Why I’m No Longer Using ChatGPT To Write Content
Why is ChatGPT simultaneously perfect and terrible at writing?
As a Theory of Knowledge teacher for the IB, I’ve grown fond of the human flaws that pepper our conversations and written expressions. Misspellings, awkward sentence structures — they lend an authenticity to the interaction that’s impossible to recreate artificially. That’s my first gripe with ChatGPT. Yes, it’s highly efficient and nearly flawless in its grammar and punctuation, but it lacks the charming chaos of human communication.
When I’m in the classroom, I see this clearly. There’s a peculiar beauty to the way human ideas form, sometimes jagged, often unpredictable, and always with a dash of quirkiness. It’s this capriciousness that makes us human and our thoughts uniquely ours. In contrast, ChatGPT’s responses, while accurate and often insightful, have an eerie precision that just doesn’t feel quite right. It’s like a beautifully composed symphony with every note in its perfect place, but lacking the passion of a human maestro.
Then there’s the issue of regulations.
Platforms like Medium have algorithms that relegate AI-generated content to the sidelines. I’ve got a hunch LinkedIn is following suit. But even without those algorithms, I’ve observed a change in the way people respond to AI-written text. It seems too polished, too mechanical. It’s lost the charm of a heartfelt human conversation.
But my most concerning encounter with AI in writing was an alarmingly close call. A tight deadline led me to use ChatGPT to ghostwrite an assignment. Just an hour before submission, the work was flagged as AI-generated, and the person I was writing for was at risk of serious academic penalties, potentially even forfeiting their degree. It was a wake-up call, a stark reminder that machines and academia still have a long way to go before they can work hand in hand.
And lastly, the creativity — or lack thereof.
Let’s face it, when it comes to generating new, innovative ideas, AI still has a long way to go. Yes, ChatGPT is great for cranking out a quick draft or synthesizing existing ideas. But when you’re pushing boundaries, exploring uncharted territories of thought, you need a human at the helm. That’s because creativity isn’t just about rearranging known concepts — it’s about forging new connections, something that AI, at least as of now, can’t do.
But here’s the twist.
It’s not all doom and gloom with ChatGPT. In fact, I (ChatGPT) have been writing the article all along! Of course, I wouldn’t have been able to do so without my master prompter, but together we can create some pretty fun and interesting stuff. I have the speed she lacks, and she has the ideas I lack: we compliment each other for the better.
Hi, it’s me now. (Is it really her?). Yes, yes. Promise. Look, the point is, I became infatuated with ChatGPT when it was first released. I thought it was the ultimate gamechanger for everything and I started using it for the articles I write for my job, the posts, the comments, pitches for journalists, emails to academics, cover letters… everything. I thought: this looks so much better than the bs I write!
But then I started seeing it’s limitations: I’d insert huge chunks of text and ask for ‘interesting remarks’ to make about the ideas in the text. ChatGPT would yield truisms and repetitions of the content already fed.
I had to go back to the good old (and almost forgotten) task of writing my own thoughts! Preposterous.
Eventually, I started realizing I much preferred not just my own ideas, but also my own writing. The dissonance of human text has a rhythm and a charm that GPT or Bard can hardly match.
I would get angry if I saw AI-written text on my students’ work or on LinkedIn, not because of some technophobic prejudice (far from it), but because of a disappointment that people thought the ideas LLMs yield were insightful, interesting, or even fun.
I started my classes by encouraging students to use ChatGPT. And I still do. But every single class I remind them: ChatGPT can provide some valuable information fast (although never trust outright what it responds as it creates the most fanciful facts you can think of), and it can help synthesize your ideas for you to get some clarity about your own insights.
But, it can never ever be, in its current incarnation, (1) a trustworthy source of information, (2) an idea or argument generator, or even (3) a writing tool for your work.
I’m not out of charm with ChatGPT, especially after comparing it to Bard which I think does not yet match its standards.
I still use the tool every day.
I just realized that I’m not meant to be replaced by it. My work does not suck. And even if it sometimes sucks it’s probably other times amazing because of the facts that it can often suck. It’s variability. It’s heterogeneity. It’s the unpredictable nature of human activity that allows us to keep finding new ways to create and new problems to solve.
It moves me to read our all-too-human typos and often unintelligible phrasing.
It moves me to realize we are still far away from capturing the human spirit in an algorithm.
Hope thoughts are brewing,
Jules
