Don’t Let AI Replace Your Human Voice.
AI is a fantastic tool, but there’s a real danger of becoming so enamored of its capabilities that we let AI replace our creative writing, our human voice, and our thoughts. Use it to enhance your creative writing and human expression, but not replace it.
Recently, I saw an ad on TV during the Olympics about a young girl inspired by an Olympic athlete. She wants to write a letter to her hero expressing her appreciation and how the athlete has inspired her. The AI company that sponsored the ad encourages her to write her personal letter with their AI product, Google Gemini.
This disturbs me. Google is urging people to replace personal human communication with AI-generated text. Instead of the young woman writing a personal letter to her hero, the letter would be composed by a predictive text machine. To me, this is a devaluation and replacement of human thoughts and emotions.
“Writing is more than just a means to convey information; it’s a way to connect on a deeply personal level. Whether celebrating a milestone with loved ones or sharing insights in this newsletter, these moments are opportunities to express our unique perspectives and emotions. When we delegate this task to an AI robot, no matter how sophisticated, we lose a piece of that human connection.”
David Sparks, The SLM
I follow other people’s writings because I want to read their thinking and their perspectives. I’m interested in reading what unique humans have to say, not predictive text Large Language Models (LLMs).
When I receive a personal letter from someone, I want a personal letter, not something generated by a machine.
Approaching AI: A Balanced Perspective
How are we going to work with AI and get the benefits of AI but not let it replace us as creative and emotional human beings? Ethan Mollick, the author of Co-Intelligence: Living and Working with AI, suggests that in using AI, we use several approaches and understandings.
• It’s an alien mind. AI mimics human behavior but doesn’t think or feel like us. “In short, we have an AI that acts very much like a person, but in ways that aren’t quite human.” Ethan Mollick, Co-intelligence: Living and Working with AI.
It doesn’t know true from false and doesn’t “think” in the sense humans do. We need to be aware that whenever we use AI, it’s not another human we’re interfacing with.
• Always Invite AI to the table. Experiment to find out what it’s good at and not good at. Discover its strengths and limitations. That will increase your skills in how to use AI properly.
• Be the human in the loop. AI is not dependable. It makes stuff up and can sound very convincing. You, as the thinking human, need to check sources and facts. Always verify AI-generated content.
As the human in the loop, you also need to use your thinking judgment to decide what to use, what not to use, and how best to organize it. Don’t let AI set your agenda.
• Treat AI like a person, but tell it what kind of person it is. Assign roles to AI for specific tasks. For example, when I ask AI to provide headline suggestions for my blog post, I tell the AI, “You are an online marketing expert.” If I want AI to give me suggestions to improve my first draft, I first tell it, “You are an expert editor of popular writing.”
• Assume this is the worst AI you’ll ever use. With AI’s rapid improvement, it will likely get much better over time.
• Ultimately, think of AI as an intern that makes stuff up. “Imagine your AI collaborator as an infinitely fast intern, eager to please but prone to bending the truth.” Ethan Mollick, Cointelligence: Living and Working with AI.
Always fact-check its work. You would never take an intern’s article, do some basic editing, and publish it under your name. You’d first check the sources cited to make sure they are accurate and relevant and add your own thoughts.
The Risks of Relying on AI for Writing Your First Draft
In a post about AI I wrote last year, I recommended using AI to write a first draft about a topic and then use it to generate ideas before you do your own writing. However, I’ve changed my mind about this approach and don’t do or recommend it now.
I don’t want to be overly influenced by the language and content of an AI-generated first draft. Once a first draft has been written, it is as though a die has been cast. A first draft exercises too much influence over my subsequent structure and thoughts. I want my structure and thoughts to be mine, not that of an impersonal predictive text machine.
I want to do my own hard thinking and organizing, and use that as my primary guide. As Ethan Mollick wrote in his book Co-Intelligence: Living and Working with AI, “Another consequence is that we could reduce the quality and depth of our thinking and reasoning. When we use AI to generate our first drafts, we don’t have to think as hard or as deeply about what we write.”
I know of several people in the Mac community who regularly ask AI to generate a first draft, edit it, and publish it. Perhaps they can maintain completely objective independent judgment over the content and organization, but I doubt it. A first draft heavily influences most people, whether they created it or AI did. That’s why published writers use independent professional editors.
Now, I don’t ask AI to generate a first draft. Instead, I do my own thinking and research and write my own first draft. After my first draft is completed, I ask AI to draft an outline on the topic or create a list of issues relevant to the topic.
If AI suggests some ideas I might want to include, I revise my draft and write the additional material myself. I don’t ask AI to write it for me and then copy and paste it.
Best Practices for Using AI in Writing
• Do your own thinking, research, and writing, then ask AI for suggestions. After you’ve done your own thinking and research and completed your first draft, ask AI for an outline of ideas on the topic. If AI proposes additional ideas you choose to address, take the ideas and develop them using your own thoughts and research.
Write your own first draft. Don’t let AI write it for you. Let it be personal, a reflection of your thinking and your expression.
• Ask AI for suggestions to improve your first draft. Tell AI it’s an expert editor. Feed it your first draft and ask it to make suggestions for improvement. Make any revisions you feel would benefit your writing. Don’t just cut and paste. Make it personal.
• Ask AI for headline suggestions. I suck at drafting good blog post headlines, so I typically feed AI my completed blog post and ask it to give me headline suggestions. If I like a suggestion, I’ll revise it and use it as the headline for my blog post.
• Ask AI for suggested action steps. I like to include recommended action steps at the close of my blog posts. After writing one or more, I’ll ask AI if it has any suggestions for additional action steps. If there’s one or more I like, I’ll write them into my post.
Enhance and Improve, Don’t Replace
While AI can be a handy tool for writers to improve their drafts, it can also be misused to supplant and replace human creative work and communication.
Personal creative writing is hard work and takes time. Because AI can now produce college-level narratives in a few seconds, It’s very tempting to use AI to avoid the hard work and let it generate blog posts, personal letters and emails, and other forms of human communication.
I’m not a Luddite who rejects any use of LLMs as a threat to humanity. I pay a monthly fee to access the most up-to-date version of ChatGPT, and I use it regularly.
I don’t have a problem with using AI to produce impersonal, routine communications (that a human reviews before publishing). However, I’m strongly opposed to using AI to supplant and replace human creativity, thinking, emotion, and personal communication.
I encourage writers to use AI to suggest topics, improvements to first drafts, headlines, and action steps. But only after you have done your own thinking, research, and writing. Likewise, I suggest you experiment with AI to learn its strengths and weaknesses.