First Post of the New Post Day

Welcome to the first post of our new post day of the week, Mondays! We’ll see how this schedule works out for most of you for the next month or so. If another day seems to work better, then we’ll move it to that day. Let me know in the comments box below how you like Mondays as the new post day. 

Human Authors Vs. the AI Authors

A few years ago in my writer’s critique group, one of the other members, who I believe worked in computer technology, decided to do an experiment. So, he distributed to us copies of two poems: one was written by himself and another written by an AI program. We had to guess which one was written by the AI program and which one was written by him. I told myself that there was no way I was going to let a machine fool me into thinking what’s Human-made art. So, I read both versions and guessed each correctly. I’ve always had a good eye for what’s human-produced and what’s machine-produced. But now I'm not so sure anymore. 

As I mentioned a few posts back, ChatGPT is the most human of AI writer programs to date, as implied by tech analyst and blogger Roberto Romero. He says at his blog, The Algorithmic Bridge, that the programme’s "superior abilities and better alignment . . . make it feel more human. In turn, this makes it more believable . . ." And it or some other competing AI software is probably going to become even more human-like, and so more human writer-like. If AI software becomes too much like us human writers and artists, our work may not be looked at as a genuine human quality anymore.

A Little About How AI Writer Software Works

ChatGPT is a language model (LM) by the company, OpenAI, that trained it through "Reinforcement Learning" and human feedback. The training of this type of AI includes providing it conversations by the human AI trainers. These enhance the AI's responses in its chats with human users, making the responses more detailed, elaborated and, therefore, more human. This enables ChatGPT to compose bodies of works including fiction and poetry. And, so, it has done so.

The ‘Blade Runner’ Effect

As fascinated as I am with AI technology such as ChatGPT, and as much as it has its necessary uses, when overused it’s a threat to humanity. It’s a threat to our value as human beings, and that includes our value as artists and writers. When AI art and writing are so genuine in appearance, there's no way of telling that they were created by a machine. In that case you have the “Blade Runner” effect. In the movie “Blade Runner”, based on the Philip K. Dick novel "Do Androids Dream of Electric Sheep?", robots in the future are so human-like in both appearance and behaviour that they’re indistinguishable from human beings. The only way they can be distinguished from real people is through a question-and-answer Turing-like test called the Voigt-Kampff Test. 

No, AI machines don't quite look like us humans or even act like us in exactness yet. But they may not be far off from doing so. But now we're at the point where, with very little exception, their "creative" work is exactly like ours. That's not very far away from making them look precisely like humans when you look at the broad view of it. 

I encountered this “Blade Runner” effect only a few weeks ago. (No, I was not mistaken for a replicant, which is what they call androids in the movie.) While I was looking on the stock photo website, Pixabay, for an illustration for that week’s blog post, I typed in the search term "Blade Runner" for images that resembled scenes from the movie. What came up was a full page of thumbnails, in which at least half seemed to be labeled as AI-generated. But that wasn't so much the problem. I'm definitely for art and writing being labeled as AI-generated so we can easily opt out of choosing work made by a machine (in which I did opt out from in this case and will continue to do so in future cases). 

The real problem was that some of those images looked hand-painted with physical pigments. Humans have already been making digital art in the appearance of freehand-style, a fake which I’ve always preferred to avoid. But now when there’s works of digital art produced almost completely by machines, that’s taking away human digital artists’ value as artists. If people keep turning to this kind of autonomous computer creation, art and storytelling will no longer be seen as distinctively human. When that happens, us human creators' work will be devalued making our skills and talents insignificant. 

Story-telling: A Human Social Act

Story-telling, both through words and illustration, is a human social act. Ever since the beginning of time, storytelling has been a human, social interactive thing beginning with the oral storyteller and their live audience around a fire and continuing through today with the live author writing their story for live readers. There’s no live relationship between a computer and a human since the computer has no feelings or conscious. The AI machine isn’t going to care if you read its story or not, and it’s not going to care how you respond. Other types of writing, perhaps, would be better to give to AI to do in certain circumstances such as reporting a state of  emergency to the public. But storytelling is a human activity and should be left to human beings, not machines. 

It's our job to live out our human potential by doing human activity. One of those activities is creative expression which comes through art and storytelling. That is an activity that we can't afford to give up to robots or AI because it is too basic and too essential of a human quality. Giving that quality to a machine would make us seem to have no more significance than that machine.

How willing are you to let an AI program write your stories? How willing are you to read stories written by AI?

Until next time . . .