Search results

30+ ways I’m using AI in everyday writing life as a technical writer, blogger, and curious human

by Tom Johnson on Dec 4, 2023
categories: ai • writing • podcasts

In this post, I provide over 30 real-life examples of how I'm using AI on a daily basis, not just for technical writing tasks but more broadly in life, including summarizing content, explaining concepts, answering questions, troubleshooting problems, and having engaging conversations for a variety of tasks and scenarios. In my view, AI use cases are ubiquitous, equivalent to the use cases for computers or the Internet in general.

Listen here:

Intro notes, including why I wrote this post

Recently, I’ve noticed many people around me trying to figure out how to use AI. The hype is extremely high and they’re seeing so much innovation and news, but that hype + innovation isn’t reaching actual transformation of their tech doc work. They look around and try to find an expert who has figured it out, because there must be something wrong if they can’t make use of this transformative new technology that everyone keeps raving about, but which hasn’t seemed to change their own workflows. As a result, there’s a lot of discussion about identifying the right use cases for AI.

What are the best use cases for AI? I think this question is somewhat like asking what are the use cases for the computer or Internet. Nearly everything is a use case. The same question was common when these technologies first surfaced, but they quickly gave way to an enormous variety of applications in nearly every domain. This is an argument Ellis Pratt makes in his latest Cherryleaf podcast (episode 141).

In the following 30+ scenarios, I describe various ways that I’m using AI in everyday life. In this post, I decided to expand outside the realm of technical writer scenarios. As for tools, I’ve mostly used ChatGPT, Claude, and Bard.

For a previous presentation on a related topic, see AI and API docs: What works, what doesn’t. Also see the posts in my blog’s AI category.

The following are rough descriptions of 30+ ways I’m using AI. The sections aren’t in any particular order. I’ve resisted grouping them because I want to show how AI can be used pervasively for many tasks, not just limited to technical writing applications. My central argument is that AI use cases are so ubiquitous they apply to all facets of life.

All the ways I'm using AI

1. Create a readable version of a YouTube transcript

Scenario: The YouTube video is 40 minutes long and you don’t want to sit patiently watching it to see if the content is relevant to your needs. You want a quick summary of what topics are covered, and whether they mention anything about the topic you’re interested in.

Prompt: Make a more readable version of this video transcript. Don't eliminate any detail, just make it more grammatical so I can read it. [Paste in the podcast transcript]

2. Understand difficult passages in a book

Scenario: You’re reading a book with some dense philosophy that you don’t understand. Rather than speed through it, you want to deepen your understanding so that you can more thoroughly enjoy the book.

Prompt: I'm reading Robert Pirsig's Zen and the Art of Motorcycle Maintenance and am struggling to understand a difficult passage. Can you explain this paragraph? Include some concrete examples to make the ideas understandable.

Tip: Use the audio app for ChatGPT on your phone so you don’t have to type out the passage. This converts your phone into a reading buddy, helping you quickly interpret passages.

3. Create a custom-tailored learning course

Scenario: You want to learn about something technical, maybe from code you’re documenting or based on another need. In this case, you’re having trouble remembering the right syntax for links in Java files for Javadac. You ask an AI tool to create a custom course for you based on this specific topic.

Prompt: I'm a technical writer documenting a Java API. I want to better understand how to tag the source with comments that get generated as Javadoc. Create a course about Javadoc tag syntax that I can follow. I want you to teach me step by step, like a friend. Start by showing an outline of the main course topics.

4. Summarize the main point of an academic article

Scenario: You’re trying to make your way through a journal article, which is living up its reputation for tediousness and say-nothingness. Even the abstract isn’t clear or useful. You use AI to create a detailed summary of the author’s argument to see if the article is worth reading.

Prompt: Summarize the main argument of the author here. Break down the logic into a crisp construction of the main claim and supporting reasons and evidence. Provide some key points the author makes and other conclusions.

Tip: If it turns out the article is worth reading, this high-level summary will aid your comprehension as you make your way through the academic jargon.

4. Get grammar advice

Scenario: You’re reviewing some documentation and aren’t sure about the word choices someone has made. You’re tempted to suggest an alternative wording, but you want to avoid marking sentences as incorrect if it’s just personal preference and not an actual grammar rule or style. You ask AI to explain the difference between two sentence structures, and referencing any official grammar or stylistic rules/conventions at stake.

Prompt: As a technical editor for a documentation group, you are an expert in grammar and plain language. Provide guidance about the following usage options, indicating which is better and explaining the rule behind it. [Paste in the same sentence in two different sentences for comparison purposes.]

5. Write full-length, human-sounding articles

Scenario: You’re feeling lazy and want to leverage AI to be your writer, while you act more as a director of the ideas. Going paragraph by paragraph, you explain in rough words and ideas what you want the AI to articulate in more refined prose. You don’t try to have AI write the article in one go, as this usually turns out disastrous. Instead, by going slowly, paragraph by paragraph, you keep a fine handle on the reins.

Prompt: You're going to help me write an article for my blog. I'm going to lead you paragraph by paragraph describing what I want you to write. I will be the director and you will be the writer. You will articulate my ideas in readable, grammatical ways, adopting a plain-speaking, direct style. If at any time my ideas are ill-conceived, you will push back and recommend better approaches. Are you ready to begin?

Tip: Use AI to write the summary parts only, such as summaries of an article or plot details from a novel. Don’t have it write about personal experiences. Mix this summary with personal experience in balanced ways.

More info: My experience trying to write original, full-length human-sounding articles using Claude AI

6. Increase interest in the world around you

Scenario: You see something interesting that catches your eye. In this case, it’s a large fishing vessel with the words “Seafreeze - Dutch Harbor, AK” on the side. You want to understand it better, including some more interesting details about it, so you take a picture of it and ask ChatGPT to provide more details.

Prompt: [Take picture of fishing vessel.] A big fishing vessel called Seafreeze Alaska just sailed by, from Dutch Harbor, AK. Tell me about this boat. What kind of fish does it catch? Tell me some interesting facts about it, including why Dutch Harbor is in a place in Alaska ironically called “unalaska.”

7. Troubleshoot using pictures

Scenario: You’re trying to fix something, but you don’t totally understand it. In this case, you’re trying to winterize your sprinklers and you’re not sure what all the valves in the sprinkler box do. You take a picture of the sprinkler valves and ask questions about the purpose of each of the valves.

Prompt: [Take picture of valve box.] I'm trying to winterize my sprinklers by first shutting off the water to my sprinkler system. Look at this image and tell me which valves I need to turn. I see three valves, two blue and one also dark blue but on a white PVC pipe. Pipe. What are all of these valves?

8. Get ideas for what to eat from ingredients in the fridge

Scenario: You’re hungry and not sure what to eat based on looking at your fridge. You take a picture of your fridge’s contents and ask AI for some suggestions on what to eat. Although you don’t follow any of these suggestions, the AI has offered some possibilities you never imagined with your fridge’s contents.

Prompt: [Take picture of open fridge.] What can I eat for dinner?

9. See counterarguments to your position

Scenario: You’re writing a post about a topic, formulating your argument, and you feel pretty confident about your logic. But you want to anticipate objections, so you use AI to better imagine the counterargument to your position.

Prompt: In the following content, I’m arguing for [some topic]. [Paste in content.] I’m feeling pretty good about my argument, but I also want to understand the opposing perspective. What would the counterargument be for a position that argues the opposite? Expand the logic of this counterargument in detail, with supporting logic and reasons.

10. Get images for a post

Scenario: You want to get images to complement the ideas in your post. You paste in a section to ChatGPT and ask it to draw an image reflecting the ideas in the paragraph.

Prompt: Create an image illustrating this principle/concept: [Paste in paragraph or more of content.]

11. Troubleshoot technical issues

Scenario: You’re trying to fix some validation errors in your podcast feed. You’re running your XML feed through a cast validator and it’s indicating you certain errors, but those errors are hard to understand in context of the XML schema’s requirements.

Prompt: Help me fix the itunes feed for my podcast. A validator says "WARNING: Found 313 episodes with an invalid length. The length should be a nonzero value specified in the enclosure tag, and refers to the file length in bytes." Here is the feed. [Paste representative sample of the feed.] How do I fix it?

12. Fix a poorly written paragraph

Scenario: You read a paragraph in your article/documentation/blog and know it sounds off. (It’s wordy, hard to read, awkward, has the wrong diction, etc.). You use AI to see if it can create a more readable version of this problematic paragraph.

Prompt: Make the following content more readable. [Paste in problematic paragraph.]

13. Find research

Scenario: You have an idea for something you want to write about, but you need some solid academic sources to support this idea. You use AI, specifically ChatGPT with Bing, to browse the web and identify some quality sources.

Prompt: I’m trying to find research about [some topic]. Browse the web to find substantial, academic articles that discuss this topic. Avoid lightweight web content and instead find more scholarly articles. Here’s the content I’m looking to support with more scholarly research: [Paste in your existing blog post or article.]

14. Offload any writing you don’t really care about

Scenario: You need to write documentation for a new feature. You aren’t concerned about connotations of plagiarism, cheating, personal authorship, voice, personal authenticity, etc. because this is just a voiceless, plain corporate tech doc, and you want to kick out a first draft.

Prompt: You’re a technical writer creating documentation for external partners. Based on the following content, create a draft of the documentation. Use clear, simple language to relay the ideas…[Paste in engineering documents.]

Scenario: You want to write some code to automate some build and publishing tasks related to API documentation. You’re trying to build docs, move them into a specific directory, insert a conditional note, and more. You write some pseudo code for this, indicate the language you want to use (shell scripts), and ask AI to write the actual code.

Prompt: Help me with some shell scripting for my documentation. I want you to write a shell script that does the following: run the command foo, then capture foo’s output as a variable called foo_output. Using foo_output, pass this into the BAR command as a parameter.

More info: Use cases for AI: Develop build and publishing scripts

16. Distill needed updates from bug threads

Scenario: You need to write release notes about a bug issue and resolution. Engineers have filed a bug that references another bug with a long thread about the issue and resolution taken. There are 75+ comments on the thread and other docs referenced, and you need a quick summary about it.

Prompt: From the following thread and related information, describe the issue and the resolution taken. This information will be included in release notes for documentation.

More info: Use cases for AI: Distill needed updates from bug threads

17. Create article summaries

Scenario: You want to create a short summary paragraph describing an article/topics/post/chapter you’ve written. You’ll use this summary at the top of the article.

Prompt: Write a summary of the following content in a few sentences.

More info: Summarize long content

18. Synthesize insights from granular data

Scenario: You have lots of little pieces of data and you want to synthesize larger insights from the data. In this case, you’re preparing notes for a book club based on a book you’ve read. You have 100 passages highlighted in the book. From these passages, you want to extrapolate major themes and arguments. You also have 10 book reviews identified.

Prompt: Based on the following content, identify some major themes from the book. Highlight the author’s main arguments and supporting reasons. Identify some questions that might be relevant to a book club involving tech workers working at [some tech company or project].

19. Arrange content into information type patterns

Scenario: You have a lot of information you’ve gathered about a new feature, and you want to sort this information into a first draft of documentation. You specifically want the information to fit a predefined pattern or template.

Prompt: You are a technical writer creating documentation, specifically a task. You follow a list of rules and arrange information into a given template. The task you are writing is how to make a basic API call.

Rules: Here are the rules for writing tasks: [paste in rules]

Template: Here is the template for writing tasks: [paste in template, such as from the Good Docs project]

Instruction: Make the following content fit into the task template above.

More info: Arrange content into information type patterns

20. Format content (HTML, XML, YAML)

Scenario: You have a long HTML table with some incorrectly coded <tr> tags or some other error, and you want it fixed. You paste the content into an AI tool and ask it to fix the errors.

Prompt: Something is wrong with the HTML for this table. Find and fix the error.

21. Compare API responses

Scenario: You have an API response for a new API you’re documenting, and you want to see if the response matches all the fields you’re documenting. You take the sample responses and ask AI to compare them against the documentation to identify any discrepancies, taking into consideration optional versus required fields. You also want to identify discrepancies in casing or spelling.

Prompt: The documentation for the OpenWeatherMap Forecast API is as follows: [Paste in documentation of the API]. Here’s a sample response from the API: [Paste in response]. Identify any fields that were present in the documentation but which are not present in the response, and vice versa.

More info: Compare API responses to identify discrepancies

22. Draft glossary definitions

Scenario: You want to build out a glossary for specialized terms. First, you want to identify the terms that might be jargon to readers, and then provide definitions for the terms.

Prompt: From the following PDF, identify all the words that would be good to add to a glossary for a course on API documentation. Provide definitions of each of the terms as well. [Paste in content or upload PDF.]

More info: Create glossary definitions

23. Get quick answers about Clickbait articles

Scenario: You’re reading the New York Times and see some op-ed articles with clickbait titles that you can’t seem to pass up. For example, “The Very Good Reason People Like George Santos Lie About Nonsense” (link).

Prompt: Provide a crisp summary of this article and explain the answer that corresponds to the title. [Paste in article.]

24. Summarize Hacker News threads

Scenario: You see a long Hacker News thread and are curious to know what all the discussion is about, but you don’t want to read through all the thread details.

Prompt: The following is a thread from a Hacker News article. Summarize the main discussion points in the thread. [Paste in Hacker News thread.]

25. Explain something in a simpler way

Scenario: You’re banging your head trying to understand a complicated concept. In this case, you’re struggling with text someone sent you that seems nonsensical, like “This REST API isn’t a resource-based API.” You aren’t sure what it means, so you paste the sentence into AI for an explanation. You ask AI to explain it like you’re a beginner and to include some examples to clarify the meaning.

Prompt: Someone has sent me a chat message saying that “this REST API isn’t a resource-based API.” Explain this to me. I thought all REST APIs consisted of resources? How is this possible? Can you explain it to me in simple terms, and maybe provide some examples to clarify?

26. Convert this proto file and engineering doc into something that resembles partner-facing documentation

Scenario: You need to write documentation for a new feature. There’s a product definition document and some other engineering documents. You use AI to quickly create a first draft of documentation, just to get a better sense of the product. The final doc you publish will likely not resemble anything similar to the AI-generated draft, but that’s okay because this first draft gives you a sense of the general shape and features of the product.

Prompt: You’re a technical writer creating some partner-facing documentation for a feature. The feature is described in this proto file and engineering doc. From the proto file and related engineering doc, create some partner-facing documentation that follows common styles and standards for documentation. [Paste in proto file and engineering doc.]

27. Understanding API example code

Scenario: You need a code sample showing how to use an API for a particular scenario. You request some code from engineers, and they give you something without any explanation. You want to better understand what’s going on in the code so you paste it into an AI tool for an explanation. You learn what each piece does and can get a sense of the logic.

Prompt: Can you explain this code? What is it doing? Break it down and explain each of the individual parts. I’m not an engineer, so make the explanations easy to understand.

28. Find out whether a book is worth reading

Scenario: You’re browsing books on O’Reilly and trying to find something that’s relevant to your interests. The book summaries are too vague to get a real sense of the contents. So you copy the introductory chapter into Claude and ask for a detailed summary. Based on this summary, you get a stronger sense of whether the book is relevant.

Prompt: The following is an introductory chapter of a book I’m considering reading. Provide a detailed summary of the chapter and what the book is about, including whether it would be of interest to someone like me who does technical writing and is interested in techniques for problem solving in technical domains. [Paste in the chapter.]

29. Find supporting quotes or anecdotes from a book

Scenario: You’re writing an article about a book you’ve read, and you want to pick out some quotes to support a theme you’re writing about. You’re not sure where in the 300 page book the quotes appear, but you managed to get ahold of a PDF of the book and have uploaded it into Claude.ai.

Prompt: The following is an upload of the book Zen and the Art of Motorcycle Maintenance. Identify some relevant quotes to a part of the book that talks about the barbecue assembly instructions. I’m trying to stress the point that instructions shouldn’t be viewed as the only path through a process. Find some quotations that support this idea.

30. Ask curious questions while eating rather than scroll news feeds

Scenario: You’re eating lunch and are tired of scrolling through senseless news feeds while you eat. Instead, you decide to have a conversation with ChatGPT, so you ask it a question. Then based on the response, you ask it another question. You’re using the audio button and it appears to others that you’re having a phone conversation of some kind, though it really doesn’t matter because people are either too focused on their own smartphones or are talking with others. You pass the time in an interesting conversation asking questions that take you down rabbit holes you didn’t know you wanted to explore.

Prompt: Can you solve technical problems through meditation? Describe the theory and approach for this.

31. Identify weak parts of a proto file (undocumented fields, weak definitions)

Scenario: You’re documenting a new feature that’s pretty much contained in a proto file. You glance through the proto file and wonder what the content quality is for the proto comments. You ask AI to assess the proto for any undocumented fields, weak definitions, confusing areas, etc. This gives you a starting point for how to assess the reference content.

Prompt: Can you assess the following proto file to identify any weaknesses, such as missing documentation, poor field definitions, or other confusing areas? [Paste in proto file.]

Scenario: You’re trying to solve a complex problem, like how to get your ADHD child to do her homework as well as navigate the right approach between discipline and support. You aren’t sure if disciplinary measures will be interpreted as punitive and backfire for an ADHD mindset, and you also need to find strategies for persuading your spouse about why a punitive approach might backfire. You use AI to express your concerns, talk through solutions, raise objections, explore counterarguments, etc.

Prompt: I’m trying to figure out how to help my daughter, who has ADHD, complete her homework…

33. Get the latest weather or news

Scenario: You want to know what the weather report is for the next week, or maybe what the latest NFL news is for the past couple of days.

Note: As far as I know, this requires ChatGPT with Bing.

Prompt: Browse the web to to get the weather report for this week. / Browse the web to get the latest NFL news for the past 2 days.

Here’s the This Week in Tech (TWIT) podcast episode I mentioned in my video: TWIT episode 956, starting at minute 11:45.

I recently gave an STC webinar on this same topic. See Webinar recording: Experiments and use cases for AI from a tech writer’s perspective for the recording and slides.

About Tom Johnson

Tom Johnson

I'm an API technical writer based in the Seattle area. On this blog, I write about topics related to technical writing and communication — such as software documentation, API documentation, AI, information architecture, content strategy, writing processes, plain language, tech comm careers, and more. Check out my API documentation course if you're looking for more info about documenting APIs. Or see my posts on AI and AI course section for more on the latest in AI and tech comm.

If you're a technical writer and want to keep on top of the latest trends in the tech comm, be sure to subscribe to email updates below. You can also learn more about me or contact me. Finally, note that the opinions I express on my blog are my own points of view, not that of my employer.

© 2024 Tom Johnson