Skip to content

References: AI Capabilities, Limitations, and Bias

  1. Hallucination (artificial intelligence) - Wikipedia - Offers an accessible overview of Hallucination (artificial intelligence), including definitions, methods, examples, limitations, and related concepts. This foundation helps students reason carefully about testing model claims, detecting bias and hallucination, and preserving human judgment.

  2. Algorithmic bias - Wikipedia - Offers an accessible overview of Algorithmic bias, including definitions, methods, examples, limitations, and related concepts. Its examples help students evaluate evidence for testing model claims, detecting bias and hallucination, and preserving human judgment.

  3. Automation bias - Wikipedia - Offers an accessible overview of Automation bias, including definitions, methods, examples, limitations, and related concepts. It supplies useful context for decisions about testing model claims, detecting bias and hallucination, and preserving human judgment.

  4. Weapons of Math Destruction - Cathy O'Neil - Crown - Examines how opaque models can scale bias and harm when their assumptions, feedback loops, and effects escape meaningful scrutiny. Its sustained treatment supports work on testing model claims, detecting bias and hallucination, and preserving human judgment.

  5. Race After Technology - Ruha Benjamin - Polity - Analyzes how apparently neutral technologies can reproduce inequality and why social context must accompany technical evaluation. Its cases illuminate tradeoffs involved in testing model claims, detecting bias and hallucination, and preserving human judgment.

  6. Generative AI Profile - NIST - Extends the AI Risk Management Framework with generative-AI risks and suggested actions concerning confabulation, bias, privacy, information integrity, and human oversight. Its methods give teams a starting point for testing model claims, detecting bias and hallucination, and preserving human judgment.

  7. AI Risk Management Framework - NIST - Provides the Govern, Map, Measure, and Manage functions for addressing validity, reliability, transparency, privacy, fairness, accountability, and other AI risks. Its comparisons clarify choices involved in testing model claims, detecting bias and hallucination, and preserving human judgment.

  8. OECD AI Principles - OECD - Presents human-centered principles for inclusive benefit, rights, fairness, transparency, robustness, safety, accountability, and responsible stewardship of artificial intelligence. Its framework strengthens responsible work on testing model claims, detecting bias and hallucination, and preserving human judgment.

  9. Advertising and Marketing Basics - U.S. Federal Trade Commission - Explains truth-in-advertising, evidence for objective claims, endorsements, online marketing, privacy, and other compliance responsibilities relevant to AI promotions. Its examples show how evidence informs testing model claims, detecting bias and hallucination, and preserving human judgment.

  10. Recommendation on the Ethics of Artificial Intelligence - UNESCO - Connects human rights, fairness, privacy, transparency, oversight, accountability, literacy, impact assessment, and governance across the AI system lifecycle. Its implementation advice helps teams practice testing model claims, detecting bias and hallucination, and preserving human judgment.