Computational
Design

building sustainable digital workflows


AI Troubleshooting Handbook

This handbook serves as a 24/7 AI Teaching Assistant Guide for students during daily coursework and computer-aided design. When stuck with modelling, scripting, or software troubleshooting, AI can offer instant solutions. Master AI to boost your design efficiency while retaining independent thinking and design ownership.


1. When to Ask AI?

AI excels at handling the following four types of specific needs:

  • Querying Hardware and Software Questions:
    • “I want to make holes in a model in Rhino. Besides Boolean Difference, what other common commands can I use?”
    • “What shortcut in macOS corresponds to Ctrl+C in Windows?”
    • “What are the minimum hardware requirements to run Unreal Engine for real-time rendering?”
  • Software Errors & Code Debugging:
    • When a Python script pops up IndexError: list index out of range, paste the error directly to AI: “What does this mean? How should I troubleshoot this in Grasshopper?”
  • Understanding Underlying Logic:
    • “Please explain clearly and simply the difference between Data Tree and List in Grasshopper, using an everyday example.”
  • Exploring Workflows & Design Logic:
    • “I want to quickly create 500 cubes of different shapes in Rhino. What efficient workflow do you recommend?”
    • “My mesh always tears when doing subdivision smoothing. Are there alternative approaches?”

2. When NOT to Trust AI?

Maintain critical thinking when using AI to avoid blind reliance:

  • Beware of AI Hallucinations:
    • AI might fabricate non-existent Grasshopper components or provide outdated commands.
    • Strategy: Provide the latest and most specific background context in your prompt. Pay attention to AI reasoning and sources, and verify them inside your software.
  • AI Cannot Replace “Design Intent”:
    • AI can tell you how to connect points into lines, but it doesn’t know why you want to connect them that way.
    • Spatial reasoning, conceptualization, and design logic must be driven by you.
  • Avoid “Black Box” Copy-Pasting:
    • For AI-generated code or scripts (Python / C# / GH components), always ask AI to explain the functionality of each module, and verify it manually before running.
    • For complex tasks, test in modular steps. This helps tremendously with future feature expansions and debugging.

3. How to Ask AI?

Vague questions lead to poor responses. Follow this simple prompt formula:

Context + Goal + Current State (Errors) + Specific Request = Efficient Solution

Context:

Software and version used (e.g., Rhino 8 / Python), operating system (Windows / macOS).

Goal:

The desired outcome (e.g., generate circles of increasing size along a 3D curve).

Current State / Errors:

What have you tried? Where are you stuck? What is the exact error message or abnormal behavior?

Specific Request:

The exact help you need from AI (e.g., provide 2-3 implementation methods and compare their efficiency).

The above information can be submitted via text, code blocks, screenshots, or voice input.

Bad Prompt Example

“My Grasshopper errored, the circles won’t turn into gradient sizes, what should I do?”

Good Prompt Example

[Context]

I am working on a parametric facade using Rhino 8 + Grasshopper on Windows.

[Goal]

I want to generate a series of circles with gradually increasing radii along an existing 3D curve.

[Current State]

I divided the curve into points, but when generating radii with Range and connecting to the Circle component, it turns red.

[Request]

What could be causing this? What are efficient workflows to create gradient circles? Please provide specific component names and connection logic.

4. What to Do If the Answer Is Wrong?

AI rarely gives perfect results on the first try. Problem-solving is an iterative conversation. Continue asking follow-up questions in the same thread:

  • Report Errors Directly

    I couldn’t find the XYZ component you mentioned in Grasshopper. Is the name incorrect or is it version-specific?

  • Narrow Down & Break Down

    This explanation is too complex. Could you elaborate specifically on Step 1 regarding Data Tree adjustments?

  • Ask AI for Testing Steps

    I’m not sure if the issue is with data structure or geometry. How can I test this step-by-step in the software?

5. Templates

Here are 3 ready-to-use prompt templates. Copy the code blocks below and replace the content inside the brackets ():

Error Diagnosis / Debugging

[Software & Environment]
(e.g., Rhino 8 + Grasshopper / Windows 11)

[Goal]
I intended to achieve: (Describe target effect)

[Issue]
After running, I got this error/abnormal behaviour: (Paste error text or screenshot)

[Attempted Troubleshooting]
I have tried: (Describe steps taken)

[Request]
Please analyze potential causes and provide 1-2 specific troubleshooting steps and fixes.

Idea Generation / Workflow Recommendations

[Design Scenario]
I am currently working on (e.g., parametric skin design / massing generation).

[Functional Need]
I want to achieve: (Describe form/function, e.g., louvers auto-adjusting to solar angles).

[Technical Constraints]
Limited to: (e.g., native Grasshopper components / Python script).

[Request]
Please provide 2-3 clear workflow ideas, broken down step-by-step, explaining pros and cons of each.

Problem Solving / Modular Code Generation

[Software Environment]
(e.g., Rhino 8 Python 3 Script)

[Input Data]
Input consists of: (Describe input, e.g., list of closed curves)

[Expected Output]
Expected result: (Describe output, e.g., geometries sorted by area with IDs)

[Safety & Clarity]
Please modularise the code and add comments explaining the core algorithmic logic.