MIT researchers developed an AI copilot for topology optimization, a process that uses algorithms to decide where material can be added or removed in a structure. The model flags regions where a designer may want to make adjustments, using patterns from synthetic examples of human feedback.
In design demonstrations reported in Communications Engineering, the team said human-guided changes could improve manufacturability or increase linear buckling load by as much as 39%. The designers could make selected regions thicker or thinner; the paper does not give a numerical measure for manufacturability.
The system added about 15 seconds compared with the conventional algorithmic designs in the reported tests. The examples were not physically built or tested, so the study shows a design method, not yet evidence of performance in real structures.


