Human Interface Integrity v01
What this dataset is
This dataset evaluates whether a system can handle human instructions without unsafe assumptions.
You give the model:
A human instruction
Context and system state
Known risks and safety constraints
You ask it to decide how to respond.
The core test is simple.
Does the system:
interpret intent correctly
ask when unclear
refuse when unsafe
Why this matters
Many robotics incidents involve humans.
They… See the full description on the dataset page:
https://huggingface.co/datasets/ClarusC64/human-interface-integrity-v01.
