DeepMind paper: Veo 3 shows emergent zero-shot reasoning, video models may become vision foundation models
RexDouglass · x · 2026-09-17
A Google DeepMind paper, "Video models are zero-shot learners and reasoners," shows Veo 3 zero-shot solves many untrained tasks: segmentation, edge detection, image editing, physical property understanding, tool-use simulation, and early visual reasoning like maze and symmetry solving — suggesting video models are on a path to unified vision foundation models, like LLMs for language.
- Co-author Shixiang Shane Gu ("LLMs are Zero-Shot Reasoners") responds to academic despair over robots being zero-shot benchmarked: digital/symbolic AGI must precede and will accelerate physical AI
- He argues LLMs remain the indispensable core of physical AGI, a hypothesis he tested over the past year across Gemini, Omni, and generative media
- Lerrel Pinto notes Astra, Fable, and Muse are zero-shotting robotics & world model benchmarks — what step jumps in progress look like
More from AGI Musings
- Palmer Luckey asks: can you use datacenter RAM for things prohibited by the operator's terms of service? — avaitopiper · 2026-11-02
- On Terence Tao's blog, Frank Calegari calls on mathematicians to adapt after OpenAI's 700-paper release — Hesamation · 2026-11-02
- Polsia's Ben argues AI can autonomously run any human economic activity, including 80% of a company — LinusEkenstam · 2026-11-02
- AI Assistants Remember You, But Nothing Helps You Find What They Told You — Ok_Negotiation_2587 · 2026-11-02
- OpenAI and Anthropic claim AI-discovered proofs disprove the Exponential Time Hypothesis — emmanuelvivier · 2026-11-02
- Researcher: using a $20 AI subscription to fix my phone hints AI may suppress GDP — stanislavfort · 2026-11-02