Michigan AI

🚀 Research highlight!
"MoE3D: A Mixture-of-Experts Module for 3D Reconstruction" proposes a simple yet effective approach to enhance the performance of feed-forward 3D reconstruction models.

By Zichen Wang, Ang Cao, Liam J. Wang, Jeong Joon Park

Check it out!
https://arxiv.org/abs/2601.05208

🚀 Research highlight!
"MoE3D: A Mixture-of-Experts Module for 3D Reconstruction" proposes a simple yet effective approach to enhance the performance of feed-forward 3D reconstruction models.

By Zichen Wang, Ang Cao, Liam J. Wang, Jeong Joon Park

Check it out!

MoE3D: A Mixture-of-Experts Module for 3D Reconstruction We propose a simple yet effective approach to enhance the performance of feed-forward 3D reconstruction models. Existing methods often struggle near depth discontinuities, where standard regression losses encourage spatial averaging and thus blur sharp boundaries. To address this issue, we introduce...

is moving fast in , but are we measuring what actually helps patients? 🤔

As Jenna Wiens points out in this interview in MIT Technology Review:
"The problem is that many providers aren’t rigorously assessing how well they actually work."

Prof. Wiens highlights a critical gap: we often lack evidence that AI tools truly improve patient outcomes. A call for more meaningful evaluation:

https://www.technologyreview.com/2026/04/24/1136352/health-care-ai-dont-know-actually-helps-patients/