In 2008 or 2009, I started a WordPress account. I was already writing a lot on Facebook, and I thought this was probably a better way to write. I posted regularly for a time. Then life happened, and I went back and forth between writing for the web and just journaling. I often talk about how I transition from teaching via content and copywriting. My first full-time marketing role was as a copywriter. I spent several years writing blogs, emails, social content, and more long-form content like ebooks. In 2023, everyone starts using AI.” It’s the future,” they’d say. I’m generally always an early adopter of. new things, but for GenerativeAI, I was hesitant. My boss encouraged me to use it. At first, I said, “This feels like cheating”, but then I embraced it. ChatGPT Addiction I used Gen AI for just boring tasks at first. Just the things that I didn’t really want to do. I would use it for documentation and other administrative tasks, but I was still heavily editing it because it was less than legible. Then I proceeded to start using it for the things I actually enjoyed, thinking I would have more time. That’s not exactly what happened. It was hallucinating a lot. Nothing it said was reliable, but I couldn’t stop. I got to the point where I didn’t even trust myself to write a Slack message without running it through AI first. The feedback I was getting regularly before I fully adopted AI was about my tone of voice. It was too casual, too playful, too relaxed. So, I used it to make my voice sound more like theirs. The feedback was almost always saying, “ You shouldn’t talk like that.” So, basically, I promoted my AI to sound like the corporate jargon machine. The aim was to sound more “professional”. Eventually, everything I wrote I doubted. Everything I submitted for review was nitpicked. So, I gave it all over to AI. Things I could easily say in conversation, I couldn’t write. I had more typos than usual, and my autocorrect was bonkers. I was creating so much with AI, I eventually stopped editing the output. I read it so much that it started to seem “ good enough.” By September of 2024, I could barely string 5 words together at work without getting AI approval. ”What do you think, Chad?” “Edit this, Chad.” “Is this right, Chad?”It was an abysmal experience for a writer.It reminded me a bit of that scene in Wish when the King or President or Governor ( I forget) was like, “ give your wishes to me, and once a year we will grant it for you, but then it’s revealed that he’s actually evil. Reclaiming my voice Then, in October 2024, I started A Black Girl’s Guide to write about my relationship with being a Black woman in corporate spaces, and that expanded to things I wish I knew before. Things to help to navigate life. There were letters to little me, and it really opened the door to getting my brain back and trusting my voice again.My true wake-up call came when I got a “wrapped” email, telling me how many prompts I sent, how many “conversations” we had. It was so bad, and I felt ashamed of myself. I decided to QuitGPT.I wanted to do it cold turkey, but some people need bridges, Hi my name is some people. Here are the 8 steps I took to reclaim my brain * Stream of consciousness writing - I would just write what I wanted to say in paragraph form and then prompt the LLM to organize it for me. That helped me remember what my flow was. * Outlines - I would tell the LLM my thesis and what I wanted to work through, coupled with 4th stream of consciousness draft, and I would ask it to create an outline. That I would then fill in with my voice and research. * AI review only - After I stopped needing outlines from it, I would plug in completed work for review. I wanted to catch issues that would harm comprehension, as well as grammatical errors and typos. * Teach me prompts - Instead of asking it to do it for me, I would ask it to teach me.I would ask for the steps or where to go to get something done. Essentially treating it like a tutorial machine. * Setting a hard boundary - No AI writing for personal projects, I would outline and write those myself. My personal projects like the children’s books, my poetry we me. I didn’t need AI to tell me if they were good or bad cause it was my art. I also got back into writing essays for fun, and I wanted it to be my view and not the spin that an LLM wanted to put on it. * Setting a firmer boundary - No AI output for personal or work projects. I didn’t want my work to look like everyone else’s. Also, because my brain was becoming used to creating again, I did not need that support. * Training myself to use older research methods again. Google Scholar and Google News were my go-to for a time. I also reimplemented Boolean searches and leaned into going to the website to explore the resource myself. Library resources like Jstor or GALILEO have also been incredibly helpful. Wikipedia’s citation section has also been a good resource to go back to. * Going back to Grammarly, Free SEO tools, and Hemingway App. - These tools exist to help me improve my writing, NOT to think for me. they allow me to think through how I’ve phrased a sentence while considering the audience reading it. At this stage, I don’t touch it nearly as much as I used to. Writing again has been like riding a bike. I still have days where Google’s AI mode gets me, but not in the same way as ChatGPT. Once I figure out how to bypass that, we will be on the road to a mostly LLM-free life. Choose your struggle I remember a creator on Instagram saying that her therapist told her that she had the privilege of choosing her hard. She was talking about it in reference to washing her hair, but I am talking about it in reference to AI. Our brains often offload the things we don’t use. You build neural pathways and stronger brain/muscle connections when you do things over and over again. So, you can choose the challenge of maintaining your skills, or you can lose your skills to the addiction to a microwave answer ad struggle to communicate.Ultimately, having my skills stay intact is important to me. If this takes a bit more time, but looks better, reads better, and has more insight and value, this is the road I’d rather take. This isn’t a full rejection of AI. Grammarly is an AI-based tool. It has always been. Hemingway App has always been an AI-based tool. However, they aren’t doing the work for me. Turning down help seems silly to me, but help to me isn’t “doing it for me.” I have the ability to think, perceive and judge. I can use my emotions and my logic to come to conclusions that make for better content. A little honesty I’m not at all saying I don’t need or want help or guidance. I’m also not saying I only need guidance at the end of a creative journey. I am standing on the backs of the people who learned things and shared them on the internet. I also understand that the content creation space is moving at the speed of light. People are trying to stay ahead of the curve by using LLMs to take some things off their plate. I understand you. I’m not judging you. If you choose to continue to use AI… this is my advice to you. READ IT OUT LOUD. If the way it sounds annoys you to hear, it will bother a lot of other ears. Get full access to Learning with Stephanie Benoit at bennieonthejets.substack.com/subscribe
