1/ 🗞️🚨 Do great minds really think alike? 🧠🤖👨🏽‍🦱 We investigate Human-AI Complementarity in Question Answering using CAIMIRA, our item response theory based neural framework! w/ Hal Daumé III Tianyi Zhou Jordan Boyd-Graber Paper Link: arxiv.org/abs/2410.06524 #nlp #human #ai #irt #qa

Covering diverse facets of natural language processing —from multimodal and multilingual systems to education and social science settings—researchers UMD CLIP Lab have 21 papers accepted EMNLP 2025! Congratulations to our researchers and see you next month in Miami! 😎 #EMNLP2024

"If former President Donald J. Trump and his running mate J.D. Vance win, they could use the federal government to remake higher education as we know it." 🧵The The Chronicle of Higher Education lays out what's at stake: (1/x) chronicle.com/article/a-seco…

If you are concerned about imminent political threats to higher ed, this American Association of University Professors report is a great place to understand what the blueprint will be in the coming years-- aaup.org/report/report-…

I'm seriously concerned about attacks on (higher) education over the next 4 years. Restrictions on federal funding conditioned on universities playing ball is a very real possibility, among others. If you are in a US university and can join a union, please consider doing so!

Moving over to b*sky. Find me at haldaume3.bsky.social. I have no friends there yet - you could be the first! :)

🚨New dataset + challenge #EMNLP2024🚨 We release ASL STEM Wiki: the first signing dataset of STEM articles! 📰 254 Wikipedia articles 📹 ~300 hours of ASL interpretations 👋 New task: automatic sign suggestion to make STEM education more accessible microsoft.com/en-us/research… 🧵

No AI system is perfect. Instead of hiding AI's flaws, what if we could leverage them to help users? Our #CSCW2024 paper on Seamful XAI shows exposing and leveraging AI weaknesses can improve user understanding & agency. w/ Mark Riedl Vera Liao Samir Passi Hal Daumé III

New #emnlp2024 paper (oral presentation on Nov 12 (tomorrow), 12:15-12:30 Monroe): TL;DR: When a model fails, it can still guide a human to success. We show that uncertainty communication improves human decision making, in the context of navigation! Why is this exciting?

All AI systems make mistakes. 🧐 What if users could leverage AI flaws to understand it & take informed actions? 🚀 Our #CSCW2024 paper on Seamful XAI offers a process to foresee, locate, & leverage AI flaws—boosting user understanding & agency. But why should you care?⤵️ 1/n

Today is the registration deadline!

Please join us for: AI at Work: Building and Evaluating Trust Presented by our Trustworthy AI in Law & Society (TRIALS) institute. Feb 3-4 Washington DC Open to all! Details and registration at: trails.gwu.edu/trailscon-2025 Sponsorship details at: trails.gwu.edu/media/556

📌NLI Improves Compositionality in Vision-Language Models is accepted to #ICLR2025! CECE enables interpretability and achieves significant improvements in hard compositional benchmarks without fine-tuning (e.g., Winoground, EqBen) and alignment (e.g., DrawBench, EditBench).

We created COALA, a visual analytics tool that helps understand how native & non-native English speakers write differently when writing together. Our findings can help build better AI writing assistants for teams. Related #CSCW paper: dl.acm.org/doi/10.1145/36…

TRAILS Director Hal Daumé (Hal Daumé III) joined "AI or Not" the podcast with Pamela Isom to talk about his shift from technical AI research to its societal impact—touching on governance, copyright, and the need for stronger regulations. go.umd.edu/1yvy