As AI reshapes the future of architecture, design, and visualization, Chaos Talks AI is bringing together leading voices to explore how innovation and creativity can thrive side by side.
Following the success of the first episode discussing AI’s role in architecture, design, and visualization, in the second episode, streamed live on November 6, Chaos Community Ambassador Phil Read dove deep into the topic of navigating the state of AI in AEC & the future. He was joined by:
- Kate Vera (Interior Designer & Educator, Katevera.creative)
- Cassidy Shipley (Product Manager of Graphics, SketchUp)
- Roderick Bates (Director of Corporate Development, Chaos)
Here are the highlights! 👇
AI and accountability in the classroom
The initial question Phil asked to kick-start the discussion focused on the implementation of AI within the context of education and accountability. He asked Kate: How do you see AI being implemented in education—not as an easy shortcut for students, but as a way to build accountability and better design judgment? Where do you draw the boundaries? In response, Kate shared how she approaches AI use across the full design process, emphasizing that thoughtful integration must align with each stage of a project’s development.
Stage-by-stage application
The integration of AI into architectural design requires a structured approach to ensure the tools are used effectively without neglecting fundamental design rigor. Kate, who is an interior design educator, outlined a full design project workflow, which typically consists of six different stages — from early ideation to final presentation.
In the concept and ideation stage, designers can turn to AI image-generation tools, such as Midjourney, or integrated solutions like Chaos Veras to spark inspiration. These tools help explore form, mood, and style quickly, offering visual variations that guide creative direction within familiar design environments.
During modeling, AI becomes a powerful productivity booster. Tools like the Chaos AI Material Generator can transform real-world photos into ready-to-use materials, enabling designers to build realistic 3D scenes more quickly and focus on refining details rather than repetitive tasks.
As the workflow progresses into rendering, AI helps streamline iterations and deliver high-quality results more efficiently. Enabling NVIDIA’s AI denoiser denoises the image during rendering, not just at the end.
Finally, in the post-production and storytelling stages, generative tools can aid in producing diagrams, visuals, and seductive short animations that clearly and compellingly communicate the design intent.
Seduction vs. rigor
A recurring theme in the episode was the "seductive" nature of AI-generated images. The ability to create beautiful visuals quickly can deceive designers into believing a problem is solved, bypassing the essential, often difficult, process of creative struggle and refinement.
💡Interesting quotes:
"How do we maintain good judgment and fear of failure with using tools that can create such seductive images so quickly, and you think, 'Oh, I must be done. This is beautiful.' But maybe it's too easy and beautiful sometimes." - Phil
"Designers are not thinking about the proportions, the flow, the light, the function of the space anymore, and you just jump from like this perfect image before the idea has even time to mature in your mind" - Kate
This raises an essential question: how and why does judgment tie in?
Kate shared an anecdote that perfectly illustrates this gap: a client questioned why her initial concept took three weeks to develop when their niece could generate an image with Midjourney in just 15 minutes. The client saw a finished product, failing to recognize the complex, professional thought process that turns an idea into a functional, buildable space.
Shepherds and sidekicks
A powerful mental model for organizing these tools is the "Shepherds and Sidekicks" analogy, proposed by Cassidy, which helps clarify the different functionalities of AI, particularly in achieving control and flexibility in the workflow. In this framework, you are the "director" who uses two types of AI:
- Shepherds are AI’s that "know more than you" and help with research and ideation. Use them to explore new creative possibilities or check complex information, like local zoning laws.
- Sidekicks are the "worker bees" that handle transactional, repetitive tasks to save you clicks. Use them to manage your door, window, and paint schedules.
Here is a breakdown of the two distinct roles:
|
AI Role |
Definition and Primary Function |
Activities and Focus |
Human Designer Role |
|
The Shepherd |
The AI that "knows more than you" and is used for ideation. |
Helps the designer hallucinate and push creative boundaries (e.g., "Hey, Varys, do something awesome with it"). Also handles knowledge retrieval (e.g., checking zoning laws). |
The shepherd aids the "drunk part" (creative exploration). The designer remains the director. |
|
The Sidekick |
The "workerbee" that manages and executes the decisions made by the director and shepherd. |
Focuses on transactions and maximizing efficiency by saving clicks. Handles routine administrative tasks such as generating door, window, and paint schedules. |
This is considered the "vastly more compelling" version of AI because it handles the detailed execution, allowing the human to focus on judgment. |
AI assistants, agentic AI, & year-in-review
The next live session of Chaos Talks AI will take place on December 11, 2025, at 4 PM CET.
Join us as we look back at a landmark year in AI — from the rise of intelligent assistants to the emergence of agentic systems. Our panel will explore 2025’s biggest breakthroughs, key lessons, and what these innovations mean for the future of design and visualization.
Engage with the panel, join the chat, and bring your questions to the table. We look forward to an inspiring discussion with you! Register here.