Definitions

1. Generative AI (GenAI)

Definition: A specific subset of Artificial Intelligence designed to respond to prompts and produce output based on the prompt - such as text, images, code, audio, or video, based on patterns learned from massive datasets. Unlike traditional AI, which might categorize data or make predictions, Generative AI "generates" original-seeming outputs in response to a user's prompt.

• Generative AI vs. Other AI: Generative AI: If prompted to generate content, AI acts like a "creator" (e.g., Gemini writing an essay, DALL-E creating a painting). If prompted to create a product, it will perform that function. If prompted to ask questions, it generates questions, based on the context given in the prompt. The output is determined by the input.
• Generative AI Bots: AI bots (like Gemini Gems) do specified functions with whatever is submitted, based on its "programming" (usually a lengthy prompt with a lot of details). There are thousands of AI bots, some of which are specifically designed for educational contexts (e.g., asking clarifying questions about the text submitted or challenging assumptions in the text submitted). Most learning-focused uses of GenAI are stronger by using bots since these give more specific contexts suited to a specific learning experience and can have strict limitations for what the AI generates.
• Predictive/Analytical AI: Acts like a "filter" or "analyst" (e.g., Netflix recommending a movie, a bank detecting credit card fraud, or a weather app forecasting rain). These systems identify patterns in existing data but do not create "new" prose or art.
• Agentic AI: AI bots that are goal-driven, include detailed workflows, and complete work in the real world. Where generative AI produces results based on the prompt given within the chatbox, agentic AI accesses and produces work according to its instructions wherever that work lives as long as that location was given to the AI Agent bot and the user has access to it (e.g., an email inbox, a dataset, a google drive). 

2. AI Hallucination

Definition: A phenomenon where a GenAI model perceives patterns or objects that are nonexistent or imperceptible to human observers, resulting in confident but false outputs.

• In Practice: An AI might invent a historical event, cite a scientific paper that doesn't exist, or provide a mathematical solution that looks correct but is logically flawed. Under OWU policy, the user – not the AI tool – is responsible for these errors.

3. AI Literacy

Definition: The ability to understand, use, and critically reflect on AI technologies. Becoming AI literate allows you to use this technology effectively in your life while being able to recognize the limitations and societal impacts of AI technologies.

4. Attribution & Disclosure

Definition: The formal acknowledgment that an AI tool was used in the creation of a work.

• Attribution refers to specific citations (like MLA or APA) used for direct AI-generated text or ideas.
• Disclosure is a broader statement explaining the extent of use (e.g., "I used Gemini to help brainstorm this outline, but the drafting is my own").

5. Personally Identifiable Information (PII)

Definition: Any data that could potentially identify a specific individual.

• Examples: Social Security numbers, student ID numbers, home addresses, or private health information.
• Policy Context: To protect privacy and follow FERPA (Family Educational Rights and Privacy Act) regulations, PII should never be entered into public AI tools. This includes GPA and grades for specific students.

6. Prompt Engineering

Definition: The process of structuring a text query (the "prompt") provided to an AI model to get a specific or high-quality result.

• Note: Being prompt-conscious means writing clear, efficient prompts to reduce the need for repetitive, energy-intensive queries.

7. Algorithmic Bias

Definition: Systematic and repeatable errors in a computer system that create unfair outcomes, such as privileging one arbitrary group of users over others.

• Context: Because AI is trained on data created by humans, it often inherits human prejudices regarding race, gender, or culture.

8. AI-Prohibited vs. AI-Permitted

Definition: Terms used by faculty to define the rules of engagement for a specific course or assignment.

• AI-Prohibited: No use of Generative AI tools is permitted.
• AI-Permitted: Students are encouraged or allowed to use AI, usually with the requirement of proper disclosure.

Comparison Table: AI Tools