Module 1 — Understanding AI and Using It Safely

Purpose: Build a practical understanding of AI, language models, prompts, and the safeguards that make everyday use safer.

1. What is artificial intelligence?

Artificial intelligence (AI) is a broad field of computing concerned with building systems that perform tasks commonly associated with human intelligence. These tasks include recognizing spoken words, identifying objects in photographs, translating languages, finding patterns in large collections of information, and helping people make decisions. AI is not a single product or a single kind of machine. A navigation app that predicts traffic, an email service that filters spam, and a conversational assistant are different applications of AI.

Most modern AI systems learn statistical relationships from examples rather than following only a fixed set of instructions written by a programmer. During training, a system may examine large quantities of text, images, or other data and adjust internal numerical parameters to improve its performance. After training, it applies what it has learned to new inputs. This can be powerful, but learning patterns is not the same as understanding a person’s circumstances or establishing that a statement is true.

For home users, the most important distinction is between an AI tool that assists and a person who remains accountable. AI can make everyday tasks easier, but it should not replace independent judgment—especially when health, money, safety, legal rights, or personal information are involved. Different services have different capabilities and data practices; always check the specific product you are using.

2. What is a large language model (LLM)?

A large language model, or LLM, is a type of AI trained on substantial collections of language and other data to process and generate text. Many popular AI chat assistants use LLMs. When you type a question, the model considers the words and context in your request and generates a response one piece at a time. It can explain a concept, rewrite a letter, translate a paragraph, suggest ideas, or help organize a plan.

An LLM does not operate like a traditional encyclopedia that simply retrieves a verified entry. It generates a response based on patterns learned during training and any additional information or tools made available to it. Some assistants can search the web, inspect files, or use connected applications; others cannot. A model may sound certain while making a factual mistake, inventing a citation, or missing recent changes. The quality of the language is not proof of the quality of the evidence.

LLMs are useful collaborators for thinking and writing, but they are not people, licensed professionals, or independent authorities. They can reflect biases in their training material or in a user’s question. Treat the response as a draft or starting point when accuracy matters, and verify important claims using original documents, trusted institutions, or qualified professionals.

3. What AI does in everyday life

AI can classify information (such as flagging spam), recognize patterns (such as identifying objects in photos), predict likely outcomes (such as estimated travel times), recommend content, and generate new material. Generative AI can create or transform text, images, audio, video, and software code. Some systems combine these abilities—for example, describing an uploaded picture and then drafting a caption.

AI can also help structure complicated tasks: comparing options, outlining questions for an appointment, explaining unfamiliar terms, creating a household checklist, or summarizing a public article. A tool connected to a calendar, email account, or shopping service may be able to take actions rather than merely suggest them. Those connected abilities require additional care because a mistake can affect real accounts, files, or purchases.

What a system can actually do depends on the particular service, its settings, available tools, and access permissions. Never assume that an assistant has checked the internet, consulted a source, or completed an action unless you can confirm it.

4. Prompts: how to ask AI for useful help

A prompt is the instruction, question, or material you give an AI system. It might be one sentence or several paragraphs. Good prompts explain the goal, relevant non-sensitive context, the audience, and the desired output. For example, you can ask for “a plain-language explanation in five bullets” or “a comparison table with questions I should investigate.”

You do not need special commands to begin. Ask one clear question, examine the answer, and follow up: “What assumptions did you make?”, “Explain that more simply,” or “Which claims should I verify?” Do not include passwords, account numbers, confidential documents, or unnecessary personal details. You can replace identifying facts with placeholders such as “[my bank]” or “[my town].”

Five safe example prompts:

• “Explain how a home Wi-Fi router works in everyday language, and give me five security settings to check.”
• “Create a weekly household meal-planning template using inexpensive ingredients; do not assume any medical dietary needs.”
• “Help me draft a polite message asking a service provider for a written estimate. Use placeholders instead of my address or phone number.”
• “Explain the difference between a credit freeze and a fraud alert. Tell me which details I should confirm with official sources.”
• “Summarize the main arguments for and against a proposed community policy, distinguish facts from opinions, and identify what evidence is missing.”

5. What AI can—and cannot—do reliably

AI can help you learn unfamiliar subjects, brainstorm ideas, draft and revise writing, translate common phrases, prepare questions, organize notes, and explore alternatives. It can be particularly useful for creating a first draft or explaining something at different levels of difficulty. You remain responsible for deciding whether its output is suitable.

AI cannot guarantee that every fact, calculation, quotation, legal interpretation, diagnosis, or recommendation is correct. It may not know about recent events; it may misunderstand sarcasm, context, local rules, or an unusual personal situation. Even when an assistant provides a source link, the source might be irrelevant, outdated, or nonexistent. Check the original source yourself.

AI should not be treated as a substitute for a clinician, attorney, financial adviser, emergency service, or other qualified professional. It also cannot promise that information you enter will remain private: data handling depends on the provider, account type, and settings. An AI that can act through connected tools should be given only the permissions needed for the task, and important actions should require your review.

6. Ten key risks to recognize

• Incorrect answers: AI can produce plausible but false facts, calculations, or explanations.
• Fabricated sources: It may invent citations, quotations, web pages, or supporting evidence.
• Bias and unfairness: Outputs may reflect stereotypes, incomplete data, or uneven treatment.
• Privacy exposure: Entering personal details can disclose information to a service or connected tool.
• Fraud and deception: Criminals can use AI-generated messages, voices, and images to impersonate others.
• Misleading media: Synthetic images, audio, and video can make fabricated events appear authentic.
• Outdated or incomplete information: An assistant may miss current laws, prices, safety advice, or local requirements.
• Overreliance: Users may accept a confident answer without applying independent judgment.
• Unsafe connected actions: An agent with access to email, files, or payments may take an unintended action.
• Malicious content and links: AI-generated text or retrieved material may contain phishing links, harmful instructions, or attempts to manipulate the assistant.

These risks are explored in greater depth in Modules 102–107, including privacy, scams, misinformation, vulnerable audiences, connected assistants, and personal safety planning.

7. Practical actions for safer AI use

Begin with low-consequence activities. Use AI to draft a grocery list, explain a public term, or brainstorm questions before relying on it for anything consequential. As the stakes increase, increase the level of checking. A recipe idea needs less verification than advice about medication, investments, or a legal deadline.

Before entering information, ask whether the assistant truly needs it. Remove names, addresses, birthdays, financial identifiers, health details, passwords, and other sensitive information whenever possible. Review the service’s privacy controls, chat retention settings, and connected-app permissions. Never paste a one-time login code or recovery phrase into a chat.

After receiving an answer, distinguish suggestions from verified facts. For important claims, consult official government sites, original documents, trusted publications, or a qualified professional. Open links independently and verify that a site is legitimate. Ask the assistant to state uncertainties and assumptions, but do not rely solely on the assistant to audit itself.

Finally, stay in control of actions. Review generated emails before sending them, inspect files before downloading or sharing them, and confirm recipients and amounts before purchases or payments. Use strong, unique passwords and multifactor authentication for the accounts connected to AI services. Disconnect integrations you no longer need.

8. Everyday examples for individual and home users

AI is most helpful when it reduces routine work while leaving decisions with the person using it. The following situations show how to use it as an assistant without giving away unnecessary information or treating the output as final authority.

• Household maintenance: Ask for a seasonal home-maintenance checklist, then check appliance manuals and local safety guidance before attempting repairs.
• Learning a new subject: Ask for a plain-language explanation of a scientific or historical topic, then compare key facts with a library or university source.
• Planning a purchase: Ask which product features to compare; independently confirm current prices, warranties, and return policies.
• Preparing for an appointment: Ask for general questions to discuss with a clinician, attorney, or financial professional without submitting sensitive records.
• Writing everyday correspondence: Draft a courteous email to a landlord or service company using placeholders, then check the wording and recipient before sending.

9. End-of-module safety checklist

• I can explain in simple terms what AI is and why its answers may be wrong.
• I understand that an LLM generates language and is not a guaranteed source of truth.
• I write clear prompts that specify the task and desired format without unnecessary personal details.
• I never enter passwords, passkeys, one-time security codes, or account-recovery phrases into AI chats.
• I avoid sharing account numbers, identity documents, medical records, or private family information unless there is a justified, trusted, authorized need.
• I use unique passwords and multifactor authentication to protect my AI and connected accounts.
• I verify important facts, quotations, calculations, and links using independent authoritative sources.
• I recognize that AI-generated voices, images, and messages can be used for scams or impersonation.
• I review permissions and require confirmation before an AI assistant sends messages, shares files, or spends money.
• I keep human judgment—and appropriate professional advice—in charge of health, legal, financial, and safety decisions.

10. Further reading and trusted references

1. United States — NIST AI Risk Management Framework — A practical framework for understanding and managing AI risks.
2. — Guidance on risks particularly relevant to generative AI.
3. Europe — European Commission, AI literacy and the AI Act — European policy and educational context for understanding AI.
4. Japan — AI Safety Institute (AISI), Information-technology Promotion Agency — Japanese AI safety resources and institutional initiatives.
5. International — OECD AI Principles — International principles for trustworthy AI.

Copyright and Use

© 2026 Institute for AI Governance, Cybersecurity & Risk Management (IAIGCRM). All rights reserved.

These educational materials are provided for individual learning and personal reference. Except as permitted by applicable law, reproduction, redistribution, adaptation, commercial use, or incorporation into third-party training programs requires prior written permission.

For licensing and permissions, contact: stephen.buerle.phd@gmail.com.

Educational Disclaimer

These materials provide general educational information about artificial intelligence safety, privacy, cybersecurity, and responsible use. They do not constitute individualized legal, medical, financial, or professional cybersecurity advice.

Although reasonable care has been taken in preparing these materials, technologies, threats, regulations, and recommended practices may change. The Institute does not guarantee that the materials are complete, error-free, or suitable for every situation.

Readers should independently verify information relevant to consequential decisions and consult appropriately qualified professionals when necessary.