Join the discussion on this paper page

Previous AI alignment efforts have focused primarily on first-order social norms -- teaching models what is socially acceptable or unacceptable (e.g., `do not steal'). However, social intelligence depends not only on norm recognition, but also on anticipating who will enforce it and how (e.g., public shame or even imprisonment). These second-order expectations, known as metanorms, govern how people respond when social rules are broken. We introduce a novel framework for evaluating metanorm reasoning in Large Language Models (LLMs) along two dimensions: emotional appraisal and behavioral response, and propose new classification tasks, namely, predicting self-regulation in violators, and other-regulation in observers. We release a multi-perspective dataset, NormReact, of 450 norm violation scenarios, hand-annotated for emotions and behavioral responses across norm violators' gender and observers' social closeness. Current LLMs portray a harsher social world: across six models, they overp

Buy Smart AI Robot Companion & Alarm Clock for Kids at dealbuddies.store! Fast US shipping. 100% money back guarantee.

MLST is the top technical AI podcast on Spotify. Subscribe now! Welcome! We bring you the latest in advanced AI research, from the best AI experts in the world. Our approach is unrivalled in terms of ...

1 likes, 0 comments - eye.candy.ai on March 9, 2024: "🌟Futuristic Visionary🌟 Eyes wide open to the future, she sees the world in a spectrum of p..."

1 likes, 0 comments - eye.candy.ai on January 21, 2024: "🌟 Timeless Beauty 🌟 Captivating simplicity within a glance. This portrait is a celebration..."

1 likes, 0 comments - eye.candy.ai on January 20, 2024: "🌟 Twin Visions 🌟 Mirror images with a shared soul; these digital twins capture the allure ..."

26K likes, 411 comments - ellecordova on November 29, 2023: "The ghost in the machine is about to start ghosting you #artificialintelligence #ai #machinelear..."

Opinions are divided about the potential impact of artificial intelligence as the response to a recent viral essay shows

New release expands AI actions, streamlines ticketing and omnichannel insights to help support teams resolve issues more efficiently...

The readings shared in Bluesky on 19 April 2025 are Knuckledragger: A low barrier proof assistant. ~ Philip Zucker. #ITP #SMT #Z3 #Python Neural theorem proving tutorial. ~ Sean Welleck. #MachineLear