Operational Procedures · Topic 9 of 9
Artificial Intelligence Basics
Learn how AI is built into applications, what an acceptable use policy covers, the limits of AI output, and how private and public AI services differ.
- 4 min
- AI is built into applications as copilots, chat assistants, search and document tools.
- Policy defines approved tools and appropriate use, with human review of output.
- Never put confidential, personal or regulated data into a public AI tool.
What it is
Artificial intelligence (AI) here means generative AI tools that produce text, images, code or summaries from a typed request, called a prompt. They are trained on very large amounts of data and predict a likely answer. They do not look up facts the way a database does.
For the exam, three questions matter. How is AI built into the software people use, what rules should govern its use, and what can go wrong with it.
How it works
Application integration
AI now appears inside everyday applications. Examples are copilots in office suites, chat assistants in browsers and help desk tools, AI assisted search, and features that summarize documents or meetings, draft email or suggest code.
Some features are switched on by the vendor and others by an administrator. A technician should know what is enabled, what data each feature can reach, and where the setting lives, because an AI feature can read whatever the user can open.
Policy
A company policy sets the rules for AI. It usually sits beside the acceptable use policy.
- Appropriate use. The policy lists the approved tools and the allowed tasks, such as drafting or summarizing. A person must review the output before it is used, and confidential data must stay out of prompts.
- Plagiarism and copyright. AI output may copy or closely resemble existing work. Presenting it as your own, or publishing it without checking, can break copyright or academic and company rules. Cite sources and verify them.
Limitations
| Limitation | What it means | What to do |
|---|---|---|
| Bias | Skewed or incomplete training data produces unfair or one sided output | Review results for fairness and do not rely on AI alone for decisions about people |
| Hallucinations | The tool states something false or invented, including fake sources, in a confident tone | Verify every claim and reference |
| Accuracy | Answers can be outdated, incomplete or simply wrong | Check against trusted, current sources |
A fluent, confident answer is not evidence that it is correct. Treat AI output as a draft that a person must check.
Private and public AI
| Aspect | Public AI service | Private AI service |
|---|---|---|
| Where it runs | Provider's shared cloud, open to anyone | Inside the company or in a dedicated, company controlled environment |
| Data security | Prompts may be stored, reviewed or used to improve the model | The company controls storage and access |
| Data source | Trained on broad data, often from the public internet | Can be grounded on the company's own approved documents |
| Data privacy | Risk if personal or regulated data is entered | Easier to meet privacy and compliance rules |
Data security. Whatever is typed into a public tool leaves the organization. It may be kept by the provider and could appear in later results for others.
Data source. Ask where the training and reference data came from and whether its licence allows the use. The source affects accuracy, bias and copyright risk.
Data privacy. Do not enter personally identifiable information (PII), health data, payment data, passwords or trade secrets into a public tool. Privacy laws and company rules still apply when the data goes into a prompt.
If a scenario has an employee pasting customer data or source code into a public chatbot, the issue is data leakage. The fix is to use the approved private tool and follow the policy.
What to remember for the exam
- AI is built into applications as copilots, chat assistants, search and document tools.
- Policy defines approved tools and appropriate use, with human review of output.
- Never put confidential, personal or regulated data into a public AI tool.
- Bias comes from the training data. Hallucinations are confident false output. Accuracy can be poor or out of date.
- Verify AI output against trusted sources, and cite where it came from.
- AI generated content can raise plagiarism and copyright problems.
- Private AI keeps data under company control. Public AI may store and reuse prompts.
Common mistakes
Trusting the answer because it sounds sure. Hallucinations are written as confidently as facts.
Pasting real data to get a quick fix. Customer records, logs with names and source code do not belong in a public tool.
Submitting AI text as your own work. It may copy existing work and breaks many policies.
Assuming private means perfect. A private tool still gives biased or wrong output and still needs review.
Check yourself
What is a hallucination?
Output that is false or invented but presented with confidence.
Why should confidential data not be entered into a public AI tool?
The provider may store the prompt or use it for training, so the data leaves company control.
What should a user do before relying on AI output?
Verify it against trusted sources, because it may be biased, outdated or wrong.
Topic 9 of 9 in Operational Procedures
Test yourself on Operational Procedures
The full CompTIA A+ 220-1202 question bank has 835 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by CompTIA. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.