous design alternatives based on specified parameters such as site constraints, building codes, material availability, and aesthetic preferences. This allows architects to quickly evaluate various options and select the most efficient designs.
Energy Efficiency Optimization: AI can analyze architectural designs to optimize energy consumption. By simulating different scenarios, it can help in determining the best placement of windows, insulation materials, and HVAC systems to minimize energy use.
Building Information Modeling (BIM): AI can enhance BIM by automating updates and generating insights from building performance data. It helps in tracking changes and predicting maintenance needs throughout the lifecycle of the building.
Site Analysis: AI tools can assess geographical and environmental data to determine the best site for a small building. This includes analyzing local climate, soil conditions, and potential environmental impacts to optimize site selection.
Cost Estimation and Budgeting: AI can significantly improve the accuracy of cost estimates by analyzing historical data on materials, labor costs, and project timelines. It can also identify potential cost overruns early in the design phase.
3D Modeling and Visualization: AI-driven software can create highly detailed 3D models and visualizations of designs, allowing clients and stakeholders to better understand the proposed projects. This aids in gathering feedback and making informed decisions.
Space Optimization: AI can suggest optimal layouts for interior spaces to maximize functionality and comfort. By analyzing user behavior and preferences, it can recommend designs that enhance usability and flow.
Sustainability Assessment: AI tools can evaluate the environmental impact of design choices, helping architects select sustainable materials and construction methods. They can also simulate different scenarios to assess the long-term sustainability of a building.
Construction Planning and Scheduling: AI can streamline the construction process by generating efficient schedules, predicting delays, and optimizing resource allocation. This improves project management and reduces potential setbacks.
User Experience Enhancement: AI can analyze user interactions and preferences in real-time to suggest design modifications that improve the overall user experience, such as lighting, acoustics, and accessibility considerations.
Smart Building Integration: AI can facilitate the design of buildings equipped with integrated smart technologies that manage lighting, climate, security, and energy consumption, enhancing overall functionality and user comfort.
Regulatory Compliance: AI can search for codes and regulations that apply to your project and produce a ‘recipe’ for compliance.
BTW These lists were generated for me by ChatGPT 4o mini.
Limitations Of AI Today
For Architects the current AI offerings face several significant limitations. For the typical small-firm architect it is premature to embrace AI.
Technological Constraints
Computational Demands
Generative design requires substantial computational resources, which can be costly and may limit accessibility for smaller firms or individual designers. The need for significant processing power can slow down design iterations and exploration.
Software Compatibility
Integrating generative design tools with existing CAD/CAM systems can be challenging, leading to workflow disruptions and data transfer issues. This lack of seamless integration can hinder adoption and efficiency.
Design Process Challenges
Limited Iterative Capabilities
Generative design tools often struggle with accommodating design revisions and iterations. The process assumes a linear progression, making it difficult for designers to work iteratively as they typically do in real projects.
Overwhelming Number of Options
While generating numerous design alternatives is a strength, it can also be a drawback. Designers may become overwhelmed by the sheer volume of options, making it time-consuming and challenging to evaluate and select the best solutions.
Difficulty in Comparing Options
Evaluating and comparing a large number of generated designs can be arduous and complex, especially for more intricate projects like buildings. This can potentially slow down the decision-making process.
Human Factors
Skill and Knowledge Requirements
Effective use of generative design software often requires specialized skills in AI, machine learning, and parametric modeling. This learning curve can be steep for many designers and engineers.
Potential for Over-reliance on Technology
There’s a risk of over-dependence on generative design tools, potentially sidelining human creativity and intuition in the design process. This could lead to a homogenization of design solutions.
Limitations in Design Understanding
Lack of Context and Nuance
Generative design software may struggle with understanding complex contexts, cultural nuances, and subjective design elements like aesthetics or emotional impact. This can result in designs that are technically sound but lack depth or appropriateness in certain situations.
Imprecise Performance Metrics
The software often relies on quantifiable metrics for evaluation, which may not capture all aspects of design quality. This can lead to optimizing for easily measurable factors while overlooking important but less quantifiable design considerations.
Ethical and Legal Concerns
Ownership and Liability Issues
The use of AI in generative design raises questions about design ownership, responsibility, and liability, which are not yet fully resolved in legal frameworks.
Data Privacy and Security
Handling large amounts of data for generative design processes can raise concerns about data privacy and security, especially when dealing with sensitive project information.
Some of these limitations remind me of the move to CAD back in the 80s, but AI has lots more.
Next Steps
So here’s what I recommend.
Try It
There are a lot of AI apps for you to experiment with. Find a free one. My go-to app is Perplexity.ai. It is free, and the free version is all you need to get started. The Pro version is $20/mo if you start to find it valuable. I use Perplexity instead of Google for searches ever since I tried it.
Claude is another app that I have used.
AI is probably embedded in several apps you use all the time. That is another way to try it out.
Prompts
You will quickly learn that garbage In - garbage out applies to AI apps. So you should be on the lookout for some guidance on how to configure your prompts (requests) for best results. Here’s one I got from Seth Godin.
Watch for AI architectural apps
Start by putting this prompt into your selected AI app: “What AI tools exist today for small-firm architects”. Do this every month or so to keep abreast.
Work on your processes
Using AI will be even more powerful for you if your processes are documented including the prompts that you use to get the results that you want.