Workshop 01: Human AI Teaming: A Paradigm Shift from Predictive Models to Explainable Coupled Intelligence
Workshop provider:
Pia Fricker, Aalto University, pia.fricker@aalto.fi
Chaowen Yao, Aalto University, chaowen.yao@aalto.fi
Yichao Shi, Georgia Institute of Technology, yshi431@gatech.edu
Zubin Tan, Aalto University, zubin.tan@aalto.fi
Workshop description:
This workshop explores a critical topic in generative design: How to integrate design scenario generation and performance evaluation into an integrated workflow that is both computationally efficient and analytically interpretable?
Design Optimization Needs a New Paradigm – This workshop introduces a fast, easy to understand generative design workflow by combining design generation, simulation, surrogate modeling, and AI. Traditional environmental simulations are often too slow for the thousands of iterations required in modern design optimization. Replacing these slow simulations with AI and surrogate models creates a faster process while keeping the results clear and easy to interpret. The workshop focuses on three main research frontiers:
1. Explainable AI. This frontier moves away from “black box” systems. It focuses on transparency, trust, finding bias, ensuring fairness, and keeping human oversight in critical decisions.
2. Next Generation Optimization and Recursive AI. This approach links optimization with machine learning. The AI continuously learns from optimization results, improves its own predictions, speeds up convergence, and uncovers hidden design patterns.
3. Generative Design. This performance based method automatically creates many options within a multidimensional space. It then evaluates them against environmental, functional, or social goals to find the best solutions.
In this workshop, The teaching team will guide participants through an end to end workflow that connects generative design, performance simulation, AI based surrogate modeling, and explainable AI across multiple scales. The primary goal is to examine how performance-informed design workflows can become faster, more flexible, and easier to understand in architecture and urban design settings.
Expected Outcomes: 1. Participants will be able to build a compact but testable workflow involving surrogate model training, model interpretation, simulation and design generation in their own fields.
2. Participants will be able to pack a light-weight Grasshopper plugin for environment analysis or generative design.
Workshop type:
a 2-day hybrid (delivered on site with the opportunity to join online) workshop, with the 2nd day as an optional choice.
Number of Participants:
15-20, but up to 30 participants.
Workshop Schedule
Day 1, 07.Sep.2026
Morning:
Introductory Lecture discussing surrogate model, explainable AI, and generative design;
Software Environment Setup for the training and simulation.
Afternoon:
Case Workshop based on our Grasshopper plugin: Betula – A Rhino Plugin to Compute Tree Carbon Capture in Urban Environments;Q&A.
Day 2, 08.Sep.2026 (Optional)
Morning:
Participant Work Session; Q&A
Afternoon:
Feedback and Closing.
Required software and equipment:
A Laptop
Rhinoceros 3D and Grasshopper
Python (for coding and using libraries) or Jupyter/Colab Notebook.
Prerequisite:
Familiarity with basic Python.
Proficiency in Rhinoceros 3D and Grasshopper.
Familiarity with basic urban environmental analysis methods.
