AI Literacy Glossary: 25 AI Terms Every Quilter Should Know (and What They Mean)
Every guild meeting now has a moment where someone says a word like “prompt” or “hallucination” and half the room nods without knowing why. This AI glossary for quilters fixes that, one plain-English term at a time.
How to use this AI glossary for quilters
I built this AI glossary for quilters the way I’d explain any new tool at the studio table: no jargon left unexplained, no assumption that you already know what any of this means. Skim the table below for the fast version, or jump to a section for the full explanation. Each term gets a plain definition first, then a quilting-world example so it actually sticks.
| Term | What It Means |
|---|---|
| Artificial intelligence (AI) | Technology that performs tasks normally requiring human judgment |
| Generative AI | AI that creates new text, images, or audio from learned patterns |
| Machine learning | How a computer improves at a task by studying data, not fixed rules |
| Large language model (LLM) | An AI trained on huge amounts of text to generate humanlike language |
| Chatbot | A program you talk to, built to answer questions or hold a conversation |
| Prompt | The instruction or question you type to tell an AI what you want |
| Prompt engineering | Wording your instructions carefully to get a better AI result |
| Context | The information an AI has available from earlier in your conversation |
| Token | A small chunk of text an AI reads or generates, roughly a word |
| Output | Whatever the AI produces in response to your prompt |
| Training data | The material an AI studied to learn how to generate responses |
| Algorithm | A set of steps a computer follows to solve a problem |
| Neural network | An AI system loosely modeled on how brain neurons connect |
| Fine-tuning | Training an existing AI model further on specific, narrower data |
| Bias | A skew in AI output caused by gaps or imbalances in its training data |
| Hallucination | When an AI states something false with complete confidence |
| AI slop | Low-quality, mass-produced AI content made without human judgment |
| Deepfake | A manipulated video, image, or audio that makes someone appear to say or do something they didn’t |
| Scraping | Automated pulling of text, images, or data from websites, often for training AI |
| AI detection | Software that tries to identify whether content was made by an AI |
| Copyright and AI | The legal and ethical questions around who owns AI-generated content |
| Opt-out | A setting or request that keeps a creator’s work out of AI training data |
| Open source | AI models whose code (and sometimes training data) is public |
| AI agent | An AI system that takes multi-step action on your behalf, not just answers |
| AI ethics | The field concerned with responsible, fair use of AI |
Jump to a section:
The basics
Artificial intelligence (AI)
Artificial intelligence, or AI, is technology that performs tasks which normally require human judgment, like recognizing patterns, making predictions, or generating text and images. It’s a broad umbrella term, not one single tool. Think of it the way you’d think of “power tools.” A rotary cutter and a longarm machine are both power tools, but they do very different jobs and carry very different risks. AI is a useful tool, and like a rotary cutter, it’s most likely to cause trouble the moment you stop paying attention to what it’s actually doing.
Generative AI
Generative AI is a category of artificial intelligence that creates new content, such as text, images, audio, or video, based on patterns it learned from existing examples. It’s the technology behind an AI tool that generates a quilt block design from a text description, or writes a first draft of your guild newsletter. The word “generative” is the key part. It’s not looking things up, it’s producing something new based on what it has seen before. That distinction matters, because it means the output can be original and also completely wrong at the same time.
Machine learning
Machine learning is the process by which a computer system improves at a task by analyzing data instead of following a fixed set of programmed rules. So instead of a programmer writing “if fabric is red, pair with green,” the system studies thousands of fabric pairings and works out its own patterns. This is the foundation underneath most modern AI tools, including the ones covered in our roundup of the best AI tools for quilters. It learns by example, the same way you learned color pairing by looking at a hundred quilts before you trusted your own eye.
Large language model (LLM)
A large language model, or LLM, is an AI system trained on enormous amounts of text so it can predict and generate humanlike language. ChatGPT, Claude, and Gemini are all built on an LLM. When you ask one of these tools to draft a pattern description or explain a color theory concept, an LLM is doing the writing. It doesn’t “know” things the way you do. It’s predicting the most likely next word based on everything it read during training, which is why it can sound confident and still be wrong.
Chatbot
A chatbot is a program you interact with by typing or talking, built to answer questions or carry on a conversation. Older chatbots followed rigid scripts and broke the second you asked something unexpected. The ones built on LLMs today can hold a real back-and-forth, closer to texting a knowledgeable friend than filling out a form. If you’ve ever asked an AI tool to help you troubleshoot a wonky bobbin tension mid-project, you were using a chatbot.
AI 101 Live
Talking to AI
Prompt
A prompt is the instruction or question you type into an AI tool to tell it what you want. “Write a caption for my finished quilt” is a prompt. So is “Explain fabric value in one paragraph.” The prompt is the only steering wheel you have, so vague prompts tend to get vague, generic answers back. The more specific you are about what you want and why, the closer the output lands to something usable.
Prompt engineering
Prompt engineering is the practice of wording your instructions to an AI system so it gives you a more accurate or useful result. It sounds more technical than it is. It’s closer to adjusting a pattern for your own measurements than writing code. If your first prompt gets you a bland answer, adding detail, examples, or a specific tone is prompt engineering in action, even if you never use that phrase out loud.
Context
In AI, context is the information a system has available when it generates a response, including anything you’ve told it earlier in the conversation. If you tell a chatbot you’re working on a scrap quilt in blues and yellows, and then ask “what should I add,” it uses that earlier detail as context for its answer. Once you close the conversation or start a new one, that context is usually gone. That’s why picking up a project days later often means re-explaining what you’re making.
Token
A token is the small chunk of text, roughly a word or part of a word, that an AI system reads and generates one piece at a time. You’ll mostly run into this word when a tool tells you you’ve hit a “token limit,” meaning you’ve fed it more text than it can process in one go. Think of it as yardage. Every AI tool has a bolt of a certain length, and long documents or long conversations eat through it fast.
Output
Output is whatever the AI produces in response to your prompt, whether that’s text, an image, or a block of code. Output is a draft, not a finished piece, the same way a muslin test block isn’t the quilt. It needs your eye on it before it goes anywhere near a customer, a class, or a pattern for sale.
How it learns
Training data
Training data is the collection of text, images, or other material an AI system studied in order to learn patterns and generate its own responses. An AI tool trained mostly on modern, minimalist quilt photography is going to be far better at generating that style than a traditional appliqué design, simply because it saw more of one than the other. The output is only ever as broad as what it learned from.
Algorithm
An algorithm is a set of steps a computer follows to solve a problem or complete a task. It’s the recipe, not the finished dish. A pattern’s cutting instructions are an algorithm for a human to follow. An AI algorithm is the same idea, just written for a machine, and running far faster and on far more data than a person could manage by hand.
Neural network
A neural network is a type of AI system loosely modeled on how neurons in the brain connect and pass along information. Picture piecing a quilt in layers: the first pass gets the blocks roughly right, the second pass refines the seams, the third pass squares everything up. A neural network processes information in layers the same way, each one refining what the last one produced, until the final output comes out the other end.
Fine-tuning
Fine-tuning is the process of taking an existing AI model and training it further on a smaller, specific set of data to make it better at a particular task. It’s the difference between a general pattern and one you’ve adjusted to fit your own measurements exactly. A general AI model can write in a decent, generic voice. Fine-tuning is what makes it sound like you instead of like everyone else’s chatbot.
Bias
Bias, in an AI context, is a skew in a system’s output that comes from patterns, gaps, or imbalances in the data it was trained on. If an AI tool was trained mostly on images of one style or tradition of quilting, ask it to generate “a quilt” and it will default to that style every time, even though quilting traditions are far wider than that. Bias isn’t usually intentional. It’s a mirror of whatever the training data did and didn’t include.
When it goes sideways
Hallucination
A hallucination is when an AI states something false with complete confidence. It’s like a guild member who answers every question whether she knows the answer or not, and says it with the same certainty either way. Ask an AI tool for the history of a specific quilt block and it may invent a plausible-sounding origin story that never happened. IBM has a solid technical breakdown of why this happens if you want to go deeper, but the practical takeaway for quilters is simpler: verify anything factual before you repeat it. To read more about why this happens, read my blog post on Why AI Hallucinates.
AI slop
AI slop is a term for low-quality, mass-produced AI content made quickly and without much human judgment or editing. You’ve probably scrolled past it on Pinterest or Etsy: quilt “patterns” and images that look almost right until you notice the seams don’t logically connect or the block repeats in a way no human would design. AI slop isn’t a failure of the technology. It’s a failure of someone skipping the editing step.
Deepfake
A deepfake is a manipulated video, image, or audio recording that uses AI to make someone appear to say or do something they never actually did. It’s not a quilting-specific term, but the quilting world isn’t immune to it. Scammers have used AI-generated video of well-known makers to sell fake courses or counterfeit patterns that the real person never made or endorsed. If a video of a maker you follow seems out of character, that’s worth a second look before you trust it or buy anything from it.
Scraping
Scraping is the automated process of pulling large amounts of text, images, or data from websites, often used to build training data for AI systems. This is the practice at the center of a lot of the frustration you’ll hear from pattern designers and photographers: their work gets pulled into a training dataset without permission or payment. Understanding scraping is the first step to understanding why so many creatives are wary of AI, and that wariness is reasonable.
AI detection
AI detection is software that attempts to identify whether a piece of text, image, or other content was created by an AI system rather than a person. These tools exist, but they’re far from reliable, and they regularly flag human writing as AI-made and miss the reverse. If you’re being told your own writing “reads as AI,” take the result with real skepticism rather than treating it as fact.
Â
The bigger conversation
Copyright and AI
Copyright and AI refers to the ongoing legal and ethical questions about who owns content that AI systems generate, and whether using copyrighted material to train those systems without permission is fair. This one isn’t settled, legally or ethically, and I’m not going to pretend it is. What I can tell you is that if you’re using AI to help design or write something you plan to sell, you should understand what you’re allowed to claim ownership of before you do. I lay out where I stand and why in my AI ethics guide.
Opt-out
An opt-out, in AI terms, is a setting or request that lets a creator ask that their work not be used to train AI systems. Some pattern marketplaces and stock photo sites now offer this as a setting you can turn on. It doesn’t guarantee your work is excluded everywhere, since enforcement varies widely, but it’s a real option worth knowing exists if you sell your own designs or photography online.
Open source
Open source AI refers to models whose underlying code, and sometimes training data, is made publicly available for anyone to inspect, use, or build on. It’s a bit like the difference between a pattern you release for free with full permission to modify and share, versus one you sell with usage restrictions attached. Neither approach is automatically better. They just come with different tradeoffs around cost, transparency, and control.
AI agent
An AI agent is a system that can take multi-step action on your behalf, like researching a topic, drafting a response, and sending it, rather than just answering a single question. This is a newer and more capable category than a basic chatbot, and it’s changing what’s realistic for a small creative business to automate. If you want to see what that looks like in practice for quilting-adjacent work, quilter-focused AI tools are a good place to start.
AI ethics
AI ethics is the field concerned with the responsible use of AI, including issues like fairness, environmental impact, labor, and creative ownership. It’s the umbrella all of the terms in this last section fall under. I don’t think taking AI seriously as a tool and thinking critically about its costs are opposite positions. Around here, they’re the same position.
One more thing
Twenty-five terms is a lot to hold onto after one read-through, and you don’t need to memorize any of it. What matters is that the next time someone in your guild says “hallucination” or “training data,” you’ll know exactly what they mean and won’t have to nod along. If you’d rather work through all of it out loud, with real examples and time for your own questions, that’s what AI 101 Live is built for.