Category: AI – This Quote

Shannon Vallor warns that narratives diminishing human moral and political agency may be more dangerous than AI systems themselves. for today’s AI systems.

Batya Friedman’s value sensitive design integrates ethical reflection, stakeholder analysis, and empirical study throughout technology development. for today’s AI systems.

Helen Nissenbaum’s contextual integrity explains privacy through appropriate information flows rather than secrecy or data control alone. for today’s AI systems.

Emily Denton and colleagues proposed datasheets so dataset creators can document origins, composition, uses, limits, and maintenance. for today’s AI systems.

Abeba Birhane’s relational ethics centers people, power, history, and social context in the evaluation of algorithmic systems. for today’s AI systems.

Emily Bender and colleagues warned against confusing fluent text generation with meaning, understanding, or responsible language technology. for today’s AI systems.

Inioluwa Deborah Raji and colleagues designed an internal audit framework that follows AI systems throughout the development lifecycle. for today’s AI systems.

Moritz Hardt and colleagues formalized equal opportunity as one way to compare machine-learning errors across demographic groups. for today’s AI systems.

Ziad Obermeyer’s healthcare study showed how a seemingly reasonable proxy can create profound racial bias in an algorithm. for today’s AI systems.

Solon Barocas and Andrew Selbst showed how seemingly neutral data systems can reproduce longstanding exclusion and inequality. for today’s AI systems.

Cynthia Rudin argues that high-stakes decisions often need transparent models, not post-hoc stories about opaque systems. for today’s AI systems.

Finale Doshi-Velez and Been Kim called for rigorous definitions and evidence when evaluating interpretable machine learning. for today’s AI systems.