🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

We are launching HALoGEN💡, a way to systematically study *when* and *why* LLMs still hallucinate. New work w/ Shrusti Ghela* David Wadden Yejin Choi 💫 🧵 [1/n]

New blog post: let's talk about latents! sander.ai/2025/04/15/lat…

Academia or industry, expert or novice, infrastructure shouldn't be your bottleneck. Garden democratizes access to OpenCatalyst models for ALL researchers. This is how tomorrow's breakthroughs will be found. Garden is a new superpower for scientists.

Scientific discovery with LLMs has so much potential yet is underexplored. Our new benchmark **LLM-SRBench** enable rigorous evaluations of equation discovery with LLMs! 🧠Key takeaway: Even SOTA discovery models with strong LLM backbones still fail to discover mathematical

Wish I was a pre-doc! This is a great opportunity!

🚨New paper out in PNASNews ! Existential AI risks do **not** distract from immediate harms. In our study (n = 10,800), people consistently prioritize current threats - bias, misinformation, job loss - over sci-fi doom! 💥👉 pnas.org/doi/10.1073/pn…

Appreciate it.

🚨 Why Do Multi-Agent LLM Systems Fail? ⁉️ 🔥 Introducing MAST: The first multi-agent failure taxonomy - consists of 14 failure modes and 3 categories, generalizes for diverse multi-agent systems and tasks! Paper: arxiv.org/pdf/2503.13657 Code: github.com/multi-agent-sy… 🧵1/n

🚨 Our latest paper is out today in Science! We uncover stark and systematic partisan differences in the amount, content, and character of science used in policy, which mirror differences in political elites’ trust in science. Four years in the making. Led by Zander Furnas 1/n

Here's some good news! The file drawer problem may have diminished in recent years, at least in social science survey experiments "This suggests increased recognition of the importance of null results."

"A society that can no longer read complex texts may soon find itself unable to think complex thoughts [...] reading becomes not just a cognitive act, but a civic one: a rehearsal for the intellectual stamina that democracy requires." 🎯

🧪 Want an AI-generated paper draft in just 1 minute? Or dreaming of building auto-research apps but frustrated with setups? Meet tiny-scientist, a minimal package to start AI-powered research: 👉 pip install tiny-scientist 🔗 github.com/ulab-uiuc/tiny… #AIAgent #pythonpackages

Again, I think academics have perhaps not quite groked what this sort of wholesale removal of a model means for replication in science. This isn’t versioned or downloadable, and you won’t be able to recreate it

📢 New paper with Iason Gabriel is out! 2025 is being called the year of AI agents, with overwhelming headlines about them every day. But we lack a shared vocabulary to distinguish their fundamental properties. Our paper aims to bridge this gap. A 🧵

Watch our team explain how you can use the FutureHouse Platform to come up with new hypotheses and make new discoveries.

FutureHouse releases AI tools it claims can accelerate science | TechCrunch techcrunch.com/2025/05/01/fut…

A new webinar series is coming soon 🎤 Designed for anyone who writes—emails, reports, blogs, posts—and wants to use AI as a tool for clarity, not chaos. 💬 Prompt better ⚡ Write smarter 🧠 Keep your voice Watch this space 👀 #AI #writing #ChatGPT #ProductivityHacks

Lately, I've been seriously exploring what it could mean to move beyond the benchmarking paradigm in ML evaluation and it's led to some stats-y papers: (1) a critique of current experimental evals of prediction based interventions & (2) a framework for adverse events reporting.

Apply to Lucy as a #PhD student next cycle. What an opportunity. #AcademicTwitter

Seth Lazar It is necessary to invest in alternatives right now. Without it, we might see the worst aspects of our current platform economy amplified. We don't have all the answers in the paper, but we have a blueprint for where to start. Paper: arxiv.org/pdf/2505.04345