🚨 Our paper was accepted at AI, Ethics, and Society Conference (AIES) happening in October! #AIES We compare human national security experts vs. LLM simulations in wargames. The results? Surprising differences in decision-making that could impact real-world conflicts. Paper: arxiv.org/pdf/2403.03407

🚨 Our paper was accepted for Conference on Language Modeling! As we face a mental health crisis and lack of access to professional care, many turn to AI as a solution. But how does ethical automated care look like and are models safe enough for patients? Paper: arxiv.org/abs/2406.11852

Should AI be aligned with human preferences, rewards, or utility functions? Excited to finally share a preprint that Micah Carroll Matija Franklin Hal Ashton & I have worked on for almost 2 years, arguing that AI alignment has to move beyond the preference-reward-utility nexus!

Incredibly valuable work from a stellar group: must read!

Great and important work by xuan (ɕɥɛn / sh-yen) et al.!

The woman who is in Founder Mode getting called a “toxic leader” has its own version in AI research for female AI researchers as well. Breaking ppl’s preconceptions can lead to a lot of unnecessary fear and stigma. Kudos to Sophia for this post ✨

Couple weeks left to get in your abstracts. We'll also have invited papers from Henry Farrell and Hahrie Han, Arvind Narayanan, Alondra Nelson, Deirdre K. mulligan, Daniel Susskind and others—leading thinkers on politics, economics, and technologies of democracy. V keen for PhDs/ECRs to apply.

Can #LLMs generate novel research ideas? Check out our one year-long human study with 100+ NLP researchers, done by the amazing CLS

I used to be shy to publicly state that I had both of my kids during my PhD, but now I'm like: "Damn straight, I had two babies AND STILL managed to become a doctor in 5 years!" 👩‍🎓Academic Mom, PhD

The results in "LLMs Are Superhuman Forecasters" don't hold when given another set of forecasting questions. I used their codebase (models, prompts, retrieval, etc.) to evaluate a new set of 324 questions—all opened after November 2023. Findings: Their Brier score: .195 Crowd

🚨 PrivacyLens evaluates #LLMs' privacy awareness using contextual integrity theory! 🔍 It goes beyond static QA to detect privacy leaks 🔐 during LLMs agents' task execution process! Overall, a multi-level approach to evaluate #privacy awareness of LLMs in action ✨