Learn: Core Concepts, Ethics, & Uses

Terminology

Artificial Intelligence is widely used across many fields, so its terminology can be confusing. To help understand AI better, here are ten key terms that cover everything from basic concepts. See 10 Essential Terms To Understand Artificial Intelligence, by Gupta, G. (2024, February 11), ELearning Industry. 

How GenAI Models Work

• Generative artificial intelligence tools are NOT search engines, nor are they sentient minds. They are sophisticated pattern-recognition and statistical next-word prediction engines.

• When you prompt, a Large Language Model (LLM) calculates the probability of the next most likely word or phrase in a sequence based on its massive training datasets. Because their outputs are probabilistic rather than deterministic, entering the same prompt twice will often yield completely different results. 

• This random element is why interacting with an LLM feels less like working with a database and more like working with a wizard, because reproducibility is low.

Capabilities & Limitations

• AI CAN summarize long texts, brainstorm project ideas, structure outlines, translate languages, and debug basic code.

• AI CANNOT conceptualize the world, do basic arithmetic, and reliably tell you where it got its information.

• Because AI models are optimized for linguistic fluency rather than factual truth, they can fabricate (hallucinate) citations, quotes, and scientific data with absolute confidence. ALWAYS assume an AI output is a draft that requires human verification.

Why AI Detection Software is Unreliable 

Plagiarism detectors (such as Turnitin's AI detector or GPTZero) are unreliable because they have high error rates and disproportionately flag the writing of neurodivergent students and non-native English speakers.

• High Error Rates: Automated detectors are easily bypassed and highly inaccurate, so they cannot provide definitive proof of misconduct.

• Systemic Biases: High error rates occur because these non-native English speakers often use highly structured, predictable linguistic patterns that detectors are trained to identify as machine-generated.

• The Alternative? Shift your grading rubrics to evaluate process over product. Require students to submit their research journals, annotated outlines, draft iterations, and self-reflection logs as a percentage of the final grade. Consider asking students to share their draft Google Doc with you so you can use Version History (under File) to understand their writing process.

Academic Integrity

When teachers attempt to deter cheating with severe penalties, such as failing the course, the burden of proof is on the accuser. AI use is challenging to detect because detection tools are not 100% accurate, particularly when grammar tools are used. However, that doesn't mean nothing can be done to prevent cheating or ensure integrity. 

You are better off Teaching Students About Plagiarism (2:44). See CSS's AI Ethical Use Statement and AI syllabus statement.

To maintain academic integrity and protect data privacy, we recommend the following:

1. Use the College's provided tools instead of requiring students to create third-party accounts.

2. Never type names, grades, ID numbers, or proprietary research data into public AI prompts.

3. If an assignment requires a tool that has not been provided by the institution, provide a non-AI equivalent assignment for students who have ethical or privacy objections.

4. Treat AI tools purely as assistants. Never let an AI make automated grading decisions, and require students to verify every machine output.

Ethical Use of AI at CSS (Established Fall 2025):

• Human-Centered and Meaningful: AI tools are used only to facilitate our work, not displace the learner, colleague, teacher, or researcher. Human users remain accountable for judgment, context, critical thinking, and meaning-making.

• Transparent: AI use should be disclosed and/or cited when appropriate, with enough detail to understand what the tool did and what the human decided.

• Responsive to Environmental and Social Impacts: We prioritize smaller, efficient models or non-AI methods that achieve the same outcomes with less resource use. We use AI judiciously, representing our commitment to environmental stewardship.

• Aware and Prepared: AI outputs should not automatically be read as truths, because AI predicts likely next words from imperfect, biased training data.

• Equitable and Accessible: All students, faculty, and staff deserve institutionally vetted, no-cost access to AI tools. AI may serve as an approved accessibility or learning support tool.

Select AI Chatbots

A chatbot's knowledge base is huge with many different sources of information on a topic. It will give you the most likely answer to the question for the average person from a massive database, which is why specificity is important. Lastly, it wants to make you happy, so it will make up facts rather than disappoint you. 

These tools can summarize results from web search and link to the sources

• Gemini (Google) - Access from the waffle in your CSS email or Drive.

• ChatGPT (OpenAI)

• Claude.ai (Anthropic)

• Copilot (Microsoft)

• Perplexity AI

The AI Guide from the AI Pedagogy Project at Harvard University allows you to play with ChatGPT and Claude without logging in or creating an account.

Select AI Research Tools

Many generative AI systems now have a feature called “Deep Research” — sometimes called “Deep Search” or “Deep Think.” They are designed to generate long reports or literature reviews instead of short answers. These systems search ONLY open websites, not content that lives behind paywalls. 

• They search, analyze, and synthesize information from hundreds of public, online sources.

• They work for several minutes - you can go away and come back later.

• Their output is a structured report with citations to sources.

• They use an iterative approach - refining as they go.

• They are cross-disciplinary - combining information from various domains.

• They are powered by LLMs, often by reasoning models.

Deep Research is available in free and paid versions of Gemini, ChatGPT, DeepSeek, Grok, Kimi, and Perplexity.  In Claude it’s only found in the paid version. We recommend Google tools: Gemini and NotebooksLM.

• Gemini, using its Deep Research tool. Consider this question: "What are the most significant peer-reviewed findings from the last year regarding the correlation between diet and mental illness? Provide a structured summary with citations.” (Deep Research questions can take 10 minutes.) See the result. See revision. NOTE: The references may not be good or may lead to fake sources.

• NotebookLM (Google) - Access from the waffle in your CSS email or Drive. This AI-powered research hub is designed to eliminate 'hallucinations' by anchoring its intelligence strictly to the materials you add to a notebook, including PDFs, websites, text, documents, images, and public YouTube videos (not unlisted). The AI helps you analyze and synthesize information and generate visual and audio summaries without the risk of external misinformation. Learn more here.

Other tools are shared under Supporting Student Researchers.

Image Generators

• IMAGES: DALL-E (OpenAI) and Gemini (Google) with its Create Image tool. 

• SLIDES: Canva.com, Beautiful, Plus AI and here is a review of seven more.

• VIDEO: Sora (from OpenAI), Flux, and lots of APIs (that use ChatGPT but with layers of tools): here is a review of 14 of them.

• AVATAR CREATOR: HeyGen 

Prompting

• A prompt format: "As a [Role], provide a [output] in [format] that will [purpose]".  Then, refine the prompt by telling it what it did well and what you want to adjust. (from the Society for Technical Communication annual conference).

• Jose Bowen's 100+ sample prompts for students, teachers, and researchers. Better prompts contain these elements: TASK: What exactly do you want AI to do?, FORMAT: What is the specific output?, VOICE: What style of language is desired., and CONTEXT: What further context or examples can you provide?

• Writing and Refining Prompts explores strategies for supplying an AI tool with an initial prompt, adapting AI output for professional use through re-prompting and human editing, how tone and language affect AI output, and ways to prompt AI tools to write in your voice.

See Teach: Prompt Engineering for specific formats. 

Overview

GenAI tools such as ChatGPT, Claude, and Copilot can be thought of as helpful assistants or collaborative partners in teaching and learning. These AI tools can aid instructors in tasks related to instructional development and preparation, allowing them to invest more in enhancing pedagogical approaches that support student learning. 

Used well, AI can reduce busywork and widen support for learners. Used poorly, it can weaken rigour, fairness, and privacy or even make you feel like you're wasting energy on tools that never really benefit you or students.

Administrative Tasks

• Administrative Tasks (Heaps, 2024). This chapter explores how AI tools can: write first drafts of emails, agendas, templates, and other documents; and help staff manage large amounts of information to summarize, identify trends, or target a search.

Course Development

• Course Content Development (Heaps, 2024). This chapter explores: how AI tools can assist with the course planning process; and ways that AI tools can generate multimedia content for course LMS or other materials.

• Lesson Planning (Heaps, 2024). This chapter explores: how to prompt an AI tool to produce a draft lesson plan, and how to adapt it for classroom use; how to use AI tools to brainstorm classroom activities.

• AI for Allegories and Metaphors (2:47) - See an AI prompt in action and learn how to save time while crafting stories that resonate with your students.

• Using AI to Create Case Studies (3:08) - describes how to leverage GenAI to create interactive case studies.

Multimedia

• How to Use AI for Media Creation (2:26). Provides an overview on how to create AI-generated media, including images, audio, and video for multimedia projects.

• AI Image Generation: Techniques and Tools (2:49). Outlines tools and approaches you can use to create custom visuals that engage students and enhance learning.

• AI for Image Generation in Course Design (2:37). Explains how to use AI for image generation and how AI-generated images offer new opportunities to engage students in critical thinking and creative projects across disciplines.

• DALL-E: Tips for Writing Effective Prompts. D2L Brightspace Community (2024, June 20). 

Assessment

• See Assess: Redesign, Evaluate, & Reward

Grading and Feedback

• AI as a Grading Assistant – Simplifying the Process (2:10) - Shares how GenAI can be used to assist with grading writing assignments and best practices to ensure AI tools are used responsibly.

• Feedback and Grading (Heaps, 2024). This chapter explores: how to use AI tools as a source of formative feedback; how AI tools can assist with summative feedback; and fun ways to use AI to make feedback more engaging.