Heyaa
Earlier this month, I tried using ChatGPT (cgpt, henceforth) to complete a series of book blogs. I dreamt about writing 3-4 articles in a day. Yes, 3-4 articles. Every day. For a month. Apart from ongoing client work.
Yes, I was all hyped up about cgpt.
But three days in, and the realisation hit. It’s not possible. The ecstasy started wearing off as I started using cgpt for regular blogging.
I had a huge list of articles to generate. And cgpt wasn’t helping me generate them in 2 minutes, unlike those gazillion Reels and carousels promised.
Cgpt didn’t help me 10x my productivity. What a dumbass I was.
What a dumbass I was.
I was back on the ground. Back to square one, not sure what to do anymore.
So, what happened exactly?
Before we move forward, let me introduce you to today’s newsletter sponsor.
Get smarter every day with Refind
Every day Refind picks 5 articles that make you smarter. You choose your topics of interest once, and Refind sends you the content you’ll love, on repeat. Don’t want emails daily, set it down to 3x a week. Thrice feels too much; turn on the weekly digest.
Loved by 100k+ curious minds.
OK, back to my story…
What happened exactly is that cgpt failed. No, not at generating what I asked. It answered whatever I asked, respectfully. But it failed at being reliable.
Yes, I know cgpt may occasionally generate incorrect information. But in my case, it generated fictional answers. It made up information that doesn't exist.
Let me narrate the incident from the beginning.
Since the first of February, I started using cgpt for my book blog.
Things were going well for me. Cgpt generated book lists on the topics I wanted, and wrote short descriptions, too. Within two days, I had five articles. Mostly around 700-1000 words.
But then, as I was on my sixth article, I suspected the book list generated by cgpt. One of the suggestions seemed odd. The name of the writer didn’t ring a bell.
So, I searched. I searched for the book name on Bing. Couldn’t find the details. I googled; nothing showed with that name. Both sites showed another book by that writer, but not that book.
I changed the keywords, tried variations, used double quotes… just in case it’s a rare book. I went three pages deep, on Google and Bing. Nothing.
So, next I went to the author’s page on Goodreads. It showed 20-something works, but not that book. I checked on Amazon. Same. I didn’t see the book under the author’s works.
I skipped that book in my article, and manually checked the other books I included, just in case anything else is fictional. Everything else was fine. But the trust was broken. I wrote the descriptions manually. Saved my work.
And then, I checked all the books and their authors that I included in my previous articles. I knew the details for most of the books, so it was somewhat easy. But I had to sacrifice my 3 days' worth of productivity.
From Day 4 onwards, I relied on my own research to create book lists.
But I kept using cgpt to write the descriptions. I thought cgpt will help me in completing those monotonous articles. But no, it didn’t.
It seemed unreliable in that case, too. I noticed it was repeating generic information instead of providing me with relevant book descriptions.
So, all my hopes, shattered; dreams crashed and crumbled.
I’m back to using my own wits.
And my case isn’t standalone, actually.
Fio had a similar encounter with ChatGPT spewing bullshit. There may have been many more incidents; unnoticed, simply because the user didn’t fact-check.
So, using cgpt-generated work is only harming your credibility.
Plus, I compared the descriptions created with cgpt, without much of edits, and the later ones that I created with a mix of cgpt and my additions.
Both versions seemed the work of a noob, paid with peanuts. But the earlier “only cgpt” articles look as if someone hell-bored with their work, hunched in a cubicle, in a dingy corner of an office, typing on their keyboard like a machine.
And that’s what you're getting if you rely on ChatGPT for your content needs. You're getting mediocre content on repeat.
“The world is currently in the middle of a mediocrity storm,” says Anmol, expressing his views on how AI is killing your creativity.
So, should we stop using AI tools?
Nopes, cgpt and other AI tools are helpful. In fact, cgpt is saving me about 10-15% of my time with those articles. Yes, like 10% in suggesting article titles and writing article introductions.
And that exactly is creating the problem.
Thanks to overreliance on AI tools, people aren’t thinking. They’re relying on it to be their knight of the night. No one wants to stare at that blank document. Everyone’s just pinging some AI.
With that realisation, my use of AI tools has gone significantly down.
I still use one or the other for ideation, or to take care of something repetitive. But it’s not going to help in the long term.
Think of the long term as 10 years.
ChatGPT will be learning from the content it generated for someone today, and repurpose it for someone else, maybe even you. The same mediocre content. Because it can’t “think.” It’s not creative. But you're.
And that’s all for today.
Bidding adieu.
Live long and prosper.
See you next week, with another story.