New research on how LLMs think: vision, theory of mind and unspoken inner words
A new benchmark from the Max Planck ETH center finds that seeing is not the same as reasoning, and a theory-of-mind study compares reasoning models with non-reasoning ones. Anthropic research examined the inner, unspoken words of Claude Sonnet 4.5.
Sources
- Seeing Isn't Reasoning: New Benchmark Challenges Multimodal AI Models | Max Planck ETH Center for Learning Systemslearning-systems.org · Sep 10, 2026
- Seeing Isn't Reasoning: New Benchmark Challenges Multimodal AI Models | ELLIS Institute Tübingentue.ellis.eu · Sep 10, 2026
- To Think or Not To Think, That is The Question for Large Reasoning Models in Theory of Mind Tasks — Lacunalacuna.tiptreesystems.com · Sep 11, 2026
- Answer-Distribution Trajectories: A Stochastic-Dynamics View of LLM Reasoning | alphaXivalphaxiv.org · Sep 8, 2026
- AI ‘thinking’ words it never says: What this tells us about consciousnessindianexpress.com · Sep 7, 2026
Machine-generated summary based on the titles and excerpts of the source articles. The facts come from the linked article.