Are You Letting AI Sneak False Facts Into Your Work?

AI sounds confident, even inventing facts, filling gaps with plausible studies, organisations, and stats, often unnoticed until called out. This RAM shifts the approach: think of AI as a map or draft, and you as the destination or verifier. Use Research Mode—a super-prompt that replaces guesses with ‘Unknown’—and verify claims with 2+ sources from Google Scholar or .gov sites. Check sources for date, author credentials, citations, and bias, then tag findings as Confirmed, Unclear, or Likely Wrong. By 10 minutes, no hallucinations, ensuring your research is reliable and your credibility strong. 

By the end of this micro-course, you will be able to:

• Use the “Research Mode” Super-Prompt to force ChatGPT, Claude, or Gemini to label facts vs. inferences.
• Apply the Triangulation method to verify each key claim using the 4-step source-checking checklist.
• Tag findings as Confirmed, Unclear, or Likely Wrong with the 3-point system—ending every research session with three checked sources and zero made-up facts in your work.

CONTENTS (RAM PDF)

Section 1 – Understanding and Detecting Hallucinations

Don’t Get Tricked by AI: The Confidence Problem

Understand why AI sounds certain even when wrong—making up studies, fake experts, false statistics—and how the problem is in how we use it, not the AI itself.

The “Map vs. Destination” Mindset: The 80/20 Rule

Learn to let AI do 80% of the work (summarising, outlining) while you verify the critical 20% (fact-checking, accuracy)—using AI as an initial guide, never the sole source of truth.

The Triangulation Method: Verify Every Claim

Master systematic verification against multiple trusted sources using the 4-step checklist.

The “Bullshit Detector” Test: Spotting Warning Signs

Practice identifying hallucinations with real examples.

Section 2 – Your Trust But Verify System

The 3-Point Tagging System: Label Your Findings

Use Confirmed (two reputable sources found), Unclear (mixed evidence, use with caution), Likely Wrong (no evidence or contradiction, remove)—turning fact-checking into a clear, repeatable method.

The 4-Step Source-Checking Checklist

Apply quick verification checks to boost confidence in every source.

Your Ready-to-Use “Research Mode” Super-Prompt

Copy-paste two complete prompt templates forcing AI into Research Mode (not story mode) with safety rules, [Fact] vs. [Inference] labelling, structured output, and suggested source types—works with ChatGPT, Claude, Gemini.