Join us in person in Schulenberg, TX, for a casual lunch! 

When: Saturday, May 18th, at 1:00 pm
Where: Oakridge Smokehouse 712 N US Highway 77 Schulenburg, TX 78956

BYOL (Buy your own lunch) – STC SoCenTx will be purchasing nonalcoholic pitcher drinks and appetizers for attendees.

Other sights you may wish to check out in/near Schulenberg:

Stanzel Model Aircraft Museum

Visitors can stop by the museum during any of the regular hours listed below.
Mon: 10:30am – 4:30pm
Wed: 10:30am – 4:30pm
Fri: 10:30am – 4:30pm
Sat: 10:30am – 4:30pm
Admission is FREE for all people of all ages!!

Painted Churches Tour
*Some churches require pre-scheduled tour groups. See website.

Antiques shopping

May 21 Program Meeting

Topic: How to Communicate Effectively at Work

Join us May 21st @ 7:00 pm online

Speaker: Royce Cook

Are you frustrated at work because people don’t seem to understand your idea? Perhaps they’re confused because the idea wasn’t expressed clearly. Luckily, learning to communicate effectively is a skill you can learn! Learn the dos and don’ts from expert speaker, STC SoCenTX’s own Royce Cook!

Image by Tung Lam from Pixabay

June Gathering

We have a few things to celebrate, so we’ll have a last gathering before our summer hiatus in June. We’ll play game and celebrate winning bronze for the Community Achievement Awards!

Stay tuned for details!

Image by Tung Lam from Pixabay

Don’t forget to renew your membership!

Before your summer adventures begin, renew your STC membership!

ChatGPT: A Powerful Partner in Technical Writing Part 1

by Noel Atzmiller

In mid-December 2023, I attended a virtual webinar sponsored by the Professional and Technical Writing Programs at George Mason University (GMU). Kathryn Meeks, a PhD Candidate in Writing and Rhetoric at GMU, gave a 45-minute presentation titled “AI and Technical Writing”. Meeks focused on ChatGPT and discussed several key topics about it during the presentation. Here is my summary of these topics and my comments.

Introduction to Generative AI and ChatGPT: Generative AI relies on a process called Machine Learning (ML). ML uses algorithms to learn from data and make predictions or decisions. Large Language Models (LLMs) are advanced models that are trained from various types of data to generate text, producing original output that resembles human-created content. ChatGPT is an LLM trained on a large quantity of data from the Internet. This LLM then facilitates conversational interaction between the human user and itself to perform a variety of language-based tasks. Currently, two versions of ChatGPT are available: version 3.5 (which is free) and version 4 (which is not). Meeks recognized that version 4 offers powerful capabilities but stated, “Lots of things can be done in version 3.5.” 

Meeks conveyed this information on several well-designed and attractive slides. It was obvious to me that she was extremely knowledgeable in the topic of AI, yet she explained it in a clear, measured manner without being condescending.

Best Practices for Using ChatGPT: Meeks stressed the importance of understanding the limitations of ChatGPT. For example, ChatGPT can provide substantial information on many topics, but the user would be wise to ensure accuracy by verifying it from trusted sources. ChatGPT is constantly receiving new “training” and its knowledge is increasing, she said. However, Meeks also stated candidly that ChatGPT will sometimes provide false information (called “hallucinating”) if it does not know the answer to your request.

Meeks recommended that users treat their experience with ChatGPT like an online “chat” using a method called  “chain prompting.” When chain prompting, a user comments on each ChatGPT response by making requests for minor corrections, suggesting alternative content, or asking for different output. These refinements may assist ChatGPT to produce more acceptable responses. In other words, ChatGPT “learns” from the corrections and suggestions. which may lead to more acceptable output. 

Meeks also encouraged users to practice using ChatGPT regularly to improve their own prompting skills. The more adept a user is at prompting, the more accurate ChatGPT’s responses can be. 

Prompt Optimization: Interacting with ChatGPT can be more efficient by giving instructions or requests in steps and by being clear and specific with these requests. Instructing ChatGPT to ask questions can also result in more useful responses, according to Meeks. She mentioned that using imperatives when requesting a response from ChatGPT is good practice. Examples include “analyze the following…,” “provide the information…,” “compare the data between…” 

Ethical Considerations and Responsible Use: Meeks strongly advocated that ChatGPT users provide transparency of the information by declaring it came from AI. She expressed that honesty about declaring the information source was good technical writing practice. I thought this was a good suggestion, although I doubt if most writers will proclaim this.

She also stated that users should ensure the information provided by ChatGPT is original to avoid inadvertent plagiarism. I thought this was puzzling.Technically, all AI writing could be considered plagiarism because it is created from algorithms that extract information from human-written texts. Rather than responding so literally, though, I agree with her recommendation. 

Meeks also encouraged AI users to make certain the information from ChatGPT does not contain biases and stereotypes. She related that much of the information on the Internet used to train ChatGPT contains biases and stereotypes. Consequently, the user must be ready to edit out these characteristics from AI-generated content.

She recognized that AI in general (and ChatGPT, in particular) can be intimidating to some people. Meeks stressed that users should regard this technology as a “powerful partner” that can augment or complement human judgment. 

For more information about future GMU virtual webinars, contact:

Heidi Y. Lawrence, PhD 
Associate Professor, English
Director, Professional and Technical Writing Graduate Programs
George Mason University

ChatGPT 3.5 and 4: What’s the Difference?

The list below relies on information from Jon Martindale’s article on (28 April 2023):

ChatGPT 4

  • trained on newer information
  • considers more parameters when composing a response to a prompt (ChatGPT 3.5 may consider up to 175 parameters!)
  • comprehends images: show it an image, then watch as it composes a description

Resource: GPT-4 vs. GPT-3.5: how much difference is there?

Questions about AI? Read “Revolution or Evolution? AI and the Work of Technical Communicators” by Saul Carliner | Fellow and Samira Karim in the March/April edition of Intercom,

2024 PDF Downloads

TIM Talks

What are TIMTalks? Listen to STC President, Tim Esposito, with new STC Executive Director, Tim Shaw, in preparation for their TIMTalks webinars.

The LavaCon Conference on Content Strategy 

and Technical Communication Management

Leave a Reply

Your email address will not be published. Required fields are marked *