Human-Computer Interaction Institute (CMU HCII)
Language Technology Institute (CMU LTI)
Pitfalls in status quo human-AI interactions.
Principles and tools for enhanced NLP model analysis.
Controllable generators for model analysis and improvement.
Transparent & controllable human-AI collaborations via multi-step problem-solving.
Behavioral testing for NLP models covering broad model capabilities.
Structural analysis for unstructured text datasets.
Uncertainty-aware data labeling and visual refinement.
De-cluttering statistical graphs.