Here are some of my predictions for the future. These are based on vibes and intuition, not careful analysis, so may be worthless, although I am willing to bet money on them. I've grouped them into 3 themes, in decreasing order of confidence: AI, US Politics, and Global Issues

Are you interested in exploring questions at the ethical frontier of AI research? If so, then take a look at this new opening in the humanity, ethics and alignment research team: boards.greenhouse.io/deepmind/jobs/… HEART conducts interdisciplinary research to advance safe & beneficial AI.

Personal update: After 8 amazing years in industry, I will join Computer Science and Engineering at Michigan in fall 2025. I look forward to putting more energy into mentoring, teaching, and collaborating across disciplines and organizations I will look for PhD students this cycle and a postdoc

Have you not read the platform of Kamala Harris and Donald J. Trump because they are too long? Maybe it is the time now! One week before voting! The amazing @jmendelsohn2 is leading an effort to use LLMs to help the public stay informed. In addition to comparative summaries,

Quick personal update: I am now a visiting scientist Abridge. It is exciting to see the frontier of AI in healthcare! Abridge is hiring at all levels. DM me if you are interested in internship or full-time opportunities.

I'll be hiring an intern working on an evaluation-related topic. If that sounds like you, please fill the form!

Today, UChicago announced the launch of the Institute for Climate and Sustainable Growth, a first-of-its kind institute that aims to balance the urgency of action on climate change with society’s need for sustainable growth. Learn more:ms.spr.ly/6013WPumL

I did not know this! Then I do not see any reason not to apply to TTIC if you are interested in AI or theory.

In case that there is still some undecided voters, here is a comparative summary of Kamala Harris and Donald J. Trump 's agenda (check out more from chicagohai.github.io/election-nlp/): Economy Harris: Proposes cutting taxes for middle-class families and ensuring the wealthiest pay their

Our department of computer science at The University of Chicago is hiring: Assistant Prof. in CS (apply.interfolio.com/157493) and also Assist. Prof in Data Science (apply.interfolio.com/153879)—join us at this incredible environment at UChicagoCS !

1/ ✨ Can Multimodal LLMs generate radiology reports from chest X-ray images? In our recent evaluation paper, we show: not yet.

Chat with Rosa Zhou if you are at EMNLP! Consider hiring her too!

On Tuesday, November 19th, Google DeepMind Senior Staff Research Scientist Been Kim (Been Kim) argues for a better way to enable communication between humans and machines as part of the DSI Distinguished Speaker Series and the Computer Science Distinguished Lecture Series!

1/ 🚀 New Paper Alert! Excited to share: Literature Meets Data: A Synergistic Approach to Hypothesis Generation 📚📊! We propose a novel framework combining literature insights & observational data with LLMs for hypothesis generation. Here’s how and why it matters.

I am @chenhaotan.bsky.social at the other platform. I will spend more of my attention and time there. I suggest that you try it out too!

How do everyday narratives reveal hidden cause-and-effect patterns that shape our beliefs and behaviors? In our paper, we propose Causal Micro-Narratives to uncover narratives from real-world data. As a case study, we characterize the narratives about inflation in news.

💡Check out our project website for our latest paper! Learn about a new approach to hypothesis generation: 👉 chicagohai.github.io/hypogenic-demo/

Sasha Rush (Sasha Rush) and I co-advise the amazing Wenting Zhao (Wenting Zhao)...on the academic job market this year.

Even ChatGPT knows. It is very hard to think of a positive spin for Rosalind here. A shocking display of ignorance and arrogance.

With RL actually working on LLM (e.g. o1/o3), progress in the next few years will be very counter-intuitive. LLM will have a more different skill profile than humans, making it important to study human+AI complementary approaches. (1/3)