Towards A New Design Computation With Artificial Intelligence

Design computation are the underlying computational techniques that support computational design methods. These are things like

Notes ↓

In addition to automation (meaning reduced task time), I think it’s important to think about what the automation means in terms of peoples’ mindsets. AI tools can begin to fulfill the role of collaborators, enabling designers to shift to editors over authors, critics over generators, predictors over observers, and perhaps even more technologist and less technician.

We’ll begin thinking about design problems differently. What data can we train on, how do we get those data, how do we author, evaluate, and iterate on the models of the world we’ve created, and the AI will execute for us. We saw hints of this with genetic algorithms, but they took so long to execute and were so difficult to encode that they never really achieved their potential… But they could now.

Even today I evaluated a new model from NASA and IBM that shows the ability to predict geographic and geologic changes from satellite imagery. It’s a marvelous modern age we’re entering.

Accuracy, insights, depth of investigation What would have taken months can take minutes, and we can ask questions we wouldn’t have had time to ask before.

We might find molecules and medicines that we would never have found.

In order to make spatially native AI models, we need design computation to model things computationally. WE need to abstract reality in a way that make things