Jiahao Huang and Jun Wang, University College London · VLM Interpretability
Probing the "Conscious" Mind: Selective Access and Global Workspace in Vision-Language Models
The authors introduce the Multimodal Concept Lens (MC-Lens), a multimodal Jacobian lens that maps intermediate states of a vision-language model's vision encoder and language decoder into a shared semantic space defined by the decoder's output vocabulary. Using it, they report that concept information is readable in intermediate states and selectively accessed as tasks require, and that counterfactual interventions show this information affects downstream computation. They also propose two J-space metrics for tracking cross-modal alignment during training and report shared access patterns across several VLM families and sizes.
Open the source ↗Similar Research & Safety releases
- Sparks of Weight-Space Programming
Rikiya Takehi, Jayden Teoh, Alex Zhang, Omar Khattab (MIT CSAIL) · LLM weight editing · Week 46, 2026
- H-Neurons
Tsinghua University · Interpretability · Week 10, 2026
- Physically consistent AI models of chaotic systems
MIT Laboratory for Information and Decision Systems (Sunbochen Tang, Navid Azizan, Themistoklis Sapsis) · Physics-constrained Chaos Models · Week 48, 2026
- Posterior Matching Distillation
NVIDIA (Hidir Yesiltepe, Julius Berner, Arash Vahdat, Pinar Yanardag) · Diffusion distillation · Week 48, 2026
- Test-time adaptation for AI text detectors
MIT and Cornell Tech researchers (Kevin Ren, Manish Raghavan, Nikhil Garg) · Adaptive AI Detector · Week 48, 2026
- LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning
Eilam Shapira, Alan Arazi, Shoham Grunblat, Roi Reichart (Technion) · Target-aware tabular encoding · Week 48, 2026
Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.