Michael Alley, Penn State
Writing as an Engineer or Scientist
  • Home
    • Scientific Presentations
    • Scientific Posters
    • Scientific Film
  • Writing Topics
    • Writing with Artificial Intelligence >
      • Strong Examples of AI Writing >
        • AI Example: Marketing Analysis
        • AI Example: Report Summary >
          • Full Report
      • Telltale Signs of AI Writing
    • Grammar >
      • Section 1: Grammar
      • Section 2: Punctuation
      • Section 3: Usage
      • Comprehensive Quiz
  • Writing Documents
    • Reports >
      • Reports Tutorial: Short >
        • Reports Tutorial: Full
      • Report Formats >
        • Sample Report Format
        • Report Templates
      • Laboratory Reports
      • Design Reports
    • Emails >
      • Tutorial: Emails
      • Writing Professional Emails
      • Writing Professional Letters
    • Research Papers
  • Teaching Resources
    • Short Films
    • Canvas Quizzes for Tutorials
    • Craft of Scientific Writing >
      • Lessons >
        • 2: Being Precise and Clear
        • 3: Avoiding Ambiguity
        • 4: Sustaining Energy
        • 5: Connecting Your Ideas
        • 6: Being Familiar
        • 9: Emphasizing details
        • 10: Incorporating Illustrations
      • Errata
    • Why Our Students Struggle With Scientific Writing

Strong Example of AI Writing:
Summary of Engineering Report

Michael Alley
August 2025
Beginning in Spring 2023, my team has asked students of an engineering writing course to use AI to create a summary of a technical report that they had written and received feedback on. In general, the summaries that AI created were less effective than the summaries the students (with guidance) created with traditional means.
        The main problem with the AI summaries was not emphasizing the report's most important details for the primary audience (a technical manager interested in the design process). With ChatGPT 3.5, the students gave few technical details. With ChatGPT 4, the students gave many technical details but did not emphasize the most important ones. However, in August 2025, ChatGPT 5.0 was able not only to recognize who the most important audience was but also to create an effective summary for that audience. Moreover, with just three short iterations to match the AI style with the expected style for this document, we reached the quality level of the traditional summary.
Picture

Summary Written by Traditional Means


In-Class Instruction. As a class, we voted on and then discussed the most important details in the report. Then we reviewed a traditional method of writing summaries: highlighting and then copying the most important sentences into a first draft; revising that draft to make the ideas (and sentences) connect; and proofreading.


​
​Summary (Traditional)
​        
To assist owners with service dogs, an engineering course at Pennsylvania State University has asked our team to design a leash lengthener for low dexterity users. This report presents the customer needs and accompanying metrics and specifications that will be considered in our team’s design. The primary customer needs include being accessible and safely retractable. For the need of being accessible, one specification assigned was that the controls be at least 3.3 inches to accommodate users with low dexterity. Additionally, for the need of being safe, one specification assigned was that the leash locks securely at a minimum of 2 feet and a maximum of 3.5 feet. Secondary needs include being portable and durable to ensure ease of use and long-term function. Moving forward in the design process, our team’s next step is to generate initial design concepts and select one of these concepts to prototype.​
​
Analysis: This summary efficiently summarizes the design problem and states the scope of the report. Included then are the two most important needs that the team identified. Also included is a specification for each of those needs. The summary concludes with a secondary result and a precise description of the next step.
Summaries Written with AI (No Iterations)

​
​Author's Prompt:  Please write a summary (no more than 150 words) of the attached report.


​
  

​

​
Summary (ChatGPT 5.0)
    This report outlines initial research for designing a dog leash lengthener for low-dexterity users served by Canine Companions for Independence (CCI). Current CCI leashes, at 3 feet with buckle adjustments, pose safety and usability challenges for individuals with limited finger coordination and strength. The goal is to create a leash that extends from 2 to 3.5 feet, is easy to operate, and prevents unintended extension. Primary needs include accessibility for low dexterity, safe retractability, durability, and portability. Key specifications are: operation with ≤1 pound of force, control size ≥3.3 inches, maximum force resistance of 130 pounds, total weight ≤3 pounds, and compact dimensions (≤4×6×1 inches). Safety concerns with traditional retractable leashes—such as excessive length and delayed locking—inform the design approach. The next phase involves ideation, prototyping, and testing, with ongoing feedback from CCI. The project aims to deliver a functional prototype by May 2025.
​
Strengths: This version gives many technical details from the report. These details would interest the primary audience, who is a technical manager.

Weaknesses: This version has the following problems:
  1. Not clear which needs and specifications are the most important.
  2. Lack of variety in sentence openers--every sentence begins with the subject-noun. This weakness is a tell-tale sign of ChatGPT.
  3. Informal language: "max" and 4X6X1"
  4. Pretentious words:  "accessibility" and "portability"
  5. Imprecision in second to last sentence: The next phase is concept generation and then selection. Prototyping does not occur in the last phase. 

Summary (Traditional)
​        
To assist owners with service dogs, an engineering course at Pennsylvania State University has asked our team to design a leash lengthener for low dexterity users. This report presents the customer needs and accompanying metrics and specifications that will be considered in our team’s design. The primary customer needs include being accessible and safely retractable. For the need of being accessible, one specification assigned was that the controls be at least 3.3 inches to accommodate users with low dexterity. Additionally, for the need of being safe, one specification assigned was that the leash locks securely at a minimum of 2 feet and a maximum of 3.5 feet. Secondary needs include being portable and durable to ensure ease of use and long-term function. Moving forward in the design process, our team’s next step is to generate initial design concepts and select one of these concepts to prototype.​
Analysis. Same traditional summary as above.
Summary ChatGPT 5.0 (Three Iterations)
        To assist owners with service dogs, an engineering course at Pennsylvania State University tasked our team with designing a leash lengthener for users with limited finger coordination and strength. In this report, we present the customer needs along with the metrics and specifications that will guide our design. Primary customer needs are that the leash is to be easy to handle and remain secure at both short and long lengths. For ease of handling, one specification is that the controls are to be at least 3.3 inches wide to suit low-dexterity users. For safety, another specification is that the leash is to lock securely at a minimum of 2 feet and a maximum of 3.5 feet. Secondary needs include a leash that is light to carry and strong enough for long-term use. In the next phase, our team will generate design concepts and select one to prototype.
    
​​Analysis. This AI version is as strong as the human version. Only 15 minutes was needed to produce this version. That time included writing the simple prompt and then the three iterations that revised the prompt to address issues of style noted above.

Leonhard Center, Penn State 
University Park, PA 16802

Content Editor:

Michael Alley
​
[email protected]