Once more, with feeling: where ChatGPT hits the skids
I would be less than honest if I didn't admit to a bit of trepidation when I used ChatGPT for the first time.
The artificial intelligence language tool took just seconds to write a small but good piece about itself. It contained unfamiliar terminology that might take me half an hour to learn.
ChatGPT is a language tool developed by US-based OpenAI. It is essentially a robot capable of conversing with users, writing essays, answering questions and generating codes. It is "trained" to trawl through billions of words and learn how they are used. And all this in 95 languages!
Faced with such a formidable capability, I, like many other ChatGPT users, had doubts whirling around in my mind. Would this technology someday replace me as a journalist?
I tried to find an answer by further interaction with the chatbot that everyone is talking about. As the great Chinese general and military strategist Sun Tzu famously wrote 2,500 years ago: Know thine enemy.
I was planning an interview with a fashion designer from Cameroon, so I asked it to make a list of questions I could ask and provide some background on the subject.
The chatbot did a good job. I needed only to add a few more detailed questions to finish the interview. Hmmm …. It appears that I have a good work helper.
Next, I told it to write a news story on the latest developments in the earthquake rescue work underway in Turkey. This time my helper fell flat, explaining to me that its data bank "only goes up to 2021."
That pointed to one limitation of such artificial intelligence models. They need a vast amount of "training" from existing information to form their responses. But the world is always changing at such a fast pace that it's hard for machine minds to "know" what's happening at any given time.
We humans, however, can see, hear and learn in real time, which is invaluable in journalism's ability to impart the latest news to the public as quickly as possible.
So, okay, breaking news is not ChatGPT's forte, but what about less-timely feature stories?
I asked it to write a profile story on US actor Misha Collins. I chose him because he's famous but not that famous, and he does a lot more than just acting. I wanted to see how the chatbot would handle such a story.
The result was actually quite disappointing – a dry, rather dull story that might feed the actor's Wikipedia page but lacked the juicy bits of a "people" story that attract and keep a reader's attention.
My fears of being replaced by artificial intelligence were rapidly waning. Even ChatGPT admits that it doesn't have feelings or emotions, which is a defining difference between machines and humans. And some jobs, such as writing, demand that human element of imagination.
The more I read what ChatGPT writes, the more I see that there is something vital lacking. The chatbot stories are like skeletons, without blood or flesh. By contrast, stories written by humans incorporate something of the author's feelings and personal experiences. They show us soul, love, hate, fear, whimsy and dreams.
Feelings and emotions are the source of creation. As long as the ability to create exists, I don't think we have to fear that AI will replace us. Even as AI evolves, we can evolve as well.
But what happens if artificial intelligence can someday be programmed to have "human" feelings? That's a whole different story. Let's just hope that the world doesn't go the way of "The Terminator."