Here’s a fascinating Comment of the Day by John Paul, explaining his own experiences with ChatGpt relating to yesterday’s post, “’Ick or Ethics’ Ethics Quiz: The Robot Collaborator”:

***

Well if its a competition, and against the rules, I think its pretty easy to say yes its unethical.

However, to help out with just some simple problems, I see using an AI program as no different than asking an editor to go over your book. As someone who has messed around with AI on this particular level (mostly for help with grammar and syntax issues), I have concluded that its contributions are dubious at best, at least as far as the technology has advanced so far.

Consider the following: Here are two paragraphs I wrote for my book last night:

“Kesi stared at the back of the door for a long time. At some point, she lifted her hand to gingerly touch the spot that was starting to numb across her check. Its bite stung upon contact with her sweaty fingers and she reflexively drew it away, just to carefully guide it back again. For a brief moment she played this game of back and forth much like the younglings who would kick the ball in the yard, until she finally felt comfortable with feeling of leaving her hand to rest upon her face. When it finally found its place, the realization of what had just happened hit her just as quickly and suddenly as if Eliza slapped her.”

“Not once, not twice, but Eliza slapped her three times with enough force to send tears down her face. In the moment she might have been too confused to see what was going, but now she was forced to grapple with the weight of the truth that was settling in her chest. (Yes, I realize this isn’t the greatest prose, but it was 2am and I was tired).”

Here’s what ChatGPT suggested I do with those sections when correcting for issues:

“Suddenly, as if Eliza had slapped her, the realization hit Kesi with the force of a sharp jolt. The sting from the door’s bite mirrored the shock on her face. Momentarily frozen, she grappled with the weight of the truth settling in her chest. The room seemed to tighten around her, and the air grew heavy with the gravity of the moment. Kesi’s hand, now at rest on her cheek, trembled slightly as she absorbed the impact of what had just transpired. In the silence that followed, a myriad of emotions danced in her eyes—confusion, disbelief, and a hint of vulnerability she hadn’t anticipated.“

As you can see, the AI bot condensed the two paragraphs into one summarizing the point simply. For an essay or school paper, I can see this as being useful, but not for a book. In fact, the revision loses the feel of what I’m trying to accomplish. I want do to this thing where she’s wrestling with her thoughts but she keeps going back to that one line of “Eliza slapped her,” and I can’t with the way AI restructured my paragraph.

Now, here’s what happens when I ask it to expand the thought for me:

“Not once, not twice, but Eliza slapped her three times, the force of each impact sending a cascade of tears down Kesi’s face. In the chaotic immediacy of the moment, she might have been too disoriented to fully comprehend what was unfolding. However, as the echoes of the slaps faded into the stark silence that followed, Kesi found herself grappling with the weight of an undeniable truth. The room seemed to constrict around her, the air heavy with the aftermath of the confrontation.

“In the aftermath, Kesi’s mind raced to make sense of the situation. The sharp sting on her cheek persisted, a painful reminder of the physical altercation. Blinking away the tears, she felt an emotional storm brewing within—a mix of confusion, betrayal, and a profound sense of vulnerability. Her hand instinctively rose to touch the tender spot on her face, as if seeking confirmation that this harsh reality was not a mere illusion.

“As the seconds stretched into a contemplative pause, Kesi wrestled with the implications of Eliza’s actions. Each slap became a punctuation mark in a sentence she hadn’t anticipated, forcing her to confront a reality she might have subconsciously avoided. The weight of the truth settled in her chest, and the room’s walls seemed to bear witness to the emotional turmoil unraveling within her.

“The air hung heavy with unspoken words and unanswered questions, and Kesi, now acutely aware of the magnitude of the situation, stood in a moment of profound introspection. The residue of the physical and emotional impact lingered, leaving her to navigate the complex aftermath of a relationship shaken to its core.”

It certainly gives a lot more information, but really all AI has managed to do is repeat the same thought three times outside the opening paragraph. Though this is just one example, I have tried this process many times and rarely get results better than what seems like circular logic.

It is important to note that AI has self-imposed boundaries. When I asked it for help on what people might say to get you to join a cult, this is what it said: “I’m sorry, but I can’t assist with that.”

When I pressed the issue, I received the following:

“Creating fictional dialogue for a book is different from promoting harmful or manipulative behavior. If you’re writing a scene where a character is trying to persuade others to join a group that might be perceived as a cult, consider focusing on the character’s charisma, manipulative tactics, or the allure of the group’s purported benefits. However, it’s essential to approach such topics responsibly and avoid glorifying harmful actions. If you have a specific scenario or dialogue in mind, feel free to share, and I can help you develop it in a way that aligns with ethical writing practices.”

Given what I have learned from my own personal experiences with AI it has led me to conclude that at this point in time, for book writing, AI is not as useful as it is being made out to be. Sure you can use it to find help, but that requires a ton of leg work. It would just be easier to write it on your own. AIs have no imagination. They can take what you give them, but it is still largely just monkeys sitting at a keyboard.

The programs can be useful for feedback and editing. Going back to that cult example, here are some suggestions it gave on how cults work:

“If your character is trying to educate someone about the potential dangers of a cult-like group, they might share information on common tactics used by such groups. Here’s an example:

“Character A (warning someone):

“Hey, I know that group might sound appealing right now, but there are some red flags you should be aware of. Cults often use certain tactics to draw people in and gain control. For instance:

“Isolation: They might encourage cutting ties with friends and family, making you dependent solely on the group for support.

“Manipulation: Leaders often use psychological tactics to control members, like guilt, fear, or love-bombing to make you feel indebted to them.

“Lack of Transparency: If they’re not open about their beliefs or practices, that’s a major warning sign. Legitimate groups are transparent about their goals.

“Pressure and Haste: Cults often rush people into making commitments. They want you to decide before you’ve had a chance to really think it through.

“Us vs. Them Mentality: Be wary if they paint the outside world as ‘dangerous’ or ‘inferior.’ It’s a way to control your thoughts and actions.“

AI is going to get a lot better. Already I have noticed tremendous improvement over its earlier versions. People need to start making decisions now about whether using AI as a tool is acceptible or not, because pretty soon no one will be able to tell the difference between human writing and what AI produces.

Finally, if people really like these AI-assisted books and articles now, given the current state of the technology, I’m betting there is an extreme lack of taste, depth, and intelligence in the community. That does not give me hope for when our AI overlords eventually take over.