Welcome back to @ClipUmd Mohit Iyyer (@MohitIyyer)! After earning his Ph.D. @UofMaryland and working at @UMass, he’s returned as an associate professor, bringing his lab and cutting-edge work on LLMs and web agents. Read more: https://t.co/zK50qLktJ5

We’re excited to welcome Julia Mendelsohn to @umiacs and the CLIP Lab! Mendelsohn investigates how language, politics and AI intersect—examining implicit bias in political discourse and the social risks of large language models like ChatGPT. Learn more: https://t.co/t9ErbmZM3r

A clever initiative from researchers in @UMDscience and @UMIACS, where humans team up with AI technology in quizbowl competitions to improve collaboration between humans and AI. #AIMaryland

The @UofMaryland is hosting a media event to showcase a broad range of #AI activities on campus. Several faculty and students @umiacs participated, including Hal Daumé III (@haldaume3) and Naitri Rajyaguru (@naitri_r) from the Perception and Robotics Group.

In this week’s edition of #TrustTRAILS, Susan Ariel Aaronson (@AaronsonSusan) explains the important relationship between data and AI, and the need for more accurate, complete and representative datasets that are used in AI-infused systems. Aaronson is a research professor

What does it take to advance AI literacy? Our experts—Virginia Byrne (@VirginiaLByrne) from @MorganStateU, David Broniatowski (@Broniatowski) from @gwuengineering, Hal Daumé III (@haldaume3) from @UofMaryland, and Brandeis Marshall (@csdoctorsister) of @DataedX_—explain how

“If you look at who is building AI technology in the world today, it tends to be a very narrow slice of the world’s population," says Hal Daumé III (@haldaume3), a professor of computer science @UofMaryland and director of @trails_ai. In this edition of #TrustTRAILS, Daumé

In a paper just published in Computational Linguistics, @ClipUmd's Philip Resnik (@psresnik) argues that "baked-in" bias in current #AI-infused large language models calls for deeper solutions. https://t.co/mSLFezC0dx

“The crowd chuckled when Daumé recalled how 2 early LLMs developed by Terps were named Bert & Elmo in a nod to famed UMD alum & Muppets creator Jim Henson.” Thanks @haldaume3 for mentioning work by @ClipUmd alums Jacob Devlin & Mohit Iyyer (@MohitIyyer)! https://t.co/PXrkl1UJJD

A team of @UofMaryland researchers led by Jordan Boyd-Graber (@boydgraber) is using a revamped #quizbowlformat to explore how #AI and humans can best work together. Their goal? Improve the evaluation of AI-infused question and answering systems by comparing how models and human

Two @ClipUmd alums—Weijia Xu(@weijiavxu) & Sudha Rao(@raosudha89)—published a paper @PNASNews quantifying the lack of plot diversity in LLM-generated stories, finding that they often contain idiosyncratic plot elements echoed frequently across generations.

Philip Resnik (@psresnik) is working with researchers from @umdcommdept, @UMDPublicHealth and @UMmedschool to boost HPV vaccine uptake. Backed by a $2.8M @NIH grant, they’re building an AI chatbot to help parents make informed vaccine choices. Read more: https://t.co/ZY35g8hhWC

Matthew Baney (pictured far right), is the assistant director of computational systems at UMIACS. In this role, he keeps high-performance computing and data centers running smoothly while mentoring the next generation of tech talent. Learn more: https://t.co/8Xp766k9Rx

Douglas Oard has stepped in as interim dean @INFOCollegeUMD. He’s driving interdisciplinary research collaborations at the college and @umiacs in AI, language technology, human-computer interaction and more. Read the story: https://t.co/ocHoWJbnfW @UMDscience @UMDResearch

Glad my small contribution could support Ai2's efforts to rigorously benchmark AI agents 🫡 Huge congrats to the team!!! 🎉

Do you want to see if ChatGPT can stop you from getting scooped, make the easiest $75 of your life, and chat with an amazing Ai2 intern? Join Hita's user study!

It's a beautiful morning in Vienna! We're presenting three papers today on: 1) Multiple Choice QA and Why it's Hurting AI 2) Better Automatic Evals for QA 3) What Makes a "Bad" Question and How Humans and Computers Deal with Them Videos and links to papers in the thread.

Excellent work by Pranav Goel Ph.D. ’23 (@pranavgoel.bsky.social) & @psresnik undertaken when Pranav was completing his Ph.D. in the CLIP Lab. https://t.co/gl1w8b6KQX

A new round of $750K in seed funding has been awarded to faculty and students who are advancing trustworthy AI at all four of TRAILS’ academic institutions: @UofMaryland, @GWtweets, @MorganStateU, and @Cornell. From disaster response to education, copyright law, and AI red

I am at #ICML2025! 🇨🇦🏞️ Catch me: 1️⃣ Today at the @WiMLworkshop mentoring roundtables (1-2pm in W211-214) 2️⃣ Presenting this paper👇 tomorrow 11-11:30 at East #1205 3️⃣ At the Actionable Interpretability @ActInterp workshop on Saturday in East Ballroom A (I’m an organizer!)

In earlier work, we showed that neural topic model evaluation was broken, and those models didn't improve over classical methods the way people thought. This new paper provides a replacement paradigm that's grounded in the real-world requirements of qualitative content analysis.