During this holiday season, I’m sitting back and letting someone – or something – else write my column. Let’s see. How about an essay on the creative potential of the new artificial intelligence writing program?
GPT-3 (short for Generative Pre-trained Transformer 3) is a state-of-the-art natural language processing model developed by OpenAI. It has garnered a lot of attention in the media and the tech community for its impressive ability to generate human-like text.
One aspect of GPT-3 that has particularly impressed many users is its ability to generate creative, unique text that often reads like it was written by a human.
For example, if a user wanted to generate a story about a magical adventure, they could provide GPT-3 with a prompt such as “Write a story about a group of friends who go on a magical adventure to save the world.”
In summary, GPT-3 is a powerful tool for generating creative, unique text. Its vast knowledge and capabilities allow it to generate a wide range of text styles and forms.
Wasn’t that fascinating? Is anyone still reading this? I know, I know, a perfectly formatted piece of dreck. Except for the penultimate paragraph and curtailing or deleting a few other sentences, the above is ChatGPT′s response to my prompt.
Admittedly, if I read this essay as a former instructor for Composition 101, I’d have to give it a “B.” Nearly all grammatically correct, perfectly formatted, flows well.
Why not an “A?”
1. Overuse of the verb “to be” (is, was); 2. Repeating key words in adjoining sentences without thought-provoking rhetorical effect (“text”; “generating” and “generate”); 3. Incorrect use of pronouns (“user” referred to as “they”); 4. Mind-numbingly lackluster.
In other words, functional, informational. But is that all readers read for?
ChatGPT’s capabilities, however, are impressive. In a National Public Radio interview (www.npr.org, Dec. 16, 2022), Ethan Mollick, a University of Pennsylvania professor, said:
“The thing people come away with if they play with ChatGPT … is, wow, I can cheat on essays with this thing. And then if they spend a couple more minutes, they say, well, I can cheat on creating software code or translating language.”
Mollick “created” a workable Master of Business Administration syllabus and, when interviewed by a TV station, learned afterward the questions were generated by ChatGPT.
My cousin Willie Cade, leading a national charge to implement Right to Repair laws so farmers can fix their equipment without paying $500 to haul them to a recognized John Deere service facility, had some ChatGPT Christmas fun: “‘Twas the night before Christmas and all through the house / Not a device was working, not even a mouse. / The computers were broken, the phones in disrepair / All because the right to fix them was not there.”
Cute, but without the familiar anapestic meter, disruptive to the ear.
Furthermore, writer beware: ChatGPT doesn’t know it doesn’t know everything. An article by McKinsey Global Institute’s QuantumBlackAI (www.mckinsey.com, December 2022) reveals, “ChatGPT sometimes ‘hallucinates,’ meaning it confidently generates entirely inaccurate information … and has no built-in mechanism to signal this. ... For example … asked to create a short bio … it generated several incorrect facts for the person, such as listing the wrong educational institution.”
When we let machines think for us, we’re doomed. Not physically deceased, but spiritually, creatively defunct.
I asked ChatGPT to write a paragraph about its own dangers.
“ChatGPT is not a real person and does not have the same level of understanding or context as a human. It may generate responses that are inappropriate, offensive or even harmful.”
From the horse’s mouth.
• Rick Holinger’s poetry and prose have appeared in more than 100 literary journals. His poetry book “North of Crivitz” and essay collection “Kangaroo Rabbits and Galvanized Fences” are available at local bookstores, Amazon, or richardholinger.net. Contact him at editorial@kcchronicle.com.