MAIA is a multimodal agent for neural network interpretability tasks developed at MIT CSAIL. It uses a vision-language model as a backbone and equips it with tools for experimenting on other AI systems.
MIT researchers advance automated interpretability in AI models
This was originally published on post