AI in Technical Communication & Writing

This library research guide will help you find resources on AI applications, databases to locate credible sources, tips for using AI as a writing partner, and guidance on addressing the critical ethical questions your assignment requires. Use the tabs on the left to navigate through the key steps of the research and creation process.

What is Artificial Intelligence in Technical Communication?

Before you can explain AI to your audience, it helps to have a solid grasp of the basics. In technical communication, artificial intelligence isn't just about robot writers. It's a collection of tools that can augment and streamline the work of a human communicator. Our field uses many forms of AI from general large language models to data processing to visual creation.

Definition, IBM Glossary Topics

"Artificial intelligence (AI) is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy."

Since the 1950s, innovations in machine learning (training AI systems with historical data) and deep learning (mimicking human brain functions to connect layers of data) have layered to build modern Generative AI that creates original outputs based on trained data.

Some key areas where AI is making an impact, which could serve as examples for your flyer, include:

• Content Generation & Augmentation: Tools that can help draft emails, reports, or documentation from a simple prompt. They can also summarize long articles or rephrase content for different audiences. Think of Copilot in your Outlook and Word files from Microsoft. We also see GenAI integration into content managers such as MadCap.
• User Assistance & Support: Intelligent chatbots and virtual assistants that provide instant, 24/7 support by drawing from knowledge bases created by technical communicators. Think of the chat widgets that pop up on almost every service to help you update your account or find a help document.
• Content Analysis & Usability: AI can analyze vast amounts of user feedback to identify common problems, assess the readability of your writing, or even predict where users might struggle with a document or interface. Companies often hire data engineers and machine learning consultants to build custom models to handle their data and analysis needs.
• Translation & Localization: AI and machine assisted translation has helped international writers take their content into new markets and serve global locales by starting with direct word-for-word translations. Then technical writers and consultants from those cultural areas edit, refine, and revise the work. Think of Google Translate...but larger and integrated directly into our localization project management software.
• Multimedia & Design: AI tools can generate images from text descriptions and transcribe audio and video, making content more accessible. Think of the auto-captions on YouTube or TikTok; we also have more advanced models embedded into Adobe products and other industry-grade tools.

The Research Assistant from the Library is NOT an AI-bot: the librarians are actually chatting with you! You can ask us for help locating resources, refining a research question, and digging into your reference list.

The Basics of Large Language Models for Writers

When you hear about AI in writing, you're most often hearing about Large Language Models, or LLMs. These are the engines that power tools like ChatGPT, Google Gemini, and others. The simplest way to think of an LLM is as a highly advanced (like, super advanced based on deep learning algorithms) version of the autocomplete or predictive text on your phone.

At its core, an LLM works by predicting the next most probable word in a sentence based on the patterns it learned from massive amounts of text data. By stringing these predictions together one word at a time, it can perform impressive tasks relevant to technical communicators:

• Generate coherent paragraphs and entire documents
• Answer questions in a conversational way
• Summarize long texts
• Translate languages

Understanding this core function is key. When you write a prompt, you are giving the LLM a starting sequence of words. Its response is a statistically probable completion of that text, and that probability can be adjusted by the "temperature" or degree of randomness built into the model. This is why the quality of your prompt is so important—a detailed prompt gives the model a better path to follow to generate a useful and relevant response for your specific task.

The LLMs are not “magic” replacements for writing skills–they still have many limitations–but they are great assets for a writer to produce more and potentially higher-quality work.

A Brief History of LLMS that are important for your field to know:

• The first chatbot was created in the 1950s/60s but was rule-based. So, for a long time, chatbots were hardcoded and more of a keyword and retrieval index.
• In the 80s, stats got more involved in the game to enable speech recognition using models of language patterns. And once they unlocked the math and statistics of language, that created space for computers to really do their thing.
• So for the next few decades, the math folks, computer scientists, and engineers collaborated and figured out Neural Networks to launch the idea of “machine learning” from wider, connected data sets, came up with the Transformer architecture, and ultimately enabled our current generative models.

Locate more information about the development of AI technology using these library databases and sources.