5 tips for using AI in technical writing 

At 3di, we love keeping abreast of cutting-edge technology. Not only do we love new tech, it also helps us develop and improve the services we can offer our customers. So, when AI exploded onto the scene, we started looking into using AI tools in our technical writing.

Although AI has been around since at least the 1950s when Alan Turing wrote his paper “Computing Machinery and Intelligence”, its now mainstream use is recent. Since OpenAI unleashed ChatGPT on an unsuspecting world, we’ve seen a deluge of AI tools generating anything from text and automated email responses, all the way to complex graphics and music.

Such pacey developments have led to both excitement and panic. Worries around copyright, job replacement, and the spread of misinformation have borne so-called AI anxiety. In the technical writing world, there have been understandable concerns that AI could threaten the craft of technical writing. At 3di, we are taking a balanced approach, looking at how we can add AI to our tool stack while keeping in mind that AI is not infallible.

So, what did we find out? Well, read on to learn about our five tips for using AI in technical writing. 

1. Saving time

Life as a technical author can involve a fair bit of repetition. How many times have you typed the same or similar input when creating or editing documentation, writing emails, or logging time? Time spent on repetitive tasks can add up. Recently, a colleague of mine recommended a handy little AI tool called Text Blaze that allows you to create snippets of text for reuse. This will be a familiar concept if you use MadCap Flare – snippets are a powerful way to reuse content. Text Blaze allows you to insert saved snippets with an assigned shortcut – great for reducing the time you spend on typing those repetitive bits of text.

Another time-intensive task is searching for specific information in a large amount of text. You know the scenario – you’ve been assigned a task and need to quickly get up to speed. With project management tools like Atlassian’s Jira, that can mean not only reading through a description, but also a stream of comments. The good news is that tools like Jira now have a ‘Summarize’ function that uses AI to analyse text and extract key information using it to provide a concise summary. We’ve found this a great way to enhance both efficiency and decision-making. Although it may not always find the information you’re after, it’s certainly worth trying out.

2. Sorting information

Finding and sorting information can be a laborious job. For example, you may have collated a long list of terms and want to add the definitions, or perhaps you want to identify commonly used terms in a large document, and then define those terms. Sure, you can use a search function to find the terms you know are used, add them to a spreadsheet and sort them. But that’s only part of the job – you also need to source and add the definitions. And what about the terms in your document you didn’t think of?

Well, AI can be a big help here. Depending on the AI tool you use, you can upload a document and ask it to find the common terms and supply the definitions. You can also ask for the terms to be ordered, for example, by requesting a table in which it adds the terms alphabetically along with their respective definitions. Providing context is important here though – many acronyms, for instance, have multiple meanings. If you don’t provide specifics, the AI tool won’t necessarily know which definition you need and will probably offer the most used one – in technical writing, a HAT is not something you wear on your head. 

Something to be aware of is that not all AI tools allow you to upload documents for free, and there may be file size limits. It’s also essential that you check the Terms of Use for any AI tool you use to understand how the information you upload is used. 

3. Creating content

Content creation is a feature many AI tools advertise – from website copy and marketing documents to social media and blog posts (and before you ask, no, AI wasn’t used to write this blog post). But what about creating technical content?

Well, the very nature of technical content doesn’t lend itself easily to AI creation. Technical writing must accurately convey complex information in a clear, concise way, and to do that, technical authors must have an intimate knowledge of the subject matter. If you are creating content that requires a subject matter expert to demonstrate complex functionality in bespoke software, AI can’t really help with drafting a procedure.

Where it can be useful is for drafting topics that need an overview of general functionality. For example, you can ask AI to summarise a marketing document for a topic about a new feature, or maybe you have lots of bits of information that you want to combine into a coherent document. Of course, if you, like us, use an (award-winning) style guide to ensure quality, the AI-generated text will only ever be a rough draft that needs rewriting using your own style conventions.

Read about the award we won for our internal Style Guide here. 

4. Enhancing content

Experienced technical authors should really be their own content enhancers, but sometimes we all need a little help. You may be new to technical authoring and have written a topic you aren’t happy with, or perhaps a reviewer asks you to rewrite some content, and you just can’t get started.

AI tools can look at your content and suggest additional text that might add value or make improvements based on the context. For example, AI can help you expand on certain points, give more examples, or even suggest a different approach. Microsoft’s Notebook function in Copilot also gives you the power to iteratively improve your content. Rather than just rewriting the text or generating completely new text when you edit the prompt, Notebook lets you change your prompt without losing your existing input.

AI can also help if you need to make sure your content is optimised for search engines. Feed an AI tool your content and it can suggest relevant keywords and the best places for them in your text, generate meta descriptions, and suggest relevant links.

5. Reviewing content

Having your technical writing reviewed by a peer is a crucial part of producing high-quality documentation. Authors at 3di are, first and foremost, their own peer reviewers. The content we send for review must be free of typos and conform to style conventions. However, no writer’s attention to detail is perfect, not even that of a technical author. Proofreading your own work is challenging, not least because our eyes tend to see what we think is there. How many times have you checked your work for errors, only for a reviewer to point out a glaring typo you didn’t spot?

AI can be useful here. Yes, there are many spell-checking programs available, but as anyone who has used Word or Grammarly knows nuanced errors can be missed and incorrect suggestions offered. From what we’ve found so far, AI tools tend to perform better at proofing content, plus they have the added benefit of suggesting some useful improvements. A note of caution though – some AI tools feel the need to correct your writing without actually telling you. This can be an issue where deliberately used plain English words are replaced or subject-matter-specific terms are changed. As any technical author knows, subject matter knowledge is essential, and that means AI tools won’t be replacing our human peer reviews any time soon. But if you do use AI to give your content a once over before sending it for review, the key is to be as explicit as possible with your prompts.

We hope you found our tips useful. We’ve found that AI can help with saving time and sorting information, and with some technical writing tasks like drafting and reviewing content. However, because technical writing demands accuracy and understanding of complex subject matter, it’s essential to balance the use of AI tools with human oversight.

*Please note that while all the images in this article were generated using AI, the article was written by our very human Technical Author, Sarah Farnborough.