3 papers indexed
Chih-Ting Liao, Fei Shen, Xin Cao, Tat-Seng Chua
The standard way to read latent knowledge out of a model, a linear probe confirmed by a steering recovery, can systematically overstate what a vision-language model (VLM) actually grounds in the image. We show this on spatial reasoning, where the error is invisible to both probin…
Cong Wang, Haiyu Wu, Zhiwei Jiang, Zifeng Cheng, Fei Shen, Yafeng Yin, et al.
Concept erasure aims to prevent image generative models from producing unsafe content while preserving their general generative capability. Meanwhile, next-scale autoregressive (AR) image generation has recently emerged as a new generative paradigm characterized by next-scale pre…
Abstract As artificial intelligence (AI) increasingly reshapes human communication, the attribution of agency to machines has become a central scholarly concern. Yet there is little consensus on how users psychologically construe machine agency in human–AI interaction. To address…