I spent a lot of time cranking out those frames of manually drawn bedrooms. I thought to myself, lets build some dynamic blocks in autocad to help? So I ended up asking chatGPT for assistance in building an app that could help me make vaporware. Wait. Its the AI age. Why can’t I simply build a bad configurator with AI?
Well, I was successful in automating what would have been a 30 minute drafting task by using my entire labor day to have chat gpt walk me through everything to build this.
To be clear, I regret everything. This was tedious. Asking this AI to do things was fun, but needing to consistently check its work for errors became annoying. BUT I was able to generate some kitchens.
What I learned:
1)Generating a thing is the easy part
2) Generating a thing that takes into account adjacencies is a complete disaster
3) I think the feature to make beautiful or intuitive is how we allow users to set adjacencies.
4) If we don’t automate adjacencies, we will need to create a really fun slide things around interface
5) Perhaps the slide things around interface is the thing that creates the adjacency matrix.
6) How do we make a table of adjacencies make intuitive sense?
| Fridge | Sink | Range | Dishwasher | Cabinet | Wall | |
|---|---|---|---|---|---|---|
| Fridge | N | Y | N | Y | Y | Y |
| Sink | Y | N | Y | Y | Y | Y |
| Range | N | Y | N | N | Y | Y |
| Dishwasher | Y | Y | N | N | Y | Y |
| Cabinet | Y | Y | Y | Y | N | Y |
| Wall | Y | Y | Y | Y | Y |
Or ask the robot to explain this madness back to you:
Cabinet and Wall: Yes, Standard placement.
Fridge and Sink: Yes, Convenient but not ideal for workflow.
Fridge and Range: No, Heat and cold conflict.
Fridge and Dishwasher: Yes, No major conflicts.
Fridge and Cabinet: Yes, Common for storage.
Fridge and Fridge: No, Inefficient use of space.
Sink and Range: Yes, Safety risk, but possible.
Sink and Dishwasher: Yes, Ideal for dishwashing workflow.
Sink and Cabinet: Yes, Proximity to storage.
Sink and Sink: No, Redundant placement.
Range and Dishwasher: No, Heat damages dishwasher.
Range and Cabinet: Yes, Store cooking tools.
Range and Range: No, Redundant and unsafe.
Dishwasher and Cabinet: Yes, Store dishware.
Dishwasher and Wall: Yes, Saves space.
Dishwasher and Dishwasher: No, Redundant.
Cabinet and Cabinet: No, Avoid bulk, prefer variety.
The problem as I see it, might be similar to having adjacencies within every level of the model. Buildings, Rooms, Objects. If the input is a building, then we can surely record its adjacencies. Regardless, this to me is the simplest way to conceptually look at the data: