Represent the world
How language, observations, and actions become useful internal representations.
Research
Our research agenda spans language, world, and multimodal models, with a shared emphasis on reasoning, evaluation, and efficient computation.
A path into the collection
How language, observations, and actions become useful internal representations.
How intelligent systems plan, use tools, and revise their decisions through feedback.
Where capability holds, where it breaks, and what it costs to put it to work.
Explore the discipline
Explore the questions that motivate each area, the methods that make progress possible, and the evidence needed to know whether a system works. These pages describe NYIC’s research interests and technical perspectives; published methods are attributed to their original authors.
Research 7 min read
Investigating how language models acquire representations, adapt to specialist work, and reason reliably with evidence.
↗Research 7 min read
Studying learned environment representations, action-conditioned dynamics, planning, and uncertainty over time.
↗Research 7 min read
Connecting language, images, audio, video, and documents through grounded representations and traceable evidence.
↗Research 7 min read
Investigating planning, tool use, memory, and coordination in systems that act over extended workflows.
↗Research 7 min read
Developing evidence about model capability, robustness, uncertainty, and behavior within accountable systems.
↗Research 7 min read
Investigating adaptation, compression, serving, and compute allocation under explicit quality and latency requirements.
↗No perspectives match these filters. Try a broader term or choose all topics.