AI Literacy Key Terms Sheet

These terms are designed to help educators and students move from being passive users of AI to critical, active “drivers” of the technology.

Foundational Pedagogical Frameworks

• Productive Struggle: The process of engaging with challenging tasks that require sustained effort. Real cognitive development occurs when students wrestle with problems just beyond their current ability; AI can hinder this process by providing answers without the learning.
• Critical Digital Pedagogy: An approach applied in any classroom where students must evaluate the credibility, perspective, and potential bias of a source, specifically within digital and algorithmic contexts.
• The “Driver” vs. “Passenger” Model: A framework for AI use where a “driver” uses the tool as an active, critical partner to enhance their own agency and thinking. A “passenger” passively accepts algorithmic output without critical thought.

Technical Foundations

• Generative AI: A type of artificial intelligence that can create new content (text, images, video) by predicting patterns based on the data it was trained on.
• Large Language Model (LLM): A sophisticated AI model trained on vast amounts of internet text; it stitches together language patterns without a conscious understanding of meaning, morality, or factual truth.
• Stochastic Parrot: A term used by researchers to describe how LLMs “parrot” language patterns and biases from their training data through statistical probability rather than actual comprehension.
• Vector Space (High-Dimensional): The mathematical “neighborhood” where AI maps words and concepts based on how frequently they appear near each other in training data.
  • Example: AI systems often position terms associated with “Islam” closer to negative words like “crisis” or “war” because that is how they frequently appear in Western news and forum data.
• Garbage In, Garbage Out: The principle that an AI’s output is only as good as the data used to train it; if the training data is biased or flawed, the AI will reproduce and amplify those flaws.

Bias & Ethical Concepts

• Algorithmic Bias: When a computer system or artificial intelligence produces results that systematically discriminate against certain content, individuals, or groups due to the system’s design or unrepresentative training data. The skewed results reflects the human bias encoded into the mathematical models.
• Hallucinations: Confidently stated falsehoods generated by an AI. Because AI predicts the “most likely” next word, it often speaks with authority even when the information is incorrect.
  • Example: An AI providing a fabricated Bible verse or a misattributed quote from the Quran.
• WEIRD Bias: A skew in AI models toward values that are Western, Educated, Industrialized, Rich, and D
• Model Collapse: A theoretical “loop” where future AI models are trained on the synthetic, biased content produced by current AI, leading to the erasure of nuance and the amplification of stereotypes over time.
• Temporal Bias (Denial of Coevalness): The tendency of AI to depict non-Western traditions as existing only in the past, often using ancient imagery or sepia tones.
  • Example: AI-generated images of Hinduism or Sikhism featuring only ancient architecture rather than modern contexts.

Classroom Strategies (Terms & Definitions)

• Prompt Engineering (Coding for Diversity): The intentional practice of writing specific, detailed instructions to guide an AI’s output.
  • Example: Specifically prompting for “a Sikh wedding ceremony in Punjab” to force the AI to look beyond its statistical Western default.
• Red Teaming: A strategy where students act as “ethical hackers” to intentionally probe an AI tool to find its limits, errors, and cultural or religious biases.
• AI vs. the Text (Comparative Analysis): An activity where students compare an AI-generated summary of a core concept (e.g., “Karma” or “Grace”) against a primary source or authoritative text to identify “hallucinations” or missing context.
• Visual Literacy (Critiquing the Image): The process of analyzing AI-generated images to determine what is accurate versus what is a stereotype or a reduction of complex identities.
• The Socratic Tutor (Flipping the Script): Using AI as a debate or thinking partner by feeding it an argument and asking it to provide “hard questions” to challenge the student’s view.
• Source Detectives: An exercise where students demystify the “black box” of AI by tracing biased outputs back to the original internet sources, such as forum posts or skewed news, that the AI likely “read” during training.
• Classroom Charter (Co-Creating Norms): The collaborative process of writing a “Constitution” for AI use, where students own the ethical commitments, such as verification, transparency, and respect for faith traditions.

This resource was designed by Brittany Molenda, a member of the 2025-2026 Tanenbaum Network of Inclusive Educators. Please download this resource to access figures and tables.