When AI Sounds Human
Artificial Intelligence (AI) large language models, such as ChatGPT, Notebook LM, and Grok, are rapidly improving.
AI-generated writing is rapidly improving, blurring the lines between human and machine authors. Earlier AI-generated writing was easy to pick out. But now, it’s often unclear whether an AI or a human being wrote something.
Should writers clearly disclose their use of AI? I believe they should—and here’s why.
When AI Replaces Human Creativity
A problem with AI that writes well is that it encourages writers to skip the hard creative writing process and depend upon AI as a shortcut to the work of writing.
It’s okay to use AI to complete what I call “business transaction” types of writing, such as business emails, but it destroys creative writing by taking its place.
Likewise, I don’t have an issue with using AI for grammar and spelling checking. It’s not a replacement of my creativity for AI to suggest a blog post headline or section heading. Those are stylistic summaries of my creative writing.
I follow and read human authors because I want exposure to their unique, personal, and authentic human voices. I want to hear their ideas expressed how they express them, not how AI expresses them.
Giving Ideas to AI Isn’t ‘Crafting’
Some argue that AI-generated writing meets the definition of creative writing if they give the AI their ideas and let it write a draft expressing them. They then “lightly edit” the AI text and publish these documents, claiming they “created” or “crafted” them.
I respectfully disagree that writers who use AI to do their writing “created” or “crafted” what they publish. To me, it sounds like a rationalization to let AI do the hard work of writing while still claiming you are a creator or a crafter of documents.
If I gave a human writer some ideas and they wrote an article or book based on those ideas, it wouldn’t be ethical for me to claim I wrote the piece or crafted/created the piece. Someone else created the prose and used their writing voice to write it. It’s their creation, not mine.
Likewise, if I gave a musician some ideas for a musical piece, but the musician then took my ideas and created musical notes and lyrics to make a song, I couldn’t legitimately claim that I created or crafted the musical piece. The musician took my ideas and through his own prism came up with the idea for a song, and created the lyrics and music.
Some argue that using AI is like collaborating with another person. However, unlike human collaboration, AI lacks genuine human creativity and voice, reducing authenticity. AI does not think or create like a human. Also, when humans collaborate to create something, that collaboration is acknowledged.
The Need for Clear and Transparent AI Disclosure
The time has come when writers should seriously consider disclosing their use of AI in their writing. AI has become so human-sounding that it’s sometimes difficult for readers to identify and distinguish between human-written and machine-written content. Transparency is needed here.
The disclosure must be specific enough that readers can understand to what extent AI was used and for what purposes. I recently read a disclosure statement that said the author ”crafted” the article with “writing help” from AI.
That doesn’t help me much because it doesn’t tell me what I need to know. Does “crafted” mean he wrote the initial draft himself or that he gave ideas to the AI and the AI drafted the prose?
And what does “writing help” mean exactly? Again, did the AI write the first draft, and then the writer lightly edited it? Or did the AI provide grammar and spelling checking and maybe gave suggestions for other ideas he might want to weave into this article?
How I’m Disclosing My AI Use
These thoughts prompted me to create (with no AI use) a more explicit disclosure to use with my writing. It reads, “AI Note: I wrote this blog post myself, using my own words for the initial draft. I used AI only to suggest headlines, section headings, and improvements to the text.”.
I want to know this about someone else’s writing to help me decide whether I’m reading unique human expressions and creations or primarily machine-generated language.
To me, revealing whether a human or AI wrote the initial draft is extremely important since the first draft sets the basic structure and expressions for a published work. (See my blog post The Art of Writing: Why AI Shouldn’t Write Your First Draft.)
“Light editing” doesn’t change that or make it the writer’s own words or creation. And if heavy editing is done to the point of overcoming the AI-created model, why not just write the initial draft oneself?
Toward Better AI Transparency
I’m sure this version of my AI disclosure statement is not the final version. I’ve already modified it several times, and I’m sure I’ll continue to refine and rewrite it.
I encourage other writers to draft their own AI disclosure statements. Readers deserve transparency about what they’re reading, and writers maintain trust by clearly stating how AI helped—or didn’t.
It would also be appropriate for readers to ask writers about their use of AI so they can make an informed decision about what to read and who to follow.
AI Note: I wrote this blog post myself, using my own words for the initial draft. I used AI only to suggest headlines, section headings, and improvements to the text.
When I post links to product pages on Amazon, my links include a referral code so that when products are purchased after clicking on the link, I sometimes receive a very small percentage of the sale. While the amount that I receive is small, it does help to defray some of the cost of running this site and gives me a small vested interest in having readers of OriginalMacGuy.com purchase products using these links. That said, I do my best to only include links to products I believe are worth buying.