AI News 2025-12-11

General

• Sentiment:
  • 2025 Edelman Trust Barometer
  • How the U.S. Public and AI Experts View Artificial Intelligence
  • Andrew Ng (text)

Research Insights

• Large Causal Models from Large Language Models. The causal model is generated by querying the LLM, yielding surprisingly robust causal knowledge.
• The Missing Layer of AGI: From Pattern Alchemy to Coordination Physics. Scaffolding of LLM (including system-2 inspired coordination layer) demonstrates that many of the perceived problems (hallucination, limited capabilities, etc.) are not intrinsic limits of LLM cognition, but more resultant from deployment.
• AI & Human Co-Improvement for Safer Co-Superintelligence.
• Google: Titans + MIRAS: Helping AI have long-term memory.
  • Titans: Learning to Memorize at Test Time.
  • It’s All Connected: A Journey Through Test-Time Memorization, Attentional Bias, Retention, and Online Optimization.

LLM

• OpenAI rolls out GPT 5.2.

Safety

• Google DeepMind: Difficulties with Evaluating a Deception Detector for AIs.

AI Agents

• Agentic AI Foundation, founded by OpenAI and others. Will provide open standards for AI agent interoperation.

Video

• Runway ML announces several updates:
  • Gen 4.5 video model, improved model with native audio.
  • Edit audio of existing videos, and multi-shot editing.
  • GWM-1 (General World Model), allowing predictions of future states (e.g. for robotics). Three variants: GWM Worlds for explorable environments, GWM Avatars for conversational characters, and GWM Robotics for robotic manipulation.