By now, you’ve seen that AI can do almost anything. However, because AI “learns” from the information we give it, we have to be careful about what we share. In this final lesson, we will cover the “Rules of the Road” for the AI era.

1. The “Public Park” Rule

Think of a public AI (like the free version of ChatGPT or Gemini) like a public park. Anything you say there could theoretically be seen or used by the “park owners” (the AI companies) to train future versions of their models.

Never upload:

• Personal Identity Info: Your Social Security number, home address, or passport photos.

• Corporate Secrets: Unreleased product plans, internal financial spreadsheets, or private client lists.

• Passwords: Never ask an AI to “keep track” of your passwords for you.

2. Fact-Checking: Be the “Human in the Loop”

AI models are “Probabilistic,” not “Deterministic.” This means they guess the next most likely word. Sometimes, they guess wrong. This is called a Hallucination.

How to stay safe:

• The “Trust but Verify” Method: If an AI gives you a legal statute, a medical dose, or a historical date, do a quick 30-second Google search to confirm it.

• Ask for Sources: Use tools like Perplexity AI that provide citations (links) for every claim they make.

3. Ethics: Is using AI “Cheating”?

This is a common question. The answer depends on Transparency.

• In the Workplace: Using AI to draft an email is usually seen as “productivity.” Using AI to write an entire report and claiming you did it all by yourself without telling your boss can be seen as “dishonesty.”

• In Education: Always check your institution’s AI policy. Using AI as a tutor is learning; using AI to write your essay is plagiarism.

• The Golden Rule: Use AI to enhance your work, not to bypass your thinking.

4. How to “Opt-Out” of Data Training

Most AI companies allow you to turn off “Chat History & Training.” If you turn this off, your conversations won’t be used to train their models.

• ChatGPT: Go to Settings > Data Controls > Turn off “Chat History & Training.”

• Claude: Anthropic (the makers of Claude) generally does not train on your data by default on their paid/professional tiers, but it’s always good to check your “Privacy Settings.”