Building Trustworthy AI

Scientific Question:

How should AI systems communicate uncertainty to humans?

This program builds and audits research prototypes for uncertainty communication. Measurement validity and effects on human decisions remain open empirical questions.

Projects

P Engineering prototype

CheckMyCoach

Four-module routing, candidate-generation, and surface-check pipeline

K Provenance recovery

Knowledge Compiler

Structured evidence infrastructure with a provenance-recovered subset

I Frozen v1.0.0

InteractionKit

A Methodological Infrastructure for HAI Trust Calibration

I Development

PI Discovery Engine

From Research Topics to Investigator Cards