_COVER.png)
What Are Large Language Models (LLMs)?
Quality reviewed September 2026
Facts verified September 20, 2026
- Subject
- Career & Life Skills
- Grades
- Grades 9-12
- Length
- 9-page student worksheet
- Time
- About one 50-minute class period
- Great for
- Sub plans and independent work
- Includes
- Reading passage, comprehension questions + full answer key
- Format
- Print-and-go PDF
What students do
- Parameters, hallucination, and autoregressive generation are three of five terms up for matching
- Marcus's failing history essay and its fabricated sources anchor two case study questions
- Three fill-in-the-blank sentences pull text, hallucination, and generate straight from the passage
- Trace how an LLM predicts one word at a time instead of retrieving a stored answer
What it covers
- Six passages on how large language models are trained, generate text, and hit their limits
- A case study on Marcus submitting an AI-written history essay full of invented facts
- A key info table naming OpenAI, Google, Anthropic, and the knowledge cutoff problem
Start reading this worksheet
The Machine That Learned to Read the Internet
If you have ever typed a question into ChatGPT, Gemini, or Claude and gotten a full, coherent answer back, you were talking to a Large Language Model, or LLM. These are AI systems trained to understand and generate human language, and they are behind almost every AI chatbot, writing assistant, and customer service bot you encounter today. But what exactly are they, and why are they so different from the software that came before? Traditional computer programs follow rules. A programmer writes instructions, and the computer executes them exactly. If you type in a math problem, the program runs a formula. If you ask it something outside its programmed rules, it fails. LLMs work completely differently. Instead of following written rules, they learn patterns from reading enormous amounts of text, and then they use those patterns to predict what words should come next in any given situation.
Trained on Text, Not Programmed With Rules
The word 'large' in Large Language Model refers to two things: the amount of text used to train the model and the number of internal settings, called parameters, the model uses to make predictions. Modern LLMs like GPT-4 have been trained on hundreds of billions of words pulled from books, websites, news articles, Wikipedia, academic papers, and much more. They have hundreds of billions to trillions of parameters. During training, the model reads text and tries to predict the next word in a sentence over and over again, billions of times. Every time it gets it wrong, the system adjusts its internal settings slightly. By the time training is done, the model has essentially memorized the statistical patterns of human language. It knows that after the word 'the,' a noun often follows. It knows that 'president' often appears near words like 'White House,' 'election,' and 'policy.' It has absorbed the style, structure, and meaning of an enormous amount of human communication. This process is called unsupervised learning. Unlike earlier AI systems that needed humans to label every example, LLMs train themselves by reading raw text and learning from patterns. No one had to label each sentence. The model figured it out on its own.
How LLMs Generate Text
When you type a question into an LLM, it does not look up an answer in a database. It predicts, word by word, what the most statistically likely response would be given everything it learned during training. This is called autoregressive generation. The model picks the most probable next word, adds it to the sentence, then picks the next most probable word after that, and keeps going until the response is complete. This is why LLMs can write in so many different styles, answer questions about almost any topic, and even tell jokes or write poetry. They are not retrieving stored answers. They are constructing responses from scratch using patterns they absorbed during training. It is also why they can make things up convincingly. If the pattern points toward a certain answer, the model will produce it even if the answer is wrong. This is called hallucination, and it is one of the biggest limitations of LLMs.
Like what you read? The full reading, the case study, and every question set are in the free download below, along with the answer key.
Why this matters
Every student in this class will eventually type a question into ChatGPT, Gemini, or Claude, and this reading explains what is actually happening when they hit enter. Once a student understands that the model is predicting plausible words rather than looking up verified facts, Marcus's mistake, submitting an essay with invented events and dead source links, stops being a one-off embarrassment and becomes a warning about how these tools fail. That understanding changes how a student treats an AI answer on a research paper, a lab report, or a college application essay, because the confident tone of the response is not evidence that the content is true.
From one teacher to another
A large language model is not running search behind the scenes, and this worksheet makes that distinction concrete. Students read how models like GPT-4 train on hundreds of billions of words by repeatedly guessing the next word and adjusting billions of internal parameters when they guess wrong. A passage compares this autoregressive, word-by-word generation to a search engine's retrieval process, then explains why LLMs cost tens to hundreds of millions of dollars to build. The reading also covers knowledge cutoffs and the emergent abilities researchers noticed once models crossed a certain scale, such as GPT-3 suddenly handling arithmetic and analogies. The case study follows Marcus, a junior who submitted an AI-written history essay with three wrong facts and two dead source links.
What you get
Download
Where should I send the next one?
Everything here is free and it stays that way. Leave your email and I'll send you the new lessons as they go up. You can unsubscribe from any one of them.
Related resources
How AI Generates Images: From Words to Pictures
How AI Generates Voice and Video
How AI Learns: Training Data Explained
The Black Box Problem: When AI Can't Explain Itself
The Cost of AI: Energy, Water, and Data Centers
The Internet as a Textbook: Where AI Gets Its Information
Every lesson on this site is free to download and use. If one of them saved you a prep period, you can chip in a few bucks toward my three kids' daycare fund. That's genuinely where tips go.