Last Week in AI Measurement in AI Policy: Opportunities and Challenges
Oct 20, 2020
Jack Clark and Raymond Perrault, esteemed co-chairs of the AI Index Project at Stanford's HAI, dive into the challenges of measuring AI in policy. They discuss the intricacies of evaluating AI technologies, the need for new metrics, and the ethical dimensions of AI research. The conversation highlights the shift towards comprehensive testing methods and addresses the societal implications of deep fakes and digital resurrections. They advocate for interdisciplinary collaboration to ensure AI development aligns with ethical standards and real-world implications.
AI Snips
Chapters
Books
Transcript
Episode notes
AI's Hidden Nature
- Characterizing AI systems is like identifying an object hidden under a thick cloth.
- Different people touching the object at different points have varying interpretations of it.
Two Axes of AI Definition
- Defining AI involves two axes: what constitutes AI and how to measure its aspects.
- Measuring AI's scientific progress, investments, and impact depends on a clear definition.
Describing AI Developments
- Describe developments within a domain accurately.
- Avoid filtering out information closely related to but not strictly captured by your AI definition.
