There’s nothing wrong with experimenting with new technology. You learn where the line is between ethical and unethical use. I firmly believe in using technology ethically. It was instilled in me as a kid when I was learning math.
My parents wouldn’t teach me how to use a calculator until I learned how to do mental math. They saw falling math standards in schools which encouraged kids to use a calculator as a crutch.
I started an experiment with ChatGPT a month ago, using the same mindset. As a freelance writer and copywriter, I feel it’s critical to adapt to new writing technology, ethically.
Take a guess: Did I use ChatGPT to write this article?
Read on to find out!
The Challenge
I challenged myself to write a 1,000-word article on a topic that I have heard a lot about but don’t have expert knowledge about. I wanted to see how ChatGPT worked with the whole article process, from research to drafting.
Getting topics that you don’t know a lot about is a fact of life for writers and copywriters, even if you have a specialty. You might get a project that’s outside your specialty, but it’s still worth doing.
I deliberately chose a topic that I didn’t know a lot about. I wanted to remove my own bias. Plus, I already know credible sources for my favorite niches (travel and food). I may try the experiment with ChatGPT again with travel or food, but we’ll see.
I had previously researched my SEO keywords and created a file for them. Then, I figured out my headings for the article. You’ll find out why later.
Note: this article was entirely mine. It was not for a client. If I were doing an article for a client, I would consider the legalities, such as privacy before using ChatGPT.
ChatGPT Tutorials
After picking a topic, I started looking for good ChatGPT tutorials that piqued my curiosity and stood out among the hype. If you are new to ChatGPT, I recommend starting with the following tutorials.
It’s no secret that I absolutely LOVE Lifehack Method’s YouTube Channel! When they posted this video, I started confronting my prejudices about ChatGPT.
6 Actually Useful Ways To Use ChatGPT 🔥 | Lifehack Method
After watching Lifehack Method’s video, I found another excellent tutorial that got into more detailed explanations.
ChatGPT Tutorial — A Crash Course on Chat GPT for Beginners
I hadn’t been on Design with Canva’s YouTube Channel for a long time. When I went back to it, I found step-by-step tutorials for ChatGPT. The tutorials are best for practicing with ChatGPT in real time!
The biggest lesson from the tutorials? You need to know how to prompt ChatGPT to get the results you want. When you prompt ChatGPT, you are priming it to work a certain way. That’s how you make it work in your favor.
On to my experiment!
Brainstorming
There’s a lot of internet noise, and it can feel like wading through mud when doing research. It’s worth trying to see if finding sources on ChatGPT will save you time on a particular project. It doesn’t always save time though.
ChatGPT is not trained on current information yet, so you need to find sources that were published in the past year yourself. Plus, you need to check the legitimacy of ChatGPT-generated sources.
I asked ChatGPT for possible sub-topics to include in the main topic. While I knew the typical sub-topics, ChatGPT pointed out other sub-topics that I didn’t know about.
Then, I asked ChatGPT for a list of 50 research sources. Then, I checked which sites on the list were competitive with each other and whether they were legitimate. I told ChatGPT to get rid of the competitive and bad sources and generate another list.
I repeated the process until I got a good list of 50 sources and then asked ChatGPT to narrow them down to the top 10. ChatGPT can be inconsistent with picking sources, but it’s still worth seeing what it generates.
The sources ChatGPT came up with eliminated a lot of the noise that I would have run into with the research. Even though ChatGPT is not trained on new data, it’s worth using it to summarize the latest data.
Summarizing
I picked out the latest sources I wanted to use, skimmed them, and listed the important points. I tested how well ChatGPT could summarize the sources and whether it agreed with me on deciding the most important points.
I compared ChatGPT’s most important points with mine. The results were quite similar! I believe it had a lot to do with the prompts that I had done before because it learned my writing style.
ChatGPT can stop summarizing before you need it to, so you have to re-prompt it. After using ChatGPT to summarize the sources, I read through them properly.
Taglines, Headings, and Titles
Would I use ChatGPT to generate taglines, headings, and titles? It depends on whether the client gave them to me, or if they came to me intuitively. This time, it was intuitive which title and headers to use, but I was curious to see what ChatGPT generated.
My overall goal in my writing is to keep my reader interested and wanting to read more. Bearing that in mind, I prompted ChatGPT to create 10 potential titles to compare them to my ideas. I did the same with my headers.
Since I know how to prompt ChatGPT to use a certain tone, I got great results! I found some of the ideas were better than mine, and I used them, which is completely okay.
It’s cool to play around with ChatGPT taglines, headers, and titles and compare them with your ideas, but ONLY if it’s your ideas! You don’t want to breach any potential legal contract.
Outlining
After asking ChatGPT to summarize the sources and pick out the main points, I asked it to create an article outline that used my headers, chosen sources, and main points.
The ChatGPT-generated outline provided a good structure when you provide the title and headers, but it was repetitive and didn’t really use the information I had given it before. From there, I was able to fill in the outline with the research I had done before.
The outline didn’t make the points specific, so I had to fill in that. In some ways, that’s a good thing because you can visualize the flow of the article.
Article Draft
Would I normally have ChatGPT write an article before I write it myself? It depends on the situation.
For the sake of experimenting with ChatGPT, I had it write a draft of the article before I started writing. I discovered some merit in doing so.
It’s not at all intuitive where to put the SEO keywords sometimes, so it’s good to get an idea of where to possibly put them before you start writing. It saves a lot of time later.
I looked at the draft and found it delivered a good article structure. For important subheadings, there was a short paragraph and 3–5 bullet points, so I emulated that structure for my article.
I didn’t like how ChatGPT wrote the sentences, so I kept the main ideas, but completely rewrote them. I only kept 10% of what ChatGPT generated. The stuff I kept were things I would have written anyway.
ChatGPT sometimes stops generating in the middle of an article. It’s especially problematic if you specify a word limit. If you want to rewrite a section in a certain style, such as bullet points and a short paragraph, it doesn’t always work.
Spelling, Tone, Grammar, and Keywords
I learned from the tutorials how to prompt ChatGPT to assess the tone of an article. I checked the tone of my article and was pleased with the results.
I didn’t like what ChatGPT did for spelling and grammar checks. I would rather use my own good judgment and double-check it with Grammarly and the Hemingway app.
I tried telling ChatGPT to leave out the Oxford comma because I was deliberately doing the article in AP style. Didn’t work too well because it’s not trained in writing styles, and it told me that.
Not all the SEO keywords I needed were included in the ChatGPT-generated draft. I found it better to stick to how I normally add those keywords.
Will my habits change in the future? Hard to say. I need more practice with knowing how to re-prompt ChatGPT when it stops generating in the middle of a summary, or an article.
When Using Emojis in Your Work 😎
Using emojis in your work is subjective. If you can use them, you don’t want to be scrolling through an entire list of them to find the best one.
Using the right emoji can really pack a punch! ChatGPT is great at finding the best emojis to add to your work! Let it know where you want to add the emojis, and it will come up with some great results!
This feature is especially useful for social media captions, titles, email headers, or anything where you are trying to hook people’s attention.
If you aren’t entirely happy with the emojis, just ask ChatGPT to generate them again until you are happy with them.
Limitations
I learned a lot about the art of prompting ChatGPT properly in video tutorials, but I still need to practice. Here are some typical roadblocks you can encounter:
- Your instructions might not be specific enough
- ChatGPT is not trained in writing styles
- You still need to find the latest resources yourself
- It will stop writing at a certain point and you have to re-prompt it
- Telling it a word limit might make it stop writing
Sometimes it’s easier to do a task yourself if it’s more efficient than having ChatGPT do it. Use technology only if it improves your work, not if it causes more work for you.
I’m Not Worried
Total time to write a 1000-word article using ChatGPT, including using my own judgment = 6 hours. I saved up to 4 hours.
I was able to do the research and outlining at the end of my workday because it wasn’t so intense. That took 3 hours.
I spent 2 hours writing the article. Even though I kept 10% of what ChatGPT generated and re-wrote the rest, having the article structure helped immensely. I did the rewriting at the start of my workday when I felt at my peak.
Lastly, I spent 1 hour editing, spell-checking, and checking for tone and keywords. This was where ChatGPT was most likely to get weird. It was better to do it myself than to prime ChatGPT again by re-prompting it.
I believe in taking my time to do articles right, and I can still do that, even with ChatGPT. I’m not going to half-ass an article just because I can. I make it clear to potential clients that they are paying for quality and that’s not going to change.
Note: I did not use ChatGPT to write this article because it was all from personal experience. I used it to generate an emoji for the section header on using emojis for work, but that was it.
Is ChatGPT Useful and Ethical?
Yes, but only if you want it to be. Be aware of ChatGPT’s limitations, and continue to practice. Be aware of any legal issues if you are doing an article for a client.
If you know your craft, then saving time and brain space on your writing process is fine. Using ChatGPT is a good way to kill writer’s block, and you don’t have to spend hours struggling to finish an article. You can work when you feel fresh.
Bottom line: ethical people are always going to be ethical. Unethical people will find ways to be unethical. If they don’t use AI to be unethical, they will use something else.
I’m sure knowing how to write properly using ChatGPT will be a sought-after skill in the future, so practice now!
What do you think of my experiment? Have you experimented with ChatGPT? Let’s trade stories!
