RT @ClaBat9: 🔑 The camera-ready version of our #UAI2025 work, "The Relativity of Causal Knowledge" with @officiallydac , is online at https….

Crisis = Opportunity. I lost my funding due to federal freezes at Harvard. Seeking roles to build next-generation agentic AI using RL+LLMs. 🔥 DMs open.Expertise: #RL #LLMs #Robotics #GenAI #DataScience .Published in top-tier CS venues.

Join millions who have switched to Grok.

RT @roydanroy: This workshop looks genuinely like a great step for statistics. What I like most is that it is young star researchers as spe….

Thrilled to teach a hands-on tutorial on LLMs as Autonomous Agents at @harvard_data ! 🚀Reasoning, tool use, memory, ReAct from scratch, SFT & RL! Don’t miss the fun! The fully interactive Colab notebooks here: Please star the Github repo if useful 🤩!

RT @karl_tuyls: 🚀 What is an agent? A question that’s been tackled for decades is now resurfacing as we enter the next phase of frontier AI….

RT @eliasbareinboim: If you are interested in how to improve decision-making, and more fundamentally, how causal inference & RL are related….

RT @shagunsodhani: There are couple of other sister teams in FAIR that are hiring in areas related to LLMs, reasoning, agents, social inter….

RT @JiahengHu1: Excited to present two papers at #NeurIPS on Wednesday, Dec 11th! 🧠.1. DUSDi (: 11am–2pm, West Ball….

RT @PeterStone_TX: Check out this work on learning a basis function for all possible RL policies in a state-action space. It supports quic….

RT @francescadomin8: This is a dream job within a dream team, in close collaboration with lab and within @harvard_data.

RT @JiahengHu1: 🚀 Despite efforts to scale up Behavior Cloning for Robots, large-scale BC has yet to live up to its promise. How can we bre….

AI isn’t a panacea for climate adaptation, nor is it yet trustworthy enough for immediate deployment. But it’s a promising area of research, and we need the academic community to co-design solutions with stakeholders, addressing real-world needs. 7/🧵.

These very concerns are where AI shows promise. Foundation models excel in sample efficiency and generalization across diverse regions and communities. Unlike traditional statistical models, deep learning can handle misaligned, unharmonized data. That being said. 6/🧵.

Common concerns fueling skepticism around data-driven climate adaptation policies include (1) limited local data for many regions and (2) difficulties in harmonizing data across geographies and timescales, complicating statistical analysis. Ironically….5/🧵.

I gained insights into the current landscape of data systems, including impressive initiatives like @dhis_2. Many countries lack or are developing an interoperable climate and health data infrastructure. Smart, adaptive AI-driven policies require such systems. 4/🧵.

AI has the potential to optimize resource allocation and strengthen early warning systems as part of climate adaptation efforts. Though we still have much to accomplish, research on LLMs and GenAI can make these solutions more practical and trustworthy than ever. But. 3/🧵.

The meeting focused on developing a 3-year agenda to integrate climate and health data and surveillance systems. It brought together the WHO, WMO, technical partners, and funders to shape a collective vision. Read more about it here: 2/🧵.

Two weeks ago, I had the incredible opportunity to present our work at the @WHO and @WMO on RL and Foundation Models for Climate Adaptation (supervised by @francescadomin8). It was an inspiring event. Some of my takeaways and insights on the role of AI:.1/🧵(opening thread)