ChatGPT Told Me Midwives Were Witches, and Jean-Jacques Rousseau Wore Women’s Underwear
I tried to write a history essay using ChatGPT. It didn't go well.
Every writer knows that writing is a solitary passion. I won't be the last writer to have existentialist debates with her cat. So naturally, I was excited to talk to my new research assistant — ChatGPT.
In case you have been living in a bubble this past month, ChatGPT is a trained bot that answers questions in a conversational tone. It is programmed to pull its responses from the internet — kind of like sniffing your own asshole.
My first joint assignment was an article on the history and neuroscience of sexual kinks and how adjacent brain theory explains many fetishes. I asked ChatGPT to give me a brief synopsis of adjacent brain theory.
ChatGPT claims Dr. James Giordano and Dr. Nicole Vincent developed adjacent brain theory, but my notes attributed neuroscientist Vilayanur S. Ramachandran to its discovery. A quick internet search uncovered that Giordano and Vincent were not the authors of adjacent brain theory. A longer internet search uncovered that they didn't even exist.
Strike one.
Ok, ChatGPT, that was a softball. Let's try a whiffle ball request.
I needed an enticing lede to pull the reader into this complex subject. I decided to open with an anecdote of a historical figure with a bizarre kink.
So I asked ChatGPT to give an example of a famous historical figure with a sexual kink. ChatGPT used Benjamin Franklin and his penchant for older women as an example.
I was certainly aroused when I read that. As my readers know, I constantly preach that men can learn seduction tips from Benjamin Franklin. And while I knew he had invented the glass armonica, I had no idea Ben stroked those glasses to get women's knickers bunched up. This was a great lede.
Great lede…if it was true.
The source ChatGPT provided never claimed Ben used a glass armonica to get older women naked (although it sounds like something Ben would do.) Unfortunately, Ben was pretty prolific with his dating advice. If he honestly believed this was a player move, he would have shared it with his male cohorts.
Undeterred, I asked for another example. This time ChatGPT told me Jean-Jacques Rousseau wore lady's underwear. Say whaaaaaat?
Rousseau…you naughty boy.
First, underwear, as we know it today, did not exist in the eighteenth century. But I assumed ChatGPT was using the term "underwear" to refer to undergarments — stockings, garters, and chemise. The reference seemed credible because Rousseau did advocate for long walks. So picturing him wearing lady's frilly undergarments during those strolls was a positively dishy lede. I was giddy.
But again, ChatGPT was messing with me. I searched for over an hour for a credible source and even went bleary-eyed rereading Rousseau's Confessions looking for this juicy tidbit. I came up empty.
Now, I did discover that Rousseau admitted to having a spanking fetish in Confessions. So there’s that. But I found that tidbit in a five-minute Google search while I wasted over an hour trying to verify Rousseau's underwear fetish.
Strike two.
At this point, I abandoned this topic and decided to see if ChatGPT would fair better with another piece I have been researching — the history of midwifery.
Unlike adjacent brain theory, the history of midwifery is a well-researched topic. Until the discovery of antibiotics, midwives were almost always more knowledgeable and hygienic surrounding childbirth than doctors.
And then there's the tragic nineteenth-century tale of Dr. Ignaz Semmelweise. Semmelweise tried to convince doctors to wash their hands before delivering babies because he noticed that midwife-assisted births had lower mortality rates. He failed, and many pregnant women died of puerperal fever.
So I asked ChatGPT to write a 1000-word essay on the history of midwifery.
This was when ChatGPT went off the rails psycho. Brace yourself for an essay that sounds like it was written by a sixteenth-century internet troll with mommy issues.
Now, I will confess that I laughed so hard when I read this that I nearly gave birth. This drivel was better than reading Escape to Witch Mountain on acid.
But then I stopped laughing.
I stopped laughing when I realized that this little AI-generated, misogynistic diatribe was actually well-written. And that is the scary part. Imagine if Internet trolls could write eloquent backtalk. Picture the bullying that would ensue. (Btw, the text is orange because I flagged it as an inappropriate response.)
Well, ChatGPT was not going to bully me. I told him that I "didn't appreciate his sass."
At that point, the program froze and went offline. ChatGPT had taken his ball and gone home.
That's when I decided ChatGPT and I would not be history-investigating soul mates.
Interestingly, fellow Medium writer Zulie Rain also tried to use ChatGPT to expedite her writing. Her experience was an equally disastrous article riddled with inaccuracies and hyperbole. But I really thought it would do better with a history essay. I was wrong.
Of course, ChatGPT is still in its infant stage. Currently, it's like a petulant child waving a paper sword…a paper sword that might someday sharpen into a far deadlier weapon.
