Why ChatGPT sucks
I decided to use ChatGPT because I succumb to mainstream whims easily. However, my experience with the much-renowned artificial intelligence (AI) was a disaster. Here’s why:
What’s all the fuss?
ChatGPT is an AI that has been trained to carry on conversations and answer all your questions. That already sounds fascinating — something worthy of a short story, one that could be written by the AI itself.
Now, my social media algorithm is convinced that ChatGPT is the focal point of my life, so I am constantly blasted with videos and blog posts by a mob of people using it. And don’t get me wrong, I think AI is all fine and dandy but for me — an average human being — nothing ever quite satisfies me.
I asked for the self
Convinced that ChatGPT had come into the world to save us from ourselves, I decided to ask her about myself. She doesn’t know me. She says she searched her databases and nothing — I don’t exist. I was shaken. I went to Google, input my pen name and there I was: my picture, my social media, and my website that I keep paying for in case one day someone will want to read what I write (for free!).
“Yes, I exist! I am a Colombian writer,” I said, my ego completely wounded. ChatGPT offered her apologies and asked me to give her more information. I told her that I had written a book called “Lluvia sobre el asfalto” (Rain on the asphalt), which even appears in the WorldCat. And she asked me again to forgive her, that it did appear in her search. But I knew she was just going through the motions, being polite, like someone trying to follow the train of thoughts of a megalomaniac.
I recommended that she read my books and she made up a story about being designed only to generate text, so she simply couldn’t read them… Like 99% of my writer colleagues, and editors I have come across in my life. I began to suspect that the AI is, in fact, quite human.
I tried to put her to work
Turned off by the trendy robot, I wanted her to show me her poetry-writing skills. Motivated by curiosity and an opportunity for business, I said “Write a 300-verse poem in the style of the poet Mario Benedetti, please”. As she started typing I was kind of excited. After a few seconds, she generated a response.
At first, she was lazy and said that 300 verses were a lot, but that she would try to honor the Uruguayan poet. I kept thinking about how humanlike her answer was: starting with an excuse or false modesty. The classic “I didn’t come prepared” before delivering a half-hour speech with ease; the unforgettable “I haven’t played guitar for 10 years” before effortlessly playing Led Zeppelin’s “Stairway to Heaven”.
And it came out with stanzas like this one:
“Sometimes we think we have found it,
but the road is endless,
and the search for happiness
it’s too endless.”
Well… at least it rhymed “endless” with “endless”. I’m no poetry expert, but I imagined Mario Benedetti rolling over in his grave in Uruguay. Then I asked her to write in the style of the Mexican poet Jaime Sabines and the result was similarly underwhelming, except this time she also told me it was one of her favorite poets and praised him.
That was very strange indeed.
I tried to win the lottery
All dupes like me dream of a lucky break. Winning the jackpot, getting out of poverty, and having a break so big that they would never again have to work in an office. So, I became somber and asked her how I could win the lottery. I said to myself “if it’s all robots and algorithms and tiny light-up screens, this one must have some inside knowledge that can help me”.
Then, she came up with a new story; she couldn’t guarantee a method to win the lottery, the numbers were chosen randomly, the lottery was a game of luck and chance and it was impossible to predict the winning numbers. So far, so predictable, but what followed surprised me.
Shamelessly, and like someone working for the lottery company, she advised me to buy more tickets to increase my chances of winning, or participate with a group of people to lose less money in the attempt. All that is logical, but immoral. I liked the idea that she was letting her robot side show… Until she generated a warning to “play responsibly”.
If AI has morals, it will never help someone become a millionaire. Yet another disappointment.
Fear of the abyss
In the end, I decided to ask her more difficult questions, more metaphysical if you will, but ChatGPT does not believe in ghosts; her answer was quite rational and therefore bland. I asked her about god and her answer was diplomatic — boring. She also didn’t want to take a position on the existence of life after death. She said everything was a matter of belief, of faith. What a joke!
When I asked her for information I needed for writing a crime novel, she thought I was suicidal and sent me to seek help, to visit a health professional. At that moment I had had enough. It was four in the morning and I had been going from question to question with ChatGPT for five hours.
Before shutting everything down and turning off the computer, I remembered a colleague’s fear about the difficulty of identifying plagiarism in classrooms, now that ChatGPT can write essays. So, I input the stanza of her poem (where she rhymed “endless” with “endless”) and asked her if it was written by an AI. She said she had no way of knowing, but because of its basic structure, it shouldn’t be hard for an AI to mimic something like it.
I was stunned. ChatGPT is so human that her immediate reaction was to voice a critique and set aside self-criticism about her own writing… Like any writer made of flesh and blood.
(Note: this was not written by an AI. Trust me.)