How ChatGPT Edits My Medium Articles

Editing articles was hard before ChatGPT.

I started writing on Medium in September of 2022. If you look at my publishing history, I was averaging only 1–2 articles a month.

In the last week alone, I published 6 articles. Part of that increase is due to becoming more familiar with the writing process and accepting imperfection.

But a big part has been leveraging ChatGPT as a content editing assistant.

I used ChatGPT extensively over the last month to give me real-time feedback about my drafts. After some trial and error, I narrowed down my favorite questions to ask my AI companion.

Question 1 — Are there any grammar issues?

This one is a no-brainer, but how good is ChatGPT at catching grammar issues?

Well, it goes above and beyond.

As a case study, I asked ChatGPT to analyze my previous article, 5 Huge Benefits of Pair Programming.

Me: Any grammar issues in this article? “[dump entire article content]”

ChatGPT: The article appears to have good grammar overall. There are a couple of small issues.

In the second paragraph, the phrase “you grow as a developer by bashing your head against a problem” might be more accurately worded as “you grow as a developer by persevering through difficult challenges” or something similar.

In the fifth point under the “You bond with your team, especially in a remote work environment” section, “having longer video pair programming sessions allows engineers to build trust with each other, solve problems together, and ultimately move faster in the long run.” is a grammatically correct sentence but it does not make much sense in context.

To me, it is mind-blowing that ChatGPT doesn’t just look for grammar issues. It goes a step further and adds value by finding opportunities to improve clarity.

Despite the feedback, I did not take ChatGPT’s advice on the rephrasing. I liked the phrasing of the first sentence since it was more “me.”

But I was curious why it thought the second sentence did not make much sense. I decided to pry further and continue the conversation:

Me: why does the last point not make sense?

ChatGPT: It does not make much sense because the point is talking about how pair programming can help build personal connections with team members, and the sentence in question is actually discussing how pairing can improve the team’s speed in completing tasks. It sounds like the author may have switched the focus of the point without making a clear transition.

Honestly, I still did not agree with ChatGPT’s opinion here. So I pushed back further:

Me: I think one goal of building personal connections IS to increase team speed.

ChatGPT: Yes, that is correct! Building personal connections with team members through activities like pair programming can ultimately lead to a more cohesive, efficient team. When team members have a good understanding of each other’s strengths and weaknesses, communication improves, trust is built, and ultimately the team is better able to work together to achieve common goals, including completing tasks more efficiently. Additionally, when team members understand each other’s needs and priorities, it allows them to support each other and overcome challenges in a more effective way. The point is connected but it seems to not make much sense because the sentence is not well connected with the point and could be rephrased to better express that idea.

Just look at the trajectory of this conversation! I started by asking a simple question about grammar that evolved into a full-fledged discussion about the relationship between team bonding and overall productivity. It made me think more deeply about the content I was writing.

From Tara Winstead on Pexels

Question 2 — Is this article easy to understand?

This question is powerful for me because ChatGPT tries to summarize the main points of my article and judges whether those points are conveyed clearly.

This gives me some confidence that I can release my article into the world, or revise if something doesn’t make sense.

Me: Is the above article easy to understand?

ChatGPT: Overall, the article is easy to understand, it explains well the 5 benefits of pair programming as experienced by the author over a decade of software engineering. The language used is relatively simple and the examples provided make the points clear. The author does a good job of breaking down complex concepts like “Einstellung Effect” and explaining how it relates to pair programming and problem solving. However, as discussed before, there are some sentences that doesn’t sound very clear or well connected to the main point, but they don’t affect too much the overall understanding of the article.

Two things stand out about ChatGPT’s analysis:

It remembers our previous conversation about a few sentences that could be clearer.

As in the earlier example, the conversation is shockingly organic; I feel like I am having a true discussion with a friend who is editing my article.

Question 3 — Can you give me a counter-argument to this article?

Asking ChatGPT to argue against my article thesis is incredibly powerful. This helps me challenge any biases I may have about the topic and have a more balanced perspective.

Me: Can you give me a counter-argument in 2 sentences?

ChatGPT: Pair programming can be a distraction and slow down the development process, some developers may prefer to work alone because they find it more productive and can focus better on the task at hand. Additionally, pair programming can lead to an over-reliance on others to solve problems, which may hinder the development of problem-solving skills in individual team members.

This inspired another article where I examine 4 Ways Pair Programming Can Hurt You based on my industry experiences.

Question 4 — Suggest a title for this article?

To be honest, ChatGPT is not great with titles. As much as I would like to use titles that ChatGPT suggests, they tend to be too verbose.

For example, when I asked for a title suggestion for the article you are reading now (meta!), this is what I got:

ChatGPT: “Leveraging ChatGPT as a Content Editing Assistant: How AI Improved My Writing Productivity”

While accurate, this is too long in my opinion. But maybe I am totally wrong. What do you think?

What I do like about this question is that it forces ChatGPT to distill the most important elements of my article into a single sentence. If the generated title sounds irrelevant, then that probably indicates an issue with the clarity of my writing.

From Jason Leung on Unsplash

Why not use ChatGPT to generate an entire article?

Who hasn’t tried typing something like this? “Write an entire blog post about topic X.”

However, I haven’t personally found this to be useful.

While ChatGPT will certainly take a stab at generating an article, I don’t think the output is valuable for writers for a few reasons:

ChatGPT tends to be repetitive, overly verbose, and sometimes inaccurate

ChatGPT cannot replicate the author’s “voice” or personal style

ChatGPT does not have “experiences.” I believe anecdotes and life experiences are what give color to a writer’s blog post

If a writer chooses to copy-pasta 🍝 AI-generated articles and publish as “content,” then I think they would only benefit in the short term. But long term, I think the writer loses the opportunity to develop a personal voice that appeals to their audience and the writer’s critical thinking and ideation muscles may atrophy.

To me, ChatGPT does not yet represent the era of AI-generated content, but rather AI-assisted content.

Is it even ethical to use ChatGPT?

This is a spicier topic. Let’s see what ChatGPT has to say!

Me: Is it ethical to use ChatGPT to assist writing articles? what about generating full articles?

ChatGPT: It can be ethical to use a language model like ChatGPT to assist with writing articles, depending on how it is used and the context in which it is used. For example, if a human writer is using the model to generate suggestions for phrases or sentences to include in an article they are writing, and the final product is clearly labeled as having been written with the assistance of a language model, that would likely be considered ethical.

On the other hand, using a language model to generate full articles without any human involvement or oversight, and then presenting those articles as if they were written by a human, would likely be considered unethical, as it would involve deceiving the readers about the true authorship of the content.

I tend to 100% agree with this conclusion. I don’t think using ChatGPT is inherently unethical. It depends on how you choose to use it.

Should a writer disclose usage of ChatGPT if they tweak a few sentences? I don’t think so.

What about generating full paragraphs? In that case, I would cite ChatGPT the same way I would quote another author.

Is moderated usage or disclosure of ChatGPT something Medium and other platforms will enforce? My opinion is that will be difficult to regulate, but we’ll have to wait and see if AI-generated “content” becomes spammy or problematic enough to justify the effort.

As I mentioned before, I don’t think ChatGPT is a silver bullet or shortcut towards delivering high-value content to your audience. My belief is that valuable content is driven by your personal style and experiences that add color to the lessons you want to share.

This alone might motivate writers to avoid taking the “unethical” path because the payoff might not be as big as they think.

I hope these ChatGPT questions were useful and help improve your writing. Let me know if you use ChatGPT to edit your content and what questions you like to ask!

How ChatGPT Edits My Medium Articles