{"id":679,"date":"2026-10-20T15:35:29","date_gmt":"2026-10-20T20:35:29","guid":{"rendered":"https:\/\/sts.doit.wisc.edu\/training-materials\/?post_type=manual&#038;p=679"},"modified":"2026-10-20T20:28:46","modified_gmt":"2026-10-21T01:28:46","slug":"ai-first-steps-into-ai","status":"publish","type":"manual","link":"https:\/\/sts.doit.wisc.edu\/training-materials\/manual\/ai-first-steps-into-ai\/","title":{"rendered":"AI: First Steps into AI"},"content":{"rendered":"\n<figure class=\"wp-block-image aligncenter size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"300\" height=\"152\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2025\/12\/image-18.png\" alt=\"Software Training for Students logo: a colorful brain made of circuit lines\" class=\"wp-image-600\"\/><\/figure>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Introduction<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">This manual is a reference for artificial intelligence (AI): the key concepts, tools, and techniques a beginner needs to understand and use AI well. You can read it start to finish, or jump to any section when you need it.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"604\" height=\"760\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_ai_brain.png\" alt=\"Illustration of a glowing purple AI processor with energy radiating outward\" class=\"wp-image-1867\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_ai_brain.png 604w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_ai_brain-238x300.png 238w\" sizes=\"auto, (max-width: 604px) 100vw, 604px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Each concept, tool, and technique in this manual is explained the same way: what it is, how it works or how to use it, why it\u2019s worth learning, what it\u2019s similar to, and an example.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What Working with AI Feels Like<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Using an AI assistant feels like working with a very fast, very well-read helper who has never met you, is always confident, and is sometimes wrong. It can draft, explain, summarize, and brainstorm in seconds, but it only knows what you tell it and what it learned in training.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Experienced AI users pay attention to a few things every time:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Context:<\/strong> the more relevant detail they give, the better the answer<\/li>\n\n\n\n<li><strong>Iteration:<\/strong> the first answer is a draft, so they follow up and refine<\/li>\n\n\n\n<li><strong>Verification:<\/strong> they check facts, sources, numbers, and code before relying on them<\/li>\n\n\n\n<li><strong>Privacy:<\/strong> they think about what data they\u2019re pasting and where it goes<\/li>\n<\/ul>\n\n\n\n<div class=\"wp-block-group sts-callout-goodtoknow has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">No coding or math background is needed for this manual. If you can use a web browser, you can follow everything here.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Software Interface: Microsoft 365 Copilot Chat<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Most of the hands-on work with AI uses <strong>Microsoft 365 Copilot Chat<\/strong>, an AI chat assistant that UW-Madison provides free to all students, faculty, and staff. Other chat assistants like ChatGPT, Claude, and Gemini are laid out almost the same way, so what you learn here carries over.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Getting In<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li>Go to <strong>m365.cloud.microsoft<\/strong> in any web browser.<\/li>\n\n\n\n<li>Sign in with your UW-Madison email address and NetID password.<\/li>\n\n\n\n<li>Check for the <strong>green shield<\/strong> icon. It means enterprise data protection is on for your chats.<\/li>\n<\/ol>\n\n\n\n<div class=\"wp-block-group sts-callout-important has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">Copilot Chat only gets the university\u2019s data protection when you\u2019re signed in with your UW-Madison account. If you don\u2019t see the green shield, sign out and sign back in with your NetID.<\/p>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading\">General Layout<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Message box:<\/strong> where you type your prompt. Press Enter to send.<\/li>\n\n\n\n<li><strong>Attach button (+):<\/strong> next to the message box. Use it to upload files like PDFs, documents, or images for Copilot to read.<\/li>\n\n\n\n<li><strong>New chat:<\/strong> starts a fresh conversation with an empty context.<\/li>\n\n\n\n<li><strong>Chat history:<\/strong> your past conversations, so you can pick up where you left off.<\/li>\n\n\n\n<li><strong>Response area:<\/strong> where answers appear, often with buttons to copy text or code and links to any web sources used.<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Most Important Features for Beginners<\/h3>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>Feature<\/th><th>What it does<\/th><th>When to use it<\/th><\/tr><\/thead><tbody><tr><td>Chat<\/td><td>Answers questions, writes, edits, and explains<\/td><td>Almost everything<\/td><\/tr><tr><td>File uploads<\/td><td>Reads documents and images you attach<\/td><td>Summarizing a reading, getting feedback on a draft<\/td><\/tr><tr><td>Web search<\/td><td>Looks up current information and cites sources<\/td><td>Recent events, facts you need to check<\/td><\/tr><tr><td>Image generation<\/td><td>Creates images from a description<\/td><td>Quick visuals and concept sketches<\/td><\/tr><tr><td>Pages<\/td><td>Turns a chat answer into an editable page you can keep working on<\/td><td>Drafts you want to refine or share<\/td><\/tr><tr><td>Agents<\/td><td>Specialized helpers for particular tasks<\/td><td>Once you\u2019re comfortable with basic chat<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Features change often. For the current list and data rules, see <a href=\"https:\/\/it.wisc.edu\/services\/m365-copilot-chat\/\">UW-Madison IT\u2019s Copilot Chat page<\/a>.<\/p>\n\n\n\n<div class=\"wp-block-group sts-callout-warning has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">Never enter sensitive or restricted university data into Copilot Chat, even when the green shield is showing.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Artificial Intelligence<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Artificial intelligence (AI)<\/strong> is the field of building computer systems that do tasks we usually connect with human thinking: recognizing patterns, understanding language, making predictions, and creating content.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"780\" height=\"745\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_human_robot_hands.png\" alt=\"Illustration of a human hand reaching toward a robotic hand\" class=\"wp-image-1861\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_human_robot_hands.png 780w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_human_robot_hands-300x287.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_human_robot_hands-768x734.png 768w\" sizes=\"auto, (max-width: 780px) 100vw, 780px\" \/><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">AI is an umbrella term with several layers inside it. Each one sits inside the one before it:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"521\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_nested_circles-1024x521.png\" alt=\"Nested circles: Generative AI sits inside Deep Learning, which sits inside Machine Learning, which sits inside Artificial Intelligence.\" class=\"wp-image-1873\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_nested_circles-1024x521.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_nested_circles-300x153.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_nested_circles-768x391.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_nested_circles.png 1485w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Each term sits inside the one before it<\/figcaption><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Artificial intelligence:<\/strong> any technique that makes computers act in smart ways, including hand-written rules.<\/li>\n\n\n\n<li><strong>Machine learning:<\/strong> AI that learns patterns from data instead of following fixed rules.<\/li>\n\n\n\n<li><strong>Deep learning:<\/strong> machine learning with many-layered neural networks.<\/li>\n\n\n\n<li><strong>Generative AI:<\/strong> deep learning models that create new text, images, audio, video, and code.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">The field also has subfields that real products combine: computer vision (images and video), natural language processing (human language), robotics (physical machines), reinforcement learning (trial and reward), and search and planning (finding the best sequence of steps, like route planning in a maps app).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The ideas are over 70 years old. What changed recently is the amount of data, computing power, and a few key breakthroughs.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"410\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_history_timeline-1024x410.png\" alt=\"Timeline of AI: 1950 Alan Turing asks \u201cCan machines think?\u201d; 1956 the term artificial intelligence is coined at Dartmouth; 1997 Deep Blue beats chess champion Garry Kasparov; 2012 deep learning wins big at image recognition; 2017 Google introduces the Transformer; 2022 ChatGPT launches; 2024 on, reasoning models and AI agents.\" class=\"wp-image-1874\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_history_timeline-1024x410.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_history_timeline-300x120.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_history_timeline-768x308.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_history_timeline.png 1518w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">A short history of AI<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">These terms get used interchangeably in the news and in job posts. Knowing which is which helps you understand what a tool can actually do, and why AI handles information at a scale people can\u2019t: processing millions of data points, automating repetitive work, and personalizing experiences.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Traditional software also automates tasks, but a programmer writes every rule. With machine learning, the computer learns the rules from examples. A 1997 chess program is AI, but it isn\u2019t machine learning at all.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">You probably used AI several times today: search engines rank results, streaming apps recommend shows, spam filters sort your inbox, maps predict traffic, your phone unlocks with your face, and chat assistants answer questions.<\/p>\n\n\n\n<div class=\"wp-block-group sts-callout-goodtoknow has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">AI doesn\u2019t think the way you do. It learns patterns from lots of examples and applies them to new situations. When a photo app recognizes your friend\u2019s face, it hasn\u2019t \u201cmet\u201d your friend. It has learned which pixel patterns tend to match that face.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Machine Learning and Training Data<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Machine learning<\/strong> builds systems that learn from data instead of following hand-written rules. Instead of writing hundreds of rules to catch spam, you show a model thousands of labeled emails and it learns the patterns itself.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Every machine learning project starts with data. Two terms matter most:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Features<\/strong> are the inputs: the clues the model looks at.<\/li>\n\n\n\n<li><strong>Labels<\/strong> are the answers it\u2019s trying to predict.<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>Says \u201cfree prize\u201d?<\/th><th>Sender in contacts?<\/th><th>Number of links<\/th><th>Label<\/th><\/tr><\/thead><tbody><tr><td>Yes<\/td><td>No<\/td><td>7<\/td><td>Spam<\/td><\/tr><tr><td>No<\/td><td>Yes<\/td><td>1<\/td><td>Not spam<\/td><\/tr><tr><td>No<\/td><td>No<\/td><td>0<\/td><td>Not spam<\/td><\/tr><tr><td>Yes<\/td><td>No<\/td><td>12<\/td><td>Spam<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">The data is split into <strong>training data<\/strong>, which the model learns from, and <strong>testing data<\/strong>, which is held back to check whether it really learned.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"324\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/train_test_split-1024x324.png\" alt=\"A bar split into about 80 percent training data, like practice problems, and 20 percent testing data, like the real exam.\" class=\"wp-image-1850\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/train_test_split-1024x324.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/train_test_split-300x95.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/train_test_split-768x243.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/train_test_split.png 1458w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Training data vs. testing data<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Most projects follow the same five steps, and the last one loops back to the start:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"410\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_workflow-1024x410.png\" alt=\"Five-step AI workflow using a study room predictor: define the problem, collect and clean data, train the model, test and evaluate, deploy and monitor.\" class=\"wp-image-1872\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_workflow-1024x410.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_workflow-300x120.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_workflow-768x308.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_workflow-1536x615.png 1536w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/ai_workflow.png 1593w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">The basic AI workflow<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Almost every AI system you use was built this way. Understanding it explains both why AI is powerful and why it fails: a model only knows what its data shows it.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"746\" height=\"760\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_data_network.png\" alt=\"Illustration of colorful data points connected in a network\" class=\"wp-image-1863\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_data_network.png 746w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_data_network-294x300.png 294w\" sizes=\"auto, (max-width: 746px) 100vw, 746px\" \/><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like studying for an exam. Practice problems are the training data, and the real exam is the testing data. If you only memorize practice answers, you\u2019ll do badly on new questions. A model that does this is <strong>overfitting<\/strong>.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">To predict which study rooms at College Library will be busy, you\u2019d collect two years of booking records, train a model to find patterns (weeknights and finals weeks fill up), test it against last month\u2019s real bookings, then launch it and retrain it each semester.<\/p>\n\n\n\n<div class=\"wp-block-group sts-callout-important has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\"><strong>Garbage in, garbage out.<\/strong> Good data is relevant, accurate, diverse, and representative of the people or situations the model will face. On most real projects, collecting and cleaning data takes longer than building the model.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Types of Machine Learning<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Machine learning comes in three main types, depending on what kind of data and feedback the model gets.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"368\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/types_of_learning-1024x368.png\" alt=\"Three cards: supervised learning learns from labeled examples, unsupervised learning finds groups in unlabeled data, and reinforcement learning learns by trial and error with rewards.\" class=\"wp-image-1848\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/types_of_learning-1024x368.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/types_of_learning-300x108.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/types_of_learning-768x276.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/types_of_learning.png 1473w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Three ways machines learn<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Supervised learning<\/strong> uses labeled data. It comes in two kinds:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Classification<\/strong> predicts a category: spam or not spam, fraud or legitimate, cat or dog, positive or negative review.<\/li>\n\n\n\n<li><strong>Regression<\/strong> predicts a number: a house price, tomorrow\u2019s temperature, delivery time, next month\u2019s sales.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Unsupervised learning<\/strong> uses data with no labels. The most common task is <strong>clustering<\/strong>, which groups similar items together based on similarity alone.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"596\" height=\"760\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_clusters.png\" alt=\"Illustration of colorful dots grouped into separate clusters\" class=\"wp-image-1865\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_clusters.png 596w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_clusters-235x300.png 235w\" sizes=\"auto, (max-width: 596px) 100vw, 596px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Reinforcement learning<\/strong> works by trial and error. An agent observes its situation, takes an action, gets a reward or penalty, and improves its strategy to earn more reward over time.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Knowing the type tells you what data a project needs. Supervised learning needs labeled examples, which can be slow and expensive to make. Unsupervised learning works without labels. Reinforcement learning needs a way to score actions.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Supervised learning is like studying with an answer key. Unsupervised learning is like sorting a pile of photos into groups without being told the categories. Reinforcement learning is like training a dog with treats.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Supervised:<\/strong> your email\u2019s spam filter<\/li>\n\n\n\n<li><strong>Unsupervised:<\/strong> a music app grouping listeners with similar taste to build playlists<\/li>\n\n\n\n<li><strong>Reinforcement:<\/strong> AlphaGo learning to beat world champions at Go, and chatbots being polished with human ratings<\/li>\n<\/ul>\n\n\n\n<div class=\"wp-block-group sts-callout-goodtoknow has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">In clustering, the model finds the groups, but a person still decides what each group means.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Probability and Confidence<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">AI rarely knows anything for certain. <strong>Probability<\/strong> lets a model express how confident it is in each possible answer, instead of a simple yes or no.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A model scores every possible answer and picks the highest one. The score is its confidence.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"421\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confidence_scores-1024x421.png\" alt=\"Bar chart of an image classifier\u2019s confidence: dog 80 percent, wolf 15 percent, fox 5 percent.\" class=\"wp-image-1870\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confidence_scores-1024x421.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confidence_scores-300x123.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confidence_scores-768x315.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confidence_scores.png 1371w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Predictions come with confidence scores<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Good systems also update their guesses as new evidence arrives. This is called <strong>conditional probability<\/strong>: the chance of something given what you already know. <strong>Bayes\u2019 rule<\/strong> is the math for it: start with a <em>prior<\/em> belief, add <em>evidence<\/em>, and get an updated <em>posterior<\/em> belief.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"273\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/bayes_update-1024x273.png\" alt=\"Bayes\u2019 rule in three steps: prior belief of 20 percent spam, new evidence of a free prize subject line, posterior belief of about 95 percent spam.\" class=\"wp-image-1871\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/bayes_update-1024x273.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/bayes_update-300x80.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/bayes_update-768x205.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/bayes_update.png 1461w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Bayes\u2019 rule with a spam example<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Statistics helps here too. Models rarely see all the data in the world, so they estimate from samples: averages, rates like how often people click a recommendation, and how much the data varies.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Confidence tells you how much to trust an answer. A 95% guess and a 51% guess can give the same label, but they deserve very different levels of trust.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A weather forecast works the same way. \u201c70% chance of rain\u201d is a confidence score, and forecasters update it as new data comes in.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A medical test result can suggest a disease without proving it. Probability captures how likely the disease is, and that likelihood changes as more test results come in.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Correlation vs. Causation<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">AI is great at finding <strong>correlations<\/strong>, things that change together. It can\u2019t tell you <strong>causation<\/strong>, what actually makes something happen. Ice cream and sunglasses sales both rise every summer, but hot weather causes both. Deciding cause is still a human job.<\/p>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Evaluating a Model<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Evaluation<\/strong> checks how well a model works on data it has never seen. The goal is a model that <strong>generalizes<\/strong>: one that learned the real pattern, not just its training examples.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">First, check the fit. A model can be too simple (<strong>underfitting<\/strong>, also called high bias) or too complex (<strong>overfitting<\/strong>, also called high variance).<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"459\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/model_fit-1024x459.png\" alt=\"Three plots of the same data: underfitting draws a flat line that misses the trend, a good fit follows the trend, and overfitting zigzags through every point.\" class=\"wp-image-1858\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/model_fit-1024x459.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/model_fit-300x135.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/model_fit-768x344.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/model_fit.png 1512w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Underfitting, good fit, and overfitting<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Then, look at the kinds of mistakes it makes, not just overall accuracy:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"617\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confusion_matrix-1024x617.png\" alt=\"Confusion matrix for a disease test: true positive (caught it), false positive (false alarm), false negative (missed it), and true negative (correctly clear).\" class=\"wp-image-1869\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confusion_matrix-1024x617.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confusion_matrix-300x181.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confusion_matrix-768x463.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/confusion_matrix.png 1188w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Four ways a disease test can turn out<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Accuracy alone can fool you. If 1 in 100 people has a disease, a model that always says \u201chealthy\u201d is 99% accurate and never catches a single case.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like grading a smoke detector. One that never goes off is \u201cright\u201d on most days, and useless in a fire. One that goes off every time you make toast is annoying, but safer.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">For a disease screening tool, a false negative (missing a sick patient) is usually far worse than a false alarm. For a spam filter, a false positive (sending a real email to spam) may be the worse mistake. Pick how you measure a model based on the cost of each mistake.<\/p>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Neural Networks and Deep Learning<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A <strong>neural network<\/strong> is a model made of layers of small connected units, loosely inspired by neurons in the brain. <strong>Deep learning<\/strong> means a network with many layers.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"540\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neural_network-1024x540.png\" alt=\"Diagram of a neural network with an input layer, two hidden layers, and an output layer connected by lines.\" class=\"wp-image-1857\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neural_network-1024x540.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neural_network-300x158.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neural_network-768x405.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neural_network.png 1518w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Layers of a neural network<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Each unit takes in numbers, multiplies each by a <strong>weight<\/strong> (how much it matters), adds them up, and passes the total through an <strong>activation function<\/strong>, which lets the network learn curvy, complex patterns instead of only straight lines. The simplest version of this unit is called a <strong>perceptron<\/strong>.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"485\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neuron-1024x485.png\" alt=\"One neuron: inputs cloudy sky, rain forecast, and month are multiplied by weights, added up, passed through an activation function, and output \u201cbring an umbrella: 87%.\u201d\" class=\"wp-image-1856\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neuron-1024x485.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neuron-300x142.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neuron-768x364.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/neuron.png 1518w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Inside a single neuron<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Training adjusts the weights over and over: predict, measure the error, nudge the weights, repeat. The method for choosing which way to nudge them is called <strong>gradient descent<\/strong>.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"471\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/training_loop-1024x471.png\" alt=\"Training loop of predict, measure error, adjust weights, and repeat, next to a curve showing gradient descent stepping down to the lowest error.\" class=\"wp-image-1849\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/training_loop-1024x471.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/training_loop-300x138.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/training_loop-768x353.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/training_loop-1536x707.png 1536w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/training_loop.png 1578w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">The training loop and gradient descent<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">In a deep image model, early layers spot edges, middle layers find shapes and textures, and later layers recognize whole objects.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Neural networks power almost all modern AI: face unlock, speech-to-text, translation, self-driving perception, and every chatbot. The largest have billions of weights.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Gradient descent is like walking downhill in thick fog. You can\u2019t see the bottom, but you can feel which way the ground slopes, so you take a small step down and check again.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Computer vision<\/strong> is a common use. To a computer, a photo is a grid of pixel numbers. A vision network learns which patterns of numbers mean \u201cstop sign,\u201d \u201ccat,\u201d or \u201ctumor on an X-ray.\u201d Tasks include image classification, object detection, and face recognition, which also raises real privacy questions.<\/p>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Embeddings<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">An <strong>embedding<\/strong> is a list of numbers (a <strong>vector<\/strong>) that captures the meaning of something: a word, an image, a song, a product, or a user.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">AI represents everything as numbers. A collection of vectors is stored in a <strong>matrix<\/strong>, and linear algebra lets computers process them efficiently. Things with similar meaning get similar vectors, so they land close together.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"696\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/embeddings_map-1024x696.png\" alt=\"Map of embeddings where dog, puppy, cat, and kitten cluster together, pizza, tacos, and cheese curds cluster together, and Madison, Milwaukee, and Chicago cluster together.\" class=\"wp-image-1868\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/embeddings_map-1024x696.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/embeddings_map-300x204.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/embeddings_map-768x522.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/embeddings_map.png 1218w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Similar meanings sit close together<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Real embeddings have hundreds or thousands of dimensions. <strong>Dimensionality reduction<\/strong> methods like PCA (principal component analysis) squash them down so people can visualize them, keeping the most important structure.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Embeddings explain how AI \u201cunderstands\u201d meaning. They\u2019re the reason search can find what you meant, not just what you typed.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like a map. Cities that are close on a map are near each other in real life. Words that are close in embedding space are close in meaning.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Search that understands \u201ccheap laptop for college\u201d matches \u201caffordable student notebook\u201d<\/li>\n\n\n\n<li>\u201cYou might also like\u201d recommendations on shopping and music apps<\/li>\n\n\n\n<li>Language models representing every word they read<\/li>\n<\/ul>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Large Language Models (LLMs)<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A <strong>large language model (LLM)<\/strong> is a neural network trained on a huge amount of text to predict what comes next. LLMs power chat assistants like ChatGPT, Claude, Gemini, and Copilot. They\u2019re a type of <strong>generative AI<\/strong>, which creates new content instead of only sorting or predicting.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"720\" height=\"760\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_book_of_text.png\" alt=\"Illustration of an open book with streams of text rising from its pages\" class=\"wp-image-1866\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_book_of_text.png 720w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_book_of_text-284x300.png 284w\" sizes=\"auto, (max-width: 720px) 100vw, 720px\" \/><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">LLMs are built on the <strong>Transformer<\/strong>, a design introduced by Google researchers in 2017. Its key idea is <strong>attention<\/strong>: for each word, the model looks at every other word and decides which ones matter most. In \u201cThe trophy didn\u2019t fit in the suitcase because it was too big,\u201d attention works out that \u201cit\u201d means the trophy.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">When it writes, an LLM scores every possible next piece of text, picks one, adds it, and repeats until the answer is done.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"487\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/next_token-1024x487.png\" alt=\"Next-token prediction for \u201cOn a cold morning in Madison, I need a cup of\u201d: coffee 62 percent, hot 14 percent, tea 11 percent, cocoa 5 percent.\" class=\"wp-image-1855\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/next_token-1024x487.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/next_token-300x143.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/next_token-768x365.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/next_token.png 1368w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Predicting the next token<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">A chat assistant goes through three stages of training to get from text predictor to helpful assistant:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"329\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/llm_training-1024x329.png\" alt=\"Three stages of LLM training: pretraining on huge amounts of text, fine-tuning on example conversations, and learning from human feedback.\" class=\"wp-image-1859\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/llm_training-1024x329.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/llm_training-300x97.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/llm_training-768x247.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/llm_training.png 1470w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">From text predictor to helpful assistant<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Knowing that an LLM predicts likely text, rather than looking up facts, explains both its strengths (fluent writing, explanation, coding) and its biggest weakness: it can sound right and be wrong.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like your phone\u2019s autocomplete, scaled up enormously. Autocomplete suggests the next word; an LLM keeps going for whole essays, using everything in the conversation as context.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Ask the same question twice and you may get two different answers. There\u2019s a bit of randomness in each pick, so the model doesn\u2019t always choose the single most likely word.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Reasoning and Multimodal Models<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Newer <strong>reasoning models<\/strong> \u201cthink\u201d before answering: they work through a problem step by step and check their work. They\u2019re slower but much better at math, coding, and planning. <strong>Multimodal models<\/strong> take in more than text, like photos, screenshots, PDFs, and voice.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Natural Language Processing<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">LLMs are the newest part of <strong>natural language processing (NLP)<\/strong>, the area of AI focused on human language. NLP also includes translation, summarization, question answering, sentiment analysis, and spam detection. Language is hard for computers because meaning depends on context, tone, culture, and ambiguity.<\/p>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Tokens and Context Windows<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Tokens<\/strong> are the chunks of text an LLM actually reads, often pieces of words. The <strong>context window<\/strong> is everything the model can \u201csee\u201d at once.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"396\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/tokens_context-1024x396.png\" alt=\"The phrase \u201cBucky Badger loves unbelievably cheesy pizza\u201d split into 8 tokens, above a box showing the context window holds your messages, the AI\u2019s replies, and uploaded files.\" class=\"wp-image-1851\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/tokens_context-1024x396.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/tokens_context-300x116.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/tokens_context-768x297.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/tokens_context.png 1421w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Tokens and the context window<\/figcaption><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li>As a rule of thumb, 100 tokens is about 75 words.<\/li>\n\n\n\n<li>The context window holds your messages, the AI\u2019s replies, and any files you upload.<\/li>\n\n\n\n<li>Today\u2019s models can hold hundreds of pages at once, but very long chats can still lose track of early details.<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Usage limits and pricing for AI tools are counted in tokens. And understanding the context window explains why a chat \u201cforgets\u201d things, and how to fix it.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">The context window is like short-term memory, or the whiteboard in a meeting. Everything written on it is in view; once it fills up or you start a new meeting, older notes are gone.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">If a long chat starts giving worse answers, open a <strong>New chat<\/strong> and paste in a short summary of only what matters.<\/p>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Hallucinations<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A <strong>hallucination<\/strong> is when an AI states something false with total confidence, like a made-up fact, quote, statistic, or citation.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"644\" height=\"760\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_fact_check.png\" alt=\"Illustration of a magnifying glass inspecting lines of text\" class=\"wp-image-1862\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_fact_check.png 644w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_fact_check-254x300.png 254w\" sizes=\"auto, (max-width: 644px) 100vw, 644px\" \/><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">LLMs predict what sounds right, not what is true. They have no built-in fact check, and they\u2019re trained to always give an answer, so gaps in their knowledge get filled with plausible-sounding text.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">This is the single most important limitation to understand before relying on AI for school or work.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like a student who didn\u2019t do the reading but answers confidently anyway. The answer might sound great and still be wrong.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Ask for \u201cthree research papers on this topic, with authors and years,\u201d then search for each one. Often at least one doesn\u2019t exist.<\/p>\n\n\n\n<div class=\"wp-block-group sts-callout-warning has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\"><strong>Watch out for:<\/strong> citations, quotes, and links; very specific numbers, dates, and names; niche or very recent topics; and math done \u201cin its head.\u201d Ask for sources and open them, turn on web search or give it your own documents, and double-check anything you\u2019ll submit or act on.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Retrieval-Augmented Generation (RAG)<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>RAG<\/strong> means the AI looks up relevant information before it answers, then writes its answer from those sources.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"315\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/rag_flow-1024x315.png\" alt=\"RAG flow: your question, then a search for relevant passages, then those passages are added to the prompt, then an answer with citations.\" class=\"wp-image-1853\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/rag_flow-1024x315.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/rag_flow-300x92.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/rag_flow-768x236.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/rag_flow.png 1503w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">How retrieval-augmented generation works<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">RAG makes answers more accurate and current, and lets you check them through citations. It cuts down on hallucinations, though it doesn\u2019t remove them.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s an open-book exam instead of answering from memory.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Copilot or ChatGPT with web search turned on<\/li>\n\n\n\n<li>Google\u2019s NotebookLM answering questions from your course readings<\/li>\n\n\n\n<li>A support chatbot that answers from a company\u2019s help pages<\/li>\n<\/ul>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Technique: Writing Good Prompts<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A <strong>prompt<\/strong> is what you type to an AI. Prompting is the skill of giving it the context and instructions it needs to give you a useful answer.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How do you use it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A strong prompt often has five parts:<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>Part<\/th><th>What it tells the AI<\/th><th>Example<\/th><\/tr><\/thead><tbody><tr><td>Context<\/td><td>Who you are and what it\u2019s for<\/td><td>\u201cI\u2019m a first-year biology student studying for a quiz.\u201d<\/td><\/tr><tr><td>Task<\/td><td>Exactly what you want<\/td><td>\u201cExplain photosynthesis.\u201d<\/td><\/tr><tr><td>Constraints<\/td><td>Length, level, tone, limits<\/td><td>\u201cIn under 150 words, using a simple everyday analogy.\u201d<\/td><\/tr><tr><td>Format<\/td><td>How the answer should look<\/td><td>\u201cThen give me 3 practice questions, with answers at the end.\u201d<\/td><\/tr><tr><td>Examples<\/td><td>What good looks like<\/td><td>Paste a sample answer or a style you like<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Then treat it as a conversation. Habits that help:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Follow up.<\/strong> Your first prompt is a draft. Ask for \u201cshorter,\u201d \u201cmore examples,\u201d or \u201ctry again.\u201d<\/li>\n\n\n\n<li><strong>Share your material.<\/strong> Paste notes or attach files so it works from your stuff, not guesses.<\/li>\n\n\n\n<li><strong>Let it ask you.<\/strong> End with \u201cAsk me any questions you need before you start.\u201d<\/li>\n\n\n\n<li><strong>Break it up.<\/strong> Ask for an outline first, then each section.<\/li>\n\n\n\n<li><strong>Check the work.<\/strong> Ask it to review its answer, then verify the important parts yourself.<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">The same tool gives dramatically better results with a better prompt. It\u2019s the most useful AI skill you can build, and it works in every chat assistant.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like briefing a new teammate or writing a good email to a TA. The more specific you are about what you need, the better the help you get.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"497\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/prompt_anatomy-1024x497.png\" alt=\"A weak prompt, \u201cExplain photosynthesis,\u201d compared with a strong prompt color-coded by context, task, constraints, and format.\" class=\"wp-image-1854\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/prompt_anatomy-1024x497.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/prompt_anatomy-300x146.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/prompt_anatomy-768x373.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/prompt_anatomy.png 1428w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">A weak prompt vs. a strong prompt<\/figcaption><\/figure>\n\n\n\n<div class=\"wp-block-group sts-callout-goodtoknow has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">You don\u2019t need all five parts every time. Adding even two makes a big difference.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: AI Agents<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">An <strong>AI agent<\/strong> is an AI system that works toward a goal by planning, using tools, and taking actions, instead of only answering a question.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"540\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/agent_loop-1024x540.png\" alt=\"Agent loop of plan, act, observe, and adjust around a goal, with tools like web search, running code, editing files, and sending email.\" class=\"wp-image-1875\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/agent_loop-1024x540.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/agent_loop-300x158.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/agent_loop-768x405.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/agent_loop.png 1398w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">How an AI agent works<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Agents connect to tools through standards like <strong>MCP (Model Context Protocol)<\/strong>, an open standard that lets AI apps plug into services like Google Drive, GitHub, or Slack.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Agents are where AI is heading: from writing about a task to actually doing it. Knowing how they work helps you use them safely.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A chatbot is like asking a librarian a question. An agent is like asking a research assistant to go find sources, read them, and write up a summary for you. MCP is like a USB port for AI.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Deep research tools that read dozens of web pages and write a report<\/li>\n\n\n\n<li>Browser agents that fill out forms<\/li>\n\n\n\n<li>Coding agents that edit, run, and test a whole project<\/li>\n<\/ul>\n\n\n\n<div class=\"wp-block-group sts-callout-warning has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">Agents can take real actions, like sending messages or deleting files. Keep a human in the loop for anything that matters.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Technique: Vibe Coding<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Vibe coding<\/strong> means building software by describing what you want in plain English and letting AI write the code. AI researcher Andrej Karpathy coined the term in early 2025.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"606\" height=\"760\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_code_laptop.png\" alt=\"Illustration of a laptop with screens full of code\" class=\"wp-image-1864\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_code_laptop.png 606w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_code_laptop-239x300.png 239w\" sizes=\"auto, (max-width: 606px) 100vw, 606px\" \/><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">How do you use it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Work in small loops and test after every change:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"368\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/vibe_coding-1024x368.png\" alt=\"Vibe coding loop: describe, generate, run, test, and refine.\" class=\"wp-image-1847\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/vibe_coding-1024x368.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/vibe_coding-300x108.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/vibe_coding-768x276.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/vibe_coding.png 1509w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">The vibe coding workflow<\/figcaption><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Start small.<\/strong> Get a tiny version working before adding features.<\/li>\n\n\n\n<li><strong>Ask why.<\/strong> Have the AI explain any code you don\u2019t understand.<\/li>\n\n\n\n<li><strong>Save versions.<\/strong> Keep copies (or use Git) so you can undo mistakes.<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Anyone can now build small tools, websites, and prototypes without years of programming experience. It\u2019s also a great way to see how code works.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Popular tools include Claude Code, Cursor, GitHub Copilot, Replit, Lovable, and v0. Any chat assistant, including Copilot Chat, can write simple apps too.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Ask Copilot Chat: \u201cWrite a single HTML file for a Pomodoro study timer with start, pause, and reset buttons.\u201d Save it as an HTML file (see Saving and Opening an HTML File below), open it in your browser, then ask for one improvement at a time.<\/p>\n\n\n\n<div class=\"wp-block-group sts-callout-warning has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">Watch out for code you don\u2019t understand, security holes, and bugs that pile up quietly. Never paste passwords or API keys into prompts or code. In a programming class, check the course AI policy first.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Tool: Teachable Machine<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">A free website from Google that lets you train your own image, sound, or pose recognition model in the browser, with no code.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How do you use it?<\/h4>\n\n\n\n<ol class=\"wp-block-list\">\n<li>Go to <strong>teachablemachine.withgoogle.com<\/strong> and click <strong>Get Started<\/strong>.<\/li>\n\n\n\n<li>Choose <strong>Image Project<\/strong>, then <strong>Standard image model<\/strong>.<\/li>\n\n\n\n<li>Rename each class to the things you want to tell apart (click the pencil icon next to its name).<\/li>\n\n\n\n<li>Add examples to each class with <strong>Webcam<\/strong> (click and hold <strong>Hold to Record<\/strong>) or <strong>Upload<\/strong>.<\/li>\n\n\n\n<li>Click <strong>Train Model<\/strong>, and keep the tab open while it trains.<\/li>\n\n\n\n<li>Test your model in the <strong>Preview<\/strong> panel. The bars show its confidence for each class.<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">You see the whole machine learning workflow in about ten minutes: collect data, label it, train, and test. It makes training data and confidence scores concrete.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s a simplified version of what data scientists do with tools like Python, TensorFlow, and PyTorch.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Make two classes, \u201cThumbs up\u201d and \u201cThumbs down,\u201d record about 50 images of each, and train. Then change the lighting or background and watch the confidence scores drop. That shows why training data needs variety.<\/p>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Technique: Saving and Opening an HTML File<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">An HTML file is a web page saved on your computer. You can open it in any browser with no internet or hosting needed, which makes it the fastest way to run a small app that AI writes for you.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How do you use it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>On Windows:<\/strong> open Notepad, paste the code, click <strong>File &gt; Save As<\/strong>, set <strong>Save as type<\/strong> to <strong>All files<\/strong>, and name the file something like <code class=\"\" data-line=\"\">timer.html<\/code>. Double-click it to open it in your browser.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>On a Mac:<\/strong> open TextEdit, click <strong>Format &gt; Make Plain Text<\/strong> first, paste the code, click <strong>File &gt; Save<\/strong>, name it <code class=\"\" data-line=\"\">timer.html<\/code>, and choose <strong>Use .html<\/strong> if asked. Double-click it to open it.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It lets you try vibe coding without installing anything.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like saving a document, except the browser runs it as a web page instead of showing text.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">After any change to the code, save the file again and refresh the browser tab to see it.<\/p>\n\n\n\n<div class=\"wp-block-group sts-callout-important has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\">On a Mac, the Make Plain Text step matters. Without it, the browser shows code instead of your app. On Windows, make sure the file isn\u2019t saved as timer.html.txt.<\/p>\n<\/div>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Concept: Using AI Responsibly<\/h2>\n\n\n\n<h4 class=\"wp-block-heading\">What is it?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Responsible AI use means using AI in ways that are fair, safe, private, honest, and accountable. AI affects real people, so a system can be impressive and still cause harm.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">How does it work?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Ask four questions about any AI system: <strong>Fairness<\/strong> (does it treat all groups equally well?), <strong>Privacy<\/strong> (whose data was used, and how is yours protected?), <strong>Safety<\/strong> (could it spread false information or deepfakes?), and <strong>Accountability<\/strong> (who is responsible when it\u2019s wrong?).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Bias:<\/strong> if training data reflects unfair patterns from the past, the model learns and repeats them. Teams reduce bias with diverse, representative data, by testing accuracy for each group, and with human review of high-stakes decisions.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"729\" height=\"760\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_people_crowd.png\" alt=\"Illustration of a diverse crowd of people\" class=\"wp-image-1860\" style=\"width:340px;height:auto\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_people_crowd.png 729w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/illus_people_crowd-288x300.png 288w\" sizes=\"auto, (max-width: 729px) 100vw, 729px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Explainability:<\/strong> in healthcare, hiring, finance, education, and law, people need to understand why an AI made a decision so it can be checked and challenged.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Your data:<\/strong> anything you type into an AI tool leaves your computer. Free consumer tools may use your chats to improve their models unless you turn that off.<\/p>\n\n\n\n<div class=\"wp-block-group sts-callout-warning has-global-padding is-layout-constrained wp-block-group-is-layout-constrained\">\n<p class=\"wp-block-paragraph\"><strong>Don\u2019t paste:<\/strong> sensitive or restricted university data (even into Copilot Chat), other people\u2019s personal information like student records or health details, passwords, API keys, financial information, or unpublished research without permission.<\/p>\n<\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Coursework:<\/strong> every course sets its own AI rules, and they can differ between assignments. Check the syllabus, ask your instructor when unsure, disclose how you used AI, keep your prompts and drafts, and remember you\u2019re responsible for everything you submit.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Why is this helpful for beginners to learn about?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">AI mistakes can affect grades, jobs, health, and privacy. Good habits protect you and the people your work affects.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Is it similar to anything else?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">It\u2019s like citing sources and protecting passwords: basic habits that become second nature once you practice them.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Example<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Amazon scrapped an experimental hiring tool in 2018 after it learned, from ten years of past r\u00e9sum\u00e9s, to downgrade r\u00e9sum\u00e9s that mentioned women\u2019s groups. The model learned the bias in its history.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"572\" src=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/responsible_ai-1024x572.png\" alt=\"Responsible AI checklist: verify, protect privacy, know the limits, be transparent, and stay in charge.\" class=\"wp-image-1852\" srcset=\"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/responsible_ai-1024x572.png 1024w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/responsible_ai-300x167.png 300w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/responsible_ai-768x429.png 768w, https:\/\/sts.doit.wisc.edu\/training-materials\/wp-content\/uploads\/2026\/10\/responsible_ai.png 1338w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Your responsible AI checklist<\/figcaption><\/figure>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Glossary<\/h2>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>Term<\/th><th>Meaning<\/th><\/tr><\/thead><tbody><tr><td>Activation function<\/td><td>The step in a neuron that bends the math, so networks can learn complex patterns<\/td><\/tr><tr><td>Agent<\/td><td>An AI system that plans, uses tools, and takes actions toward a goal<\/td><\/tr><tr><td>Algorithm<\/td><td>A step-by-step set of instructions a computer follows<\/td><\/tr><tr><td>Attention<\/td><td>The Transformer\u2019s way of deciding which words matter most to each other<\/td><\/tr><tr><td>Bias (data)<\/td><td>Unfair patterns in data that a model learns and repeats<\/td><\/tr><tr><td>Classification<\/td><td>Predicting a category, like spam or not spam<\/td><\/tr><tr><td>Clustering<\/td><td>Grouping similar items together without labels<\/td><\/tr><tr><td>Context window<\/td><td>Everything a language model can see at once in a conversation<\/td><\/tr><tr><td>Deep learning<\/td><td>Machine learning with neural networks that have many layers<\/td><\/tr><tr><td>Dimensionality reduction<\/td><td>Compressing many features into fewer while keeping the important structure (for example, PCA)<\/td><\/tr><tr><td>Embedding<\/td><td>A list of numbers that captures meaning; similar things get similar numbers<\/td><\/tr><tr><td>Features<\/td><td>The inputs a model uses to make a prediction<\/td><\/tr><tr><td>Fine-tuning<\/td><td>Extra training that adapts a model to a specific job, like following instructions<\/td><\/tr><tr><td>Generative AI<\/td><td>AI that creates new text, images, audio, video, or code<\/td><\/tr><tr><td>Gradient descent<\/td><td>The method for adjusting weights a little at a time to reduce error<\/td><\/tr><tr><td>Hallucination<\/td><td>When an AI states something false with confidence<\/td><\/tr><tr><td>Labels<\/td><td>The correct answers a model learns to predict<\/td><\/tr><tr><td>Large language model (LLM)<\/td><td>A model trained on huge amounts of text to predict the next token<\/td><\/tr><tr><td>Machine learning<\/td><td>AI that learns patterns from data instead of following fixed rules<\/td><\/tr><tr><td>MCP<\/td><td>Model Context Protocol, an open standard for connecting AI apps to other tools<\/td><\/tr><tr><td>Overfitting<\/td><td>When a model memorizes training data and fails on new data<\/td><\/tr><tr><td>Perceptron<\/td><td>The simplest unit of a neural network: weighted inputs, summed, then an output<\/td><\/tr><tr><td>Prompt<\/td><td>The instructions or question you give an AI<\/td><\/tr><tr><td>RAG<\/td><td>Retrieval-augmented generation: the AI looks up sources before answering<\/td><\/tr><tr><td>Regression<\/td><td>Predicting a number, like a price or temperature<\/td><\/tr><tr><td>Reinforcement learning<\/td><td>Learning by trial and error with rewards and penalties<\/td><\/tr><tr><td>Token<\/td><td>A chunk of text, often part of a word, that a language model reads<\/td><\/tr><tr><td>Transformer<\/td><td>The neural network design behind modern chatbots, built around attention<\/td><\/tr><tr><td>Underfitting<\/td><td>When a model is too simple to capture the pattern<\/td><\/tr><tr><td>Vector<\/td><td>A list of numbers; how AI represents words, images, and users<\/td><\/tr><tr><td>Vibe coding<\/td><td>Building software by describing it in plain English and letting AI write the code<\/td><\/tr><tr><td>Weight<\/td><td>A number in a neural network that says how much an input matters<\/td><\/tr><\/tbody><\/table><\/figure>\n<\/section>\n\n\n\n<section class=\"wp-block-sts-block-sts-custom-sections\">\n<h2 class=\"wp-block-heading\">Further Reading and Resources<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">All of these are free.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/www.elementsofai.com\/\">Elements of AI<\/a>: a beginner-friendly online course on AI basics, no coding needed<\/li>\n\n\n\n<li><a href=\"https:\/\/www.3blue1brown.com\/topics\/neural-networks\">3Blue1Brown: Neural Networks<\/a>: visual videos on how neural networks and LLMs work<\/li>\n\n\n\n<li><a href=\"https:\/\/www.youtube.com\/watch?v=zjkBMFhNj_g\">Andrej Karpathy: Intro to Large Language Models<\/a>: a one-hour talk on how chatbots are built<\/li>\n\n\n\n<li><a href=\"https:\/\/developers.google.com\/machine-learning\/crash-course\">Google Machine Learning Crash Course<\/a>: hands-on machine learning with interactive exercises<\/li>\n\n\n\n<li><a href=\"https:\/\/www.kaggle.com\/learn\">Kaggle Learn<\/a>: short Python and machine learning lessons in the browser<\/li>\n\n\n\n<li><a href=\"https:\/\/teachablemachine.withgoogle.com\/\">Teachable Machine<\/a>: train your own model in the browser<\/li>\n\n\n\n<li><a href=\"https:\/\/it.wisc.edu\/services\/m365-copilot-chat\/\">Microsoft 365 Copilot Chat at UW-Madison<\/a>: access instructions and data rules<\/li>\n\n\n\n<li><a href=\"https:\/\/sts.doit.wisc.edu\">Software Training for Students<\/a>: more free workshops, office hours, and 1-on-1 help<\/li>\n<\/ul>\n<\/section>\n","protected":false},"author":19,"featured_media":0,"template":"","meta":{"_acf_changed":false,"_uw_seo_meta_title":"","_uw_seo_meta_description":"","_uw_seo_twitter_card_type":"summary_large_image","_uw_seo_meta_image":"","_uw_seo_meta_image_url":"","_uw_seo_meta_image_sizes":[],"_uw_seo_custom_meta_tags":[],"footnotes":""},"categories":[],"class_list":["post-679","manual","type-manual","status-publish","hentry"],"acf":[],"_links":{"self":[{"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/manual\/679","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/manual"}],"about":[{"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/types\/manual"}],"author":[{"embeddable":true,"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/users\/19"}],"version-history":[{"count":7,"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/manual\/679\/revisions"}],"predecessor-version":[{"id":1885,"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/manual\/679\/revisions\/1885"}],"wp:attachment":[{"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/media?parent=679"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/sts.doit.wisc.edu\/training-materials\/wp-json\/wp\/v2\/categories?post=679"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}