AI Glossary

Jump to a letter: A   B   C   D   E   F   G   H   I   L   M   N   O   P   R   S   T   U   V   Z

A

Algorithm: A set of step-by-step instructions that a computer follows to complete a task.

Algorithmic Bias: This refers to the potential for AI systems to perpetuate or amplify existing biases due to biased training data or flawed algorithms, resulting in unfair or discriminatory outcomes.

Artificial Intelligence (AI): The ability of a computer or machine to mimic human intelligence (e.g., learn, reason, solve problems).

AI Ethics: Guidelines for developing and using AI responsibly and ethically, ensuring fairness, safety, and respect for everyone.

AI Literacy: Involves the knowledge, skills, and attitudes necessary to interact with AI in a safe and effective way. This includes understanding how AI works, recognizing its potential benefits and risks, evaluating AI tools for bias and fairness, and developing the skills to use AI tools effectively and critically in teaching and learning responsibly.

AI Model: A computer program trained on a dataset to recognize patterns and perform specific tasks.

AI Safety: Measures taken to ensure AI tools are used in ways that prevent harm to individuals or society. This can encompass data privacy, bias mitigation, and responsible development.

AI Tool: AI-powered software that can automate or assist users with a variety of tasks (e.g., AI-powered writing tools, tutoring programs, or assessment tools).

B

Bias: When an AI system unfairly favors certain groups or produces prejudiced results. This can happen if the data used to train the AI is incomplete or reflects existing biases in society.

C

Chain-of-Thought (COT) Prompting: A prompting strategy that asks an AI tool to process instructions step-by-step, which can produce a better result for logical and mathematical reasoning tasks.

Context-Setting: When prompting, providing examples, information, and relevant context to an AI tool that allows the tool to produce higher quality outputs.

D

Data: Information, such as facts, numbers, and text, that is used to train AI tools.

Dataset: A large collection of organized information (like text, images, or numbers) used to train an AI model.

Data Privacy: Protecting stakeholders’ personally identifiable information (PII) when using digital tools, including AI tools.

Deep Learning: A subset of machine learning that uses neural networks to recognize complex patterns in large amounts of data. Deep learning powers GenAI capabilities, including image recognition, natural language processing.

Digital Citizenship: Responsible and ethical use of technology, encompassing online safety, privacy, critical thinking, and respectful interactions in the digital world.

E

Explainable AI: AI tools should be designed in a way that allows stakeholders to understand how they work and how decisions are made. This includes providing clear explanations of the factors considered and the logic used in the decision-making process.

F

Few-Shot Prompting: A prompting strategy that includes two or more examples of the desired input and output.

G

Generative AI (GenAI): A type of AI that can create new content, such as text, images, music, audio, or code.

H

Hallucination: Any inaccurate or misleading output from an AI tool. These can be presented as facts by the AI tool , further elevating the need to properly vet the outputs before using them more broadly.

Human-in-the-Loop: An understanding that humans should always be involved in processes involving AI, providing guidance, feedback, or making final decisions to ensure the AI is used responsibly and effectively.

I

Internal GenAI: These are restricted for use within a specific organization or domain and may require payment. One example is Google’s Gemini Enterprise, which is an add-on for Google Workspace.

L

Large Language Model (LLM): An AI model that is trained on large amounts of text to identify patterns between words, concepts, and phrases to generate effective responses to prompts.

M

Machine Learning (ML): A subset of AI focused on developing computer programs that can analyze data to make decisions or predictions.

N

Natural Language Processing (NLP): A field of AI that enables machines to parse, analyze, and generate human language.

Neural Networks: Advanced computing algorithms inspired by the human brain

O

One-Shot Prompting: A prompting strategy that includes one example of the desired input and output.

Output: The information or creative work that an AI tool produces after it is prompted, such as an answer to a question, a text summary, an image, or a piece of music.

P

Prompt: The method for interacting with an AI tool in the form of a request, question, snippet, or an example.

Public GenAI: These tools are available to anyone on the internet, such as Gemini Chat, ChatGPT, Claude, or Perplexity.

R

Reinforcement Learning: A type of ML that provides feedback to a program to improve its decisions over time.

S

Supervised Learning: A type of ML that uses labeled datasets to train a program to recognize patterns in data.

T

Transparency: Being open and honest about how AI tools are trained and how they work to all stakeholders, including being clear about when and how AI is used in the classroom.

Tokens: Units of data processed by AI models during training and output generation.

U

Unsupervised Learning: A type of ML that uses unlabeled datasets to allow a program to identify patterns in data without a specific output in mind.

V

Vendor GenAI: This type of GenAI is provided by third-party vendors whom CPS may contract to ensure data handling and usage align with District standards and policies. It is important for CPS to carefully vet any vendors to ensure alignment with District values and data privacy standards.

Z

Zero-Shot Prompting: A prompting strategy that doesn’t include any examples of the desired input and output.