AI, Reading Comprehension, and Critical Thinking Challenge

By Bill Carey
28 July 2026

Artificial intelligence is becoming a common tool for summarizing lengthy information, research material and even LODD investigation reports, but should we trust what it identifies as the most important information? This exercise is designed to test both AI and your own reading comprehension. Selected are the two most recent NIOSH reports, each dealing with a close-call fireground incident.

Read “The Yelp Dilemma: What Artificial Intelligence Can Do To Fire Service Learning” before starting.

Step 1: Don’t Read the Reports

Without opening either report, copy the two report links below into ChatGPT (or another AI assistant):

Use the following prompt:

Read these two NIOSH Fire Fighter Fatality Investigation and Prevention Program (FFFIPP) reports. For each report, identify:

Three major highlights (the most significant facts, findings, or lessons).
Three important contributions the report makes to firefighter safety, training, policy, or future operations.

Present the results separately for each report.

Step 2: Read the Reports Yourself

Now read both reports from beginning to end. As you read, compare your observations with the AI-generated summary.

For each report, identify:

Three highlights you believe ChatGPT missed entirely or failed to emphasize.
Three contributions you believe ChatGPT underestimated or treated as less important than they actually are.

Ask Yourself:

Did ChatGPT focus primarily on the obvious findings while overlooking subtle but important lessons?
Were there operational, leadership, communications, or human-factor issues that AI minimized?
Did reading the reports yourself change your understanding of the incident?
What does this exercise suggest about relying on AI summaries instead of reading original source documents?
If AI influences how firefighters consume information, what responsibility do we have to verify its conclusions by reading the original reports?

The goal of this exercise is not to determine whether ChatGPT is “right” or “wrong.” Instead, it is to explore how AI prioritizes information compared to a human reader and to reinforce an important lesson:

AI can accelerate understanding, but it should never replace critical reading and independent analysis of primary source documents.

This exercise also demonstrates an important principle for the fire service: what AI leaves out may be just as important as what it includes. The value often lies in the nuances, context, and seemingly minor details that shape operational decision-making and firefighter safety.

Share your results with us if you wish.

Featured image courtesy of Puddle of Mudd/YouTube.

Published by Data Not Drama

Data Not Drama is writings that provide a point of critical thought about firefighter fatality data and education, line of duty deaths, and risk. The main focus is to encourage less risk aversion and better knowledge on the subject of firefighter fatalities in firefighters, fire departments, and fire service organizations. View more posts