Writers, Behold!
Generative AI has exploded upon the writing scene. We can’t ignore it.
I remember clearly the moment the large language model known as ChatGPT (Chattie in our household) entered my life. My twenty-something daughter was wandering around the family room carrying her open laptop. She meandered over to me and said, “Hey Mom! Look at this!”
I peered into her laptop screen, expecting a video or an interesting photo. Instead all I saw was a white void with small boxes around phrases such as “Tell me a fun fact” near the bottom of the screen. Under that was a blinking cursor waiting for input. “It can write stories!” my daughter said. “What do you want it to write?”
I peered at a prompt line at the very bottom of the screen. “Tell it to write a story about a little girl with curly brown hair who lived in a castle. She typed it in.
Oh my gosh! Within seconds, an entertaining story unfolded before our eyes. It was eerily similar to the make-believe stories my husband used to tell my daughter at bedtime. It had believable characters, a simple plot, and a satisfactory ending. I said to my daughter in a very serious tone, “Why do I even bother to write?”
Why write, indeed! Yes, the AIs are getting better and better every day. I know because I use Chattie all the time. I have tested it up and down and right and left. Over the last year, I have seen the models improve significantly. And I have read many dire predictions about the impact this might have on our society as a whole. This technological advance gives us much to process. Before jumping to fearful conclusions, I think we should first understand the nature of this beast.
At a rudimentary level Chattie is only a highly sophisticated statistical “predict the next word” black box. It has been fed an unimaginably large compendium of written works composed by human beings and found on the internet. The result is a multi-dimensional highly complex “map” of words where like is placed with like, or words that are most likely to go together are placed together. The black box is then “trained” by applying a mathematical equation with billions of parameters. Once trained, Chattie knows how to respond to a prompt.
When we give Chattie a prompt of one or more sentences, it traverses its map of words and finds strings of words that are most likely to go with our original sentences. This “answer” may look intelligently derived, but it is not. Chattie has no intentionality. It’s simply a sophisticated pattern-matcher— a brilliant machine-learning application. It is not equivalent in any degree to someone who has a vision and makes decisions, such as a writer.
We writers have been graced with the great need to express our thoughts and emotions by putting words on paper, massaging them, sculpting them. We are motivated by desire and a sense of mission. We have something unique that we want to say, that we must say. It is our soul that sings out whenever we put pen to paper or fingertips to keyboard. It is our heart that cries with joy when inspiration arrives and we can say, “Well said!”
Chattie has neither heart nor soul. It is simply a computational tool that produces strings of words that make sense. We as writers can embrace this technology by finding ways in which it makes our lives easier. For instance, it is a great research tool. I can ask Chattie any question about a well-known event or person, subject, etc., and it provides answers. If I don’t understand a concept, I can ask Chattie to explain it to me; what a great personalized tutor! Exploring the capabilities of Chattie is pure fun.
Yes, we should use this technology wisely. Even more importantly, we should try to get a foot into the door of the natural language processing labs that are producing large language models. Wouldn’t it be dandy if we could give the researchers pointers about AI tools that would be particularly useful to those who are experts in using words?
We can’t ignore generative AI. Let’s embrace it.