4 Attributes of Trustworthiness

A translation model knows "95 languages," translates 100% for top 3 languages, but gives gibberish for 3 uncommon ones.

A system's ability to handle adversarial attacks relates most closely to:

AI bot for elderly care works accurately & fair, but users find its interface confusing. It needs improvement in:

An ML system designed to optimize engagement metrics without considering user well-being might fail in terms of:

An AI system designed to reduce energy consumption in buildings aligns with:

A recommendation system that explains why it suggested a particular product exemplifies:

A popular facial recognition system used for surveillance without proper privacy safeguards raises concerns about:

A conversational AI that adapts its responses based on user interaction demonstrates which attribute?

Metrics like precision, recall, and F1 score are most directly related to which attribute?

A fraud detection system disproportionately flags transactions from certain countries needs improvement in:

ML loan model is fair across demographics, but maximizes short-term profits over customer relationships. This concerns:

Autonomous vehicle GPS system is transparent that it performs perfectly in ideal weather but struggles in snow & rain.

Ethical considerations and societal impacts of an ML model fall under:

An un-interpretable healthcare AI with high accuracy needs improvement in:

A recommendation system that performs well for popular items but fails for niche interests needs improvement in:

"Does the model accomplish what it was designed to do?" primarily relates to:

An LLM generates highly accurate documentation but includes subtle errors that are difficult for non-experts.

A model identifies pneumonia in chest X-rays with 95% accuracy. This demonstrates which attribute of trustworthiness?

A facial recognition system performs well for all skin tones and ages. This primarily demonstrates: