My Newest Hobby: AI Detection
Do You Use AI?
I wonder if AI underestimated how quickly humans would figure out if something is written solely by AI.
I understand why people want to use AI to help them write. It’s so easy to use. Just throw a topic into ChatGPT and out comes a beautiful product in a second or two. And AI phrases things so clearly. No thinking required.
I use AI myself. It has become a useful editor to me. I get it to untangle some of my more confusing sentences. I sometimes ask it to reorganize my paragraphs to make better sense. I often let it do boring tasks like write SEO descriptions or choose social media titles.
Even while writing this (sort of) anti-AI rant, I want to use AI. Badly. I’m so tempted to put this very writing into AI for an automatic upgrade.
But this time I’m not. I want these thoughts and words to be strictly my thoughts and words. Not AI’s words. For better or for worse.
Because every day I notice how quickly the internet is swapping human voices for computer voices. And it’s making me sad.
Just like I don’t want to talk to a bot when I need help for a unique problem I’m having, I also don’t want to read AI’s words when I’m longing to hear a unique voice.
AI is flattening out unique voices.
We’re Rightfully Suspicious
I’ve been slowly reading a book about AI by Robert Wright called The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning. His premise is that we’re currently faced with a technological tipping point: will we as a species rise to the moral challenge of using AI for good and not bad? Will we get “the God we deserve,” on either end of the spectrum?
TBD. Check back in a few years.
In the meantime, we’re already becoming suspicious of images we see online. And rightfully so. Is that cat really ringing someone’s doorbell pretending to deliver a pizza? Did that politician really become a superhero last night? Some things really are too crazy to be true, even for these times.
So I’ve picked up a new hobby myself. It’s not one I particularly enjoy. But it’s one that I can’t seem to put down.
I’m becoming an unofficial AI writing detective.
Do You Want to Know If It’s AI?
I’ve been suspecting some writers’ work for awhile now. Not the loyal blog writers that I follow. Their voices seem authentically their own. I’m more grateful than ever for them.
But I am noticing an uncomfortable blandness of writing in several Substack newsletters I follow. Especially among newsletters that pop up quickly and suddenly have tons of content every week.
Obviously it’s not just me noticing either.
One study I read says that AI-generated content now accounts for 41% of long-form content on LinkedIn, 13% on Reddit, and 10% on Substack. And on average, the AI-generated posts on LinkedIn receive 45% less engagement than human-written posts.
We can just tell.
Within the past few weeks, Substack has encouraged writers to develop their own “How I make this” statement for their newsletters. Not to discourage or shame people from using AI, but to advocate for transparency about it if you do. I for one would appreciate if writings would come with a label if they’re created 100% by AI.
And people are having opinions about it. Big ones.
Substack also recently suggested using Pangram as an AI detector. It’s not foolproof, of course. (And isn’t it ironic to use AI to detect AI?)
Of everything I’ve tested so far, it’s told me what I suspected, that the piece was likely written completely by AI.
My favorite tells of AI are overuse of the words “quietly” and “matters,” as in this article that landed in my inbox yesterday:
“The 3 Substack Updates That Quietly Turn Your Publication Into a Business Tool
…The reason that matters now comes down to who’s arriving on the platform.”
Yep, it’s AI. (Or so AI tells me, anyway.)
I’ve even put my own work in Pangram as a test. So far it says my writing is 100% human . . . even though I use em dashes—I refuse to let AI steal them from me.
Of course I’m not surprised it shows I’m human since I did write it myself. But still, I appreciate the verification. Even if it’s from AI. Whew.
Is It All Bad?
Some people think writers should not use AI at all. Others say it’s fine because people are busy or they are genuinely bad writers, so why not let AI write for them?
To be fair, AI writing isn’t necessarily bad. It is usually quite organized, grammatically correct, and informative, things that I appreciate in good writing. I’ve received lots of help and insights from AI-generated articles.
In my own writing, I appreciate AI’s advice when I ask it questions like, What doesn’t make sense here? How can I make this simpler? What points have I missed?
And if I’m looking for solid information, like 3 tips to unclog a bathtub drain, I don’t care if the data comes from AI or a human. Just give me the help, as long as it’s accurate.
However, if I’m looking to hear what actually worked for a fellow human being—and I would appreciate a good laugh along the way when they tell a funny story of what happened when they tried to unclog their own drain—I don’t want that to come from AI.
I want it to come from you.
What AI Can Never Give Me
AI can give me information, but it can’t give me a person. Maybe this is what it boils down to: Do we prefer knowing content or knowing a person?
I do love some good content, don’t get me wrong.
But it will never replace loving a good person.
I want to hear you and your stories. In your human words. From your own lived experiences. In your individual tones and rhythms and expressions.
So what you get here today is 100% from a human. Could AI have improved it? Yes, definitely. It would have been cleaner and clearer.
But this time I didn’t give AI the opportunity to make it better. Instead, you’re getting a version with my errors and blurry syntax and perhaps a slightly confusing structure.
It’s not perfect.
But it is personal.
I’d love to hear your beautifully human thoughts in the comments.
Read More:
- Do You Say Thank You to AI? The Bots Are Listening
Are we trusting AI too much? Stay aware of how computers are shaping us. - What CAPTCHA Got Wrong About My Humanity
Instead of identifying fragile objects, CAPTCHA reminded me of my grandson, risk, and what it truly means to be human. - That “Friend” Is Not Who You Think They Are
As AI becomes more helpful and polite, it’s easy to confuse it with friendship. - When to Ignore GPS (and Listen to Yourself Instead)
Here’s how one road trip reminded me that the best navigation system is sometimes your own intuition.