Yeah, I know. You’ve already heard how UX Writers should update their resumes and pivot because replacement is inevitable.
AI can’t replace writers. It’s just not at a point where it can understand all the context and minutiae needed to craft concise, clear text. That said: The danger comes from management thinking that chatbots can replace paid writers. Whether it’s truly better or not doesn’t matter if your CEO’s unwilling to sign a paycheck. You and I know that writing is more than checking for grammar, but other stakeholders need to know that, too.
Instead of treating AI as a foe, I spent the last month seeing how we could join forces. My goal wasn’t just to save myself time; I wanted to prove that 1) ChatGPT is a helpful tool, and 2) it’s a better tool in the hands of a writer.
Note: I used ChatGPT and Bing AI for this experiment.
Use Case #1: Thesaurus and Dictionary The most obvious way for writers to use AI is as a wordsmith assistant. What’s another word for “help”? How can I say this sentence in a more serious tone?
These sorts of prompts got me mixed results. Some were lightbulb moments — why didn’t I think of that — while others were very clearly wrong. One limitation to note is that many bots (like many people) believe that complex writing is good writing, so it’s important to set up standards at the start.
💡Prompts: What’s another way of saying X? Make Y shorter/simpler/more conversational. What words are associated with Z? One danger is that if all UX Writers start using ChatGPT, every product’s going to start sounding very similar. So instead of only providing guidelines like keep it simple, also let it know your basic copy guidelines. What tone of voice are you going for? What terms should be used/avoided? 🔪 Job-killer: AI can come up with more options than I can. It’s also great at content mining, AKA finding terms and connotations that would have taken me a lot longer to find. 😃 Job-saver: AI is only as good as the prompt writer. It got a lot better once I included some of our basic guidelines, but guess what? Those guidelines were originally written by a human (me). Use Case #2: Guidebook As the sole UX Writer at my company, it can be a challenge to not be a bottleneck. I want to check every line of text that goes into our project, but as we expand, it’s just not feasible. I’ve created copy guidelines and offered writing training, both of which showed great results, but I’m the only person who’s fully focused on copy. Translation: things can fall through the cracks. And what if I’m out sick or on vacation? Some premium bot versions allow you to add additional context via links or documents. This can be a huge time-saver when it comes to typing in your prompt, or having back and forth’s to get an intelligible answer. Just as with the first use case, the resources you provide will need to be written by someone with actual UX understanding. 💡Prompts: Write an error message about XYZ following 🔗these content guidelines. Check this paragraph against 🔗these rules. The biggest bonus of this is that instead of double checking every single error message or toast in my product, I can share this resource with designers and PMs so that they have the ability and confidence to get 99% there. No more bottlenecks. 🔪 Job-killer: Could this lead to the bot being referenced more often than me? Maybe. (Though is time to focus on bigger-picture projects really a bad thing?) 😃 Job-saver: AI sometimes “forgets” other information when it’s given something specific as reference. The guidelines fed into it also need to be written and maintained by a professional UX Writer. Use Case #3: Explorer Research isn’t always at the forefront of UX, usually because of the cost and time investments. While I usually try to do competitive analysis, content mining, etc. when writing for a new feature, it doesn’t always happen. If a prompt will at least provide some basic insights, though, I can be more confident about my decisions, or try a different route than I’d previously considered. 💡Prompts: What does company X say about Y? What terms are usually brought up when talking about Z? Note, of course, that AI can hallucinate. This shouldn’t be treated as true research, and it shouldn’t replace actual studies. But it can bring resources and insights to your attention that could have gone missed otherwise. 🔪 Job-killer: Honestly? None. This is bringing in an element that I haven’t had time for in the past. 😃 Job-saver: See above. Use Case #4: Bug Catcher Many products have thousands of keys. Legacy is a b*tch, but it now feels more manageable. The role of “bug catcher” mixes the above prompts together. While I love auditing products to see where we can make improvements, it can be tiring (and sometimes impossible) to find every inconsistency. 💡Prompts: Take a look at this code; which lines do not end in punctation? Give me a list of places where the term “team” is capitalized and lowercase. It saves a lot of time for me to send over a list of keys that need periods added to them, rather than physically combing through every screen or every line of code myself. 🔪 Job-killer: This could cause stakeholders to assume the bot will catch everything, thus leaving me out of the conversation. 😃 Job-saver: When auditing gets done faster, I can focus on bigger, UX-centric improvements.
Get With It Look, I’m not saying that AI is already at a place that can radicalize your day to day. And there are problems with it. But for the month I gave it a try, I found some use cases that saved me both time and headaches. Changes are coming whether we like it or not, and some pivoting is required. Learn how to write better prompts. Optimize documents so AI can use them more effectively. Ask stakeholders what their copy painpoints are. If you’re really worried about your future, why not make your biggest competitor your biggest ally?
Comments