Computer Science at UT Austin

@UTAustin is launching a new School of Computing in fall 2026! With Information and Statistics & Data Science, we’ll expand student opportunities, accelerate research, and strengthen pathways to high-impact careers and grad study. Read more: utex.as/3OQFrIh

Jiayi (Jane) Chen (@Jane_JiayiChen) from @UTCompSci and @ldosexpedition gave a great talk on integrating ML for Systems in the Real World including congestion control (Unum) and CDN caching (Darwin) at UIUC Systems Seminar. Full house; great discussion! Check out her cool work: janecjy.github.io

VLA models are capable generalists. But can they continually self-improve? Such Continual Reinforcement Learning (CRL) problems are traditionally considered very challenging. Surprisingly, we found that with the right setup, the simplest CRL recipe can work really well! arxiv.org/abs/2603.11653

UT Austin is shaping the future of digital twins after major advances in AI, math and supercomputing. TAMEST Member Karen E. Willcox, Ph.D. (NAE) and the @OdenInstitute are driving physics‑informed, AI‑powered models across critical fields. Read more: buff.ly/Zww6jxd

We have seen rapid progress in humanoid control — specialist robots can reliably generate agile, acrobatic, but preset motions. Our singular focus this year: putting generalist humanoids to do real work. To progress toward this goal, we developed SONIC (nvlabs.github.io/GEAR-SONIC/), a Behavior Foundation Model for real-time, whole-body motion generation that supports teleoperation and VLA inference for loco-manipulation. Today, we’re open-sourcing SONIC on GitHub. We are excited to see what the community builds upon SONIC and to collectively push humanoid intelligence toward real-world deployment at scale. 🌐 Paper: arxiv.org/abs/2511.07820 📃 Code: github.com/NVlabs/GR00T-W…

Catastrophic forgetting has long been a challenge in continual learning. However, our new study found that pretrained Vision-Language-Action (VLA) models are surprisingly resistant to forgetting! Zero forgetting, or even positive backward transfer, is possible with simple experience replay. arxiv.org/abs/2603.03818

CMHC @ WCP (WCP 1.102) will offer extended hours this week on Monday thru Wednesday from 5–7 p.m. for students directly impacted by the tragedy downtown and seeking mental health support. 24/7 Support: 24/7 CMHC Crisis Line at 512-471-2255 TimelyCare at timelycare.com/utexas

Today it was confirmed that among those who lost their lives is one UT student. This is devastating, and we will remember her. Also, a number of our students sustained injuries during the attack, some of them very serious. I have met with many of their families and will continue to support and pray for them. In moments like this, we show our strength by coming together. Longhorns always look out for Longhorns. We lift each other up when we need it most. Now is one of those times. president.utexas.edu/message/walkin…

AI agents are here. How we design them now will shape trust, work and learning for years. Join leaders exploring ethical, human‑centered agentic AI for social good. 📅 March 4 | 9:45 a.m. 🔗 Register: bit.ly/45BGBgF #EthicalAI #TexasResearch #TexasAI

Congratulations to Zizhao Wang on becoming my 30th Ph.D. graduate with his thesis entitled "Causality-Inspired Reinforcement Learning: State Abstraction and Representation Learning". Committee members: Amy Zhang, Alessandro Lazaric, Sandeep Chinchali, and Roberto Martin-Martin.

As AI rapidly evolves, the School of Computing will allow the University of Texas to more deeply explore artificial intelligence for public use, a UT dean said. statesman.com/news/education…

Follow your curiosity to the Texas Science Festival for two full weeks of science-inspired offerings! #TexasScience #TexasScienceFestival #ScienceFestival news.utexas.edu/2026/02/11/tex…

UT Austin solves real-world problems with the power of computing. A new School of Computing in the College of Natural Sciences (@TexasScience) will combine our strengths in computer science, information, statistics, and data science to drive discoveries and develop technology leaders. Read more: utex.as/3OQFrIh

UT Austin is launching a new School of Computing, uniting computer science, information, statistics and data science. Opening Fall 2026, the school expands student opportunity and strengthens Texas’ computing talent pipeline. More: utex.as/4qH3RkR

🚨 Excited to share Reasoning Execution by Multiple Listeners (REMuL), a multi-party training method for faithful reasoning. Consistently boosts faithfulness evals (hint attribution, early answering, mistake injection) across diverse reasoning tasks while maintaining accuracy! ➡️ Faithfulness is key for CoT interpretability but current LLMs produce unfaithful reasoning that is hard to follow, with standard outcome-focused RL hurting faithfulness. ➡️ REMuL approaches faithfulness through the lens of executability. A CoT is faithful if independent "listener" models can follow/execute a truncated CoT prefix and reliably arrive at the same conclusion as the “speaker” model. ➡️ REMuL trains the speaker via GRPO to produce reasoning that achieves consistent answers among listeners, while maintaining correctness via masked supervised finetuning. ➡️ Interestingly, REMuL's multi-party training generalizes better. Directly optimizing for faithfulness metrics improves those metrics alone, but not others, while REMuL improves across metrics! 🧵👇

Alice Xiang (@SonyAI_global) will deliver the keynote at the Texas Symposium on Machine Learning, Responsible AI & Robotics Her work on human-centric AI datasets is featured on the cover of @NaturePortfolio. 🔗 View the program & register: bit.ly/45BGBgF

Join us and our partners at @TexasRobotics and the Machine Learning Lab on March 3 & 4 for a research symposium on responsible AI and robotics. 🔗 View the program & register for free: bit.ly/45BGBgF #EthicalAI #TexasResearch #TexasAI

Nearly 2 years ago, @janetlauyeung, @kanishkamisra, @valentina__py and I decided that it's time to bring discourse structure to the center of NLP teaching. We believe that discourse and coherence play important roles in today's long-form LLM generations and their evaluation, and that LLMs are great tools to study models of discourse. The challenge: how do we train students with this mindset? After sessions and sessions of brainstorming, slide-making, and homework design, and the help of amazing TA William Sheffield, we finally launched a cross-listed undergraduate course @UT_Linguistics and @UTCompSci last semester called "Computational Discourse and Natural Language Generation". We intentionally fostered an exploratory mindset: we discussed bleeding-edge work and engaged students in critical thinking and open-ended problems inside and outside of the classroom. We were really impressed by the kind of course projects students ended up doing! Teaching in the era of AI is challenging, but also what an exciting time to attempt something new! We will present this course at the @teaching_nlp workshop at #EACL2026 in 🇲🇦. Check out the preprint at arxiv.org/abs/2602.02878

On Thursday, I was honored to give a keynote at The 40th Annual AAAI Conference on Artificial Intelligence in Singapore. The title was "From *How* to learn to *What* to learn in Multiagent Systems and Robotics": underline.io/embed/146013-f…

This podcast by Carl Liebowitz is aimed at the general public. Should be accessible to all! youtube.com/watch?v=3dN7VT…