Haley L.

Grateful to @stanfordhai.bsky.social and @facct.bsky.social for highlighting our work!

Computer-vision research powers surveillance technology - Nature An analysis of research papers and citing patents indicates the extensive ties between computer-vision research and surveillance.

New paper hot off the press www.nature.com/articles/s41...

We analysed over 40,000 computer vision papers from CVPR (the longest standing CV conf) & associated patents tracing pathways from research to application. We found that 90% of papers & 86% of downstream patents power surveillance

1/

Professor Thomas Dee analyzed attendance data from five school districts in the southern Central Valley and found a spike in daily absences this past January and February. These districts, which serve over 100,000 students, saw considerable jumps in all age groups. stanford.io/3HVr6Hb

#EduSky

In race to attract data centers, states can forfeit hundreds of millions of dollars in tax revenue to tech companies Amazon, Meta and Google are among the beneficiaries of state tax exemptions for data centers that help power AI.

"Today nearly all states that have sales taxes offer data centers exemptions on computers, wires, air-conditioning units and, in some cases, even the energy required to keep it all running." www.cnbc.com/2025/06/20/t...

thanks Shaily!!

Flattered and shocked for our paper to receive the #facct2025 best paper award.

faculty resists against state control over the campus

Brilliant presentation by @sepehrvakil.bsky.social and Mahdi Ganjavi on their new @mitpress.bsky.social book about engineering activism during the Iranian revolution. They close with this picture and caption: "faculty resists against state control over the campus"

Thanks to @stanfordhai.bsky.social for supporting this project!

Our findings offer insight into the role of #ChatGPT in the reproduction of language ideologies which conflate producers of “good English" with producers of “good research." Future work should look into this effect in other writing scenarios, such as #education and #work.

Authors, aware of this development, described removing ChatGPTisms and other clues which could expose their “non-native” status to reviewers.

Though ChatGPT may mask some signs of language background, reviewers explain that they now use ChatGPT “style” and non-linguistic features as indicators of author demographics.

When we asked about their preferences for English *writing style* in scientific papers, interviewees instead indicated how they associate native English *speakers* with the quality of science described.

Descriptively analyzing almost 80k reviews, we find that for the most part, the answer is not really. Changes were subtle and mostly insignificant except in a few cases. To understand why, we interview 14 scientists from around the world.

Using the case of computer scientists submitting manuscripts to @iclr-conf.bsky.social, we ask, do scientists from countries where English is less widely spoken receive fewer writing critiques in peer reviews after ChatGPT becomes available?

"You Cannot Sound Like GPT": Signs of language discrimination and resistance in computer science publishing LLMs have been celebrated for their potential to help multilingual scientists publish their research. Rather than interpret LLMs as a solution, we hypothesize their adoption can be an indicator of exi...

Many are lauding the potential of ChatGPT to "help" people with their writing. Looking at writing not as a deficit to be fixed, but as a site of social interaction between writers *and* evaluators, our #facct2025 paper explores how this played out in a real life situation. arxiv.org/abs/2505.08127