🔥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

14/14 Thank you for reading! Please find our paper here: arxiv.org/abs/2502.04512 Work led by Ivaxi Sheth @ NAACL 2025, and done jointly with Jan Wehner, Ruta Binkyte, and Mario Fritz

Ivaxi Sheth @ NAACL 2025 Totally agree. This is really great and important work!

💥Representation Engineering is a new approach for controlling LLM behavior that has been exploding in popularity! 📜Our survey "Taxonomy, Opportunities, and Challenges of Representation Engineering for Large Language Models" spans 130 papers to answer all your questions about it

🚨 New Preprint Alert! 🚨 Are you interested in Imprecise Probability (IP)? Then check out our latest preprint "Truthful Elicitation of Imprecise Forecasts". Joint work with Siu Lun Chau and Krikamol (Hiring Postdoc). arxiv.org/abs/2503.16395 A quick thread🧵(1/3)

I am excited to present our work on safety for open-ended AI systems at the FAR.AI Alignment workshop tomorrow and SSI-FM workshop on 27th #ICLR25 🇸🇬 If you are also thinking about similar challenges or just curious, would love to chat!

Inspiring keynote by Tim Rocktäschel at ICLR on Open-endedness and automation of AI 🚀 However I hope that the unpredictable and uncontrollable nature of open-ended AI also inspires strong focus on Safety 👷‍♀️🦺👩‍🔬 x.com/sahar_abdelnab…

📢📢Our paper, "A Theory of Response Sampling in LLMs: Part Descriptive and Part Prescriptive" has been accepted at ACL 2025 (Main Conference)!! 🥳 Details below 👇🧵1/n

Hawthorne effect describes how study participants modify their behavior if they know they are being observed In our paper 📢, we study if LLMs exhibit analogous patterns🧠 Spoiler: they do⚠️ 🧵1/n