AI is Not Democratizing Writing

“I hate to write, but I love having written.”

That quote is often attributed to Dorothy Parker. However, like most great quotes, its origin is messy and unclear. There are many versions of the quote and many authors who can lay some claim to it.

But it’s a sentiment that most writers can agree with on some level. Writing is often a messy, arduous and painful process. Even someone who writes as much as I do experiences writer’s block, frustration with getting a point across and the dread of staring at a blank page.

And those issues get compounded if you don’t write regularly. Then, they get compounded again if you have other factors such as a learning disability, not writing in your first language or just plain aren’t good at/comfortable with writing.

From that perspective, AI can seem to be a saviour. Why spend time on the parts we don’t like when we can just skip to the end? With a few sentences to an AI prompt, anyone can generate a column that is guaranteed to be grammatically correct and reasonably coherent.

It may not be entirely accurate, but that’s something that can be addressed in the editing process.

This has led many to claim that AI is democratizing the writing process. But while generative AI does make it easier to fill a book, it is not a substitute for writing. Regardless of how we normalize the use of AI, it cannot replace human writing.

Writing, like all art, is a fundamentally human activity, and the effort is a big part of why it has merit.

Skipping to the End

One of the interesting things about the “I hate to write…” quote is how many versions of it there are. For example, runners and gardeners both have their own versions of it.

However, we wouldn’t let a runner skip to the finish line in a 5K. If they didn’t run the race, they don’t get to claim they finished it. Likewise, while a homeowner can certainly get a landscaper to plant and maintain their garden, we don’t call the homeowner a gardener, nor are they gardening.

If generative AI can be said to democratize anything, it’s having writing. It makes it easier to have a written thing, but the human merely possesses the final object; they did not write it.

That’s because the act of writing isn’t just putting one word in front of the other. It’s a million tiny decisions, many made subconsciously, that guide the author and ensure that no two pieces are exactly alike.

When we turn all those decisions over to a computer, we are no longer writing. Instead, we’re skipping over to the “having written” part of the quote. Those decisions matter, and that is a significant part of why plagiarism is viewed as such a serious offense and why countless authors are angry at AI companies for using their work.

Furthermore, writing is not a one-way street. Writing also helps the writer. Not only does the physical act of writing help you retain information, but putting information into your own words (paraphrasing) also helps you learn and get a deeper understanding of the information.

If you use generative AI to “write” your content, you are deprived of these benefits.

However, some would argue that there are times when those benefits are not relevant. And, in a way, they are right.

The Potential Uses of AI

One thing that educators have to wrangle with is that education is very writing-focused. Essays are often part of a student’s life from an early age. However, most people who graduate will not be in a writing-focused position, and even fewer will have to write academic papers or essays.

However, writing remains a valuable way to assess one’s understanding of a topic, and, as mentioned earlier, it aids the writer in comprehension and memory. In short, writing is a powerful tool and is important for education.

But it’s not important everywhere.

Much of the writing we do is obligatory and contractual in nature. Though dedicated writers can crank out thousands of words per day, most people don’t get anywhere near that threshold. Emails, reports, cover letters and applications are some of the rare times non-writers will be called on to write.

As a society, we need to decide which of these situations requires human writing. Are there situations where skipping the writing process might be justified?

These questions have extreme importance for those who struggle with writing. They often find themselves shut out of spaces not because of a lack of knowledge, but an inability to write.

In turn, academia is one such space. Writing is a key academic skill. However, some individuals struggle with writing for various reasons and may fail, even though their chosen field is not writing-intensive. It is also one of the top causes of plagiarism among students.

Not everyone wants to be a writer, and maybe we shouldn’t force them to be. Just like a homeowner paying for a landscaper, not every homeowner wants to be a gardener. But if they are required to have a garden, perhaps by their homeowners’ association, it’s a valid solution.

However, that still doesn’t mean that AI is democratizing writing any more than landscapers are democratizing gardening. They are, possibly, democratizing the finished product, not the process.

That is, save the spaces where it is.

Where AI IS Democratizing Writing

None of this is to say that AI isn’t democratizing writing in some ways. In fact, all the issues here are with generative AI, not other “AI-powered” tools.

We’ve seen a surge in AI-powered grammar checkers, citation aids, and other tools that have helped individuals write with greater confidence. While grammar is an important skill to learn, and there are times such tools should be (and are) barred, for most people, this helps them write rather than write for them.

Grammar is a complicated subject. While it is an important aspect of writing, the focus on it is often steeped in ableism, classism and racism. There are many situations where grammar shouldn’t matter, but clearly does. These tools restore access to those spaces for many people.

But those tools don’t require AI, and many have been around for decades. Grammarly, for example, was first launched in July 2009. While AI has made some improvements to these tools, those advancements are largely incremental.

The spaces where AI is actually democratizing writing are all areas where technology has been making improvements for decades. It’s not bringing anything new to the table, just modest improvements to established tools.

So yes, AI can help with writing by making it easier, but that isn’t what most people are talking about when they are discussing AI democratizing writing.

Bottom Line

AI doesn’t make someone a writer any more than hiring a landscaper makes you a gardener, buying a prebuilt PC makes you a PC builder or hiring a mechanic makes you a car expert.

As a society, we need to decide whether writing is more like running or like gardening. Sometimes, it’s perfectly fine to skip directly to the final product. At other times, the work is the very point.

Personally, I think that we will have different answers for different types of writing. Just as there are times when using a ghostwriter is acceptable, there will be times when using an AI is acceptable.

However, to get to that point, we have to change our language around AI. AI-generated text should be called just that, AI-generated. Calling it “AI-written” or “written” at all maligns the importance of the writing process.

As with the gradient of AI usage, if we are going to set boundaries around AI, we first have to have an effective means to talk about it. Distinguishing between generated and written text is an important step.

Though I’m not usually one for such pedantry, we need to acknowledge that there is a difference between text generated by an AI and written by a human. That meaning holds importance for both the audience and the person behind the work.

It’s a distinction that I don’t think we can afford to forget or lose.