Individual information is paramount to the case view
what metaphors can we use to design intuitive, Explicit and mutable context in LLMs?
what metaphors can we use to design intuitive, Explicit and mutable context in LLMs?
HUMAN-AI INTERACTION
CONVERSATIONAL UI
Context engineering
DIRECT MANIPULATION
INTERACTION DESIGN
Context in LLMs is hidden, ephemeral, and out of the user's hands. As conversations grow, models degrade, a phenomenon known as Context Rot. While existing research treats this as a technical bottleneck, we identify a design gap, Interactional Context Engineering, using UI to make context explicit and manipulable, so users can understand what the AI knows and actively shape it, rather than drifting into passive "instruct, serve, repeat" loops.
Context in LLMs is hidden, ephemeral, and out of the user's hands. As conversations grow, models degrade, a phenomenon known as Context Rot. While existing research treats this as a technical bottleneck, we identify a design gap, Interactional Context Engineering, using UI to make context explicit and manipulable, so users can understand what the AI knows and actively shape it, rather than drifting into passive "instruct, serve, repeat" loops.
Context in LLMs is hidden, ephemeral, and out of the user's hands. As conversations grow, models degrade, a phenomenon known as Context Rot. While existing research treats this as a technical bottleneck, we identify a design gap, Interactional Context Engineering, using UI to make context explicit and manipulable, so users can understand what the AI knows and actively shape it, rather than drifting into passive "instruct, serve, repeat" loops.
Context engineering
HUMAN-AI INTERACTION
CONVERSATIONAL UI
Context engineering
DIRECT MANIPULATION
INTERACTION DESIGN
HUMAN-AI INTERACTION
CONVERSATIONAL UI
DIRECT MANIPULATION
INTERACTION DESIGN
The project is still in progress. The deliverables and the research was presented at NYU's Research Excellence Exhibit 2026
The project is still in progress. The deliverables and the research was presented at NYU's Research Excellence Exhibit 2026
Siddharth Srivastava
Isaac Ting
Reviewed by
Vedant Daas Swain
Kevin Chow
A glimpse of Bloom DELIVERABLES
Presented our research work at NYU's 2026 Tandon Research Excellence Exhibit
My other projects
my projects
my OTHER projects
