AI FLUENCY · PHASE 1: AWARENESS
Module 3 — The Four Properties of AI | Lesson 011
Working Memory: What Is the Context Window?
Understand how AI uses earlier information during a conversation and why its working context has limits.
Watch this 54-second silent visual lesson. Then explore the explanations and examples below.
Watch on YouTube →Introduction
Have you noticed that an AI assistant can follow instructions you provided earlier in a conversation?
For example, you might tell ChatGPT, Claude, or Gemini to explain everything in simple language.
Several messages later, the assistant may continue following that instruction.
But sometimes, especially in long conversations, AI may lose track of details you previously mentioned.
This behavior relates to a fundamental property of Large Language Models: working memory, commonly explained through the context window.
1. What Is a Context Window?
A context window is the amount of information an AI model can process together when generating a response.
It may include the current question, earlier messages, instructions, and relevant documents.
The information available in the context helps the model interpret the user's request and generate an appropriate continuation.
The context window is the information the language model can work with while generating its current response.
2. How Does AI Working Memory Help?
Imagine that you are preparing a classroom lesson.
You tell an AI assistant:
"I am teaching 10-year-old students. Use simple language."
Then you ask:
"Explain climate change."
The AI can use your earlier instruction to generate an age-appropriate explanation because that instruction remains in the context.
"Climate change means the Earth is getting warmer over time. This can affect weather, animals, plants, and people."
Notice that the response uses relatively simple language.
The earlier instruction helped shape the answer to the current question.
3. How Does the Context Window Work?
An LLM processes information in units called tokens.
Tokens may represent words, parts of words, punctuation, or other text segments.
A context window has a maximum token capacity.
The model uses the available tokens to help determine what to generate next.
Information Available to the AI
Use simple language.
+
Explain climate change.
↓
Explanation in simple language.
Simplified illustration of how context influences an AI response.
The model's available context may also contain system instructions, tool results, or selected information provided by an application.
4. Is AI Working Memory Unlimited?
No.
Every language model has a limit on how much information it can process at one time.
This limit is commonly measured in tokens.
When a conversation or document exceeds the available context capacity, some information may need to be removed, summarized, or excluded.
A Simplified Context Window
Currently Available Context
Current question
Relevant earlier instructions
Selected conversation history
Potentially Outside the Context
Older messages or details that are no longer included.
Illustrative only. Applications may manage older conversations through different mechanisms.
An important instruction may therefore stop influencing a response if it is no longer included in the model's current input.
However, even information within the context is not guaranteed to be used perfectly.
5. Is Context the Same as Knowledge?
No. These concepts describe different sources of information.
Knowledge
Information and relationships learned during model training.
Context
Information available to the model while generating its current response.
Saved Memory
Information an AI application may store and retrieve separately across interactions.
A model's context window is not automatically the same as an application's saved memory.
Applications may use tools, retrieval systems, or memory features to bring relevant information into a new conversation.
What the model can actually use depends on what information is made available to it.
6. How Can We Use Context Effectively?
Understanding context helps us communicate more effectively with AI assistants.
- State your goal clearly. Explain what you want the AI to accomplish.
- Provide relevant background. Include details important to the task.
- Specify essential requirements. Mention the audience, format, constraints, or important preferences.
- Summarize long discussions. Restate important decisions when a conversation becomes lengthy.
- Repeat critical instructions when needed. Do not assume every earlier detail remains available indefinitely.
Good context improves the chances of getting a relevant response, although it does not guarantee perfect results.
7. Why Does This Matter for AI Fluency?
AI Fluency involves more than simply asking questions.
It includes understanding what information the AI needs to perform a task effectively.
When we provide relevant context, we help the model interpret our goals and constraints.
When we understand context limits, we are less likely to assume that AI has retained everything from a long discussion.
This awareness supports better communication, evaluation, and responsible AI use.
Key Takeaway
The context window is the information an AI model can process together when generating a response.
It acts as a form of working memory, but it is limited and distinct from learned knowledge or separately stored memory.
Think About It
Imagine working with an AI assistant on a project over many conversations.
You gave important requirements at the beginning, but later the AI appears to ignore them.
What might explain this behavior?
How could summarizing the project requirements help the AI produce a more relevant response?
What Have You Learned?
After completing this lesson, you should be able to:
- Explain what a context window is.
- Describe how earlier instructions can influence current responses.
- Recognize why a context window has a limited capacity.
- Distinguish context from learned knowledge and saved memory.
- Provide relevant information to improve AI interactions.
CONTINUE LEARNING
Next: Lesson 012
Steerability: Why Does AI Follow Instructions?
Now that you understand how context influences AI-generated responses, the next lesson explores steerability.
You will learn how instructions influence AI behavior, why responses change when prompts change, and why instruction-following still has limitations.
Further Learning Resources
Your AI Fluency Journey
AI Fluency develops through four connected phases.
Understanding how AI works helps us use it responsibly to accomplish meaningful goals.
