Back to Lessons
beginner10 min10 min read

AI Ethics & Hallucinations: Trusting (but verifying) AI

Understand the limitations of AI, avoid common pitfalls, and maintain data privacy.

What you will learn

  • Identify AI hallucinations and when they commonly occur
  • Apply the Verification Loop to any AI-generated content
  • Understand data privacy risks with public AI tools
  • Establish safe AI usage policies for your team

As powerful as AI is, it is not a "truth engine." It is a probabilistic model that predicts the next most likely word or pixel. This means it can, and will, make mistakes.

What is a Hallucination?

A hallucination occurs when an AI confidently states a fact that is incorrect. This often happens because the AI is optimized to be helpful and conversational, leading it to "fill in the gaps" when it doesn't have specific data.

How to spot them:

  • Check Citations: AI often makes up realistic-sounding URLs or book titles.
  • Math & Logic: LLMs still struggle with complex arithmetic and multi-step logic.
  • Recent Events: If the model hasn't been updated recently, it may guess at current events.

Data Privacy 101

Unless you are using an Enterprise or "Team" version of an AI tool, your data might be used to train future versions of the model.

Rule of Thumb: Never paste sensitive customer data, trade secrets, or personal identification numbers (SSNs, passwords) into a public AI chat.

Real-World Workflow: The Verification Loop

Always treat AI output as a draft, not a finished product.

  1. Generate: Ask the AI to draft the content.
  2. Review: Look for claims that seem too specific or "too perfect."
  3. Verify: Use a search engine or internal documents to confirm key facts.
  4. Refine: Correct the AI and ask it to rewrite if necessary.
SafetyData PrivacyHallucinations
🤖

Almost Done!

Made it to the end — nice work. Record your achievements to update your smart-assistant profile.

Scroll progress: 0% • Finish reading down to complete.