CMU Sapling Lab Lands 10 Papers at COLM 2026, Spanning Social AI
MaartenSap · x · 2026-10-07
CMU's Sapling lab (led by Maarten Sap) announced 10 papers accepted at COLM 2026, detailed in a thread focused on social AI:
- Cognitive Chain-of-Thought (CoCoT): structured multimodal reasoning about social situations
- The Hidden Puppet Master: predicting when manipulative LLM dialogue actually changes beliefs
- Mind the Sim2Real Gap: whether simulated users can stand in for real humans in agent evaluation
- Training Proactive and Personalized LLM Agents
- NormViz: benchmark + framework grounding multimodal reasoning in global cultural norms
- SOTOPIA-TOM: evaluating privacy and information management in multi-agent interaction with Theory of Mind
- GoodPoint: learning constructive paper feedback from author responses
- Social World Models: representing and reasoning about how social interactions unfold
- Useless but Safe?: utility recovery via user intent clarification in multi-turn conversations
- CCBench: cultural competence via implicitly signaled norms in health queries
More from Research
- Polar probe paper accepted at COLM 2026: linearly decoding semantic geometry in LLMs — JeanRemiKing · 2026-10-07
- AI math proofs shift to open-source-style collaboration, square packing shows — ctjlewis · 2026-10-07
- Overmind: Open Platform That Turns Production Traces Into Fine-Tuning Data for Agents — cneuralnetwork · 2026-10-07
- Study: Top-k Logits Leak as Much Information as Tuned Lens Trajectories, Far More Accessible — sineadwilliamso · 2026-10-07
- AutoAWQ Author: Reproduce Bonsai 2-Class Ternary Model for ~$43k on One B300 Node in ~4 Weeks — airesearch12 · 2026-10-07
- COLM 2026: Robust adaptation study unifies safety pretraining and midtraining — AdtRaghunathan · 2026-10-07