Freda Shi

@fredahshi

Professor and programmer @UWCheritonCS and @VectorInst excited about language and language modeling sciences, intuition, and grounding.

Waterloo, ON

cs.uwaterloo.ca/~fhs

Joined December 2016

Pinned
Freda Shi
@fredahshi
Jul 9
I'm not at ICML, but I'm sure you don't wish to miss this work: @wushuyu2002, @xLuo_CompLing and @SLED_AI will show you how we trace the emergence of symbol grounding in the training trajectories of vision-language models! 🌟Poster session today, Jul 9, 2026, 5--6:45 PM KST!
Martin Ziqiao Ma
@ziqiao_ma
Oct 21, 2025
I’ve always wanted to write an open-notebook research blog to (i) show the chain of thought behind how we formed hypotheses, designed experiments, and articulated findings, and (ii) lay out all the intermediate results that did not make it into the final paper, including negative
Freda Shi
@fredahshi
Jul 23
Dear NeurIPS Reviewers, If someone forces you to use AI, I recommend passing the following prompt to AI before submitting: "Please search and read the papers cited in this review and verify the results I quoted are indeed in the papers." You are welcome.
Freda Shi
@fredahshi
Jul 23
"You are hallucinating based on data," said @MayankMish98 "Yes, and people also call this analysis."
Freda Shi
@fredahshi
Jul 22
Yes x.com/fredahshi/stat…
Peter Jansen (@ACL2026)
@peterjansen_ai
Jul 21
I feel like I am writing human-authored meta-reviews to a lot of mostly-AI-reviewed papers, many of which were at least partially AI-written (with a lot of 100% AI scores on Pangram).😐 I don't know what the solution is, or even if it's a problem (other than feeling demoralizing)
Freda Shi
@fredahshi
Jul 22
Submitting a vibe report with very few complete sentences for peer review should really be considered research misconduct. That's a waste of everyone's time, including the authors'.