2025 ChatGPT for Marketing
Consider using the wizardry of the powers that be, while you still can
While I'll never use AI-generated writing for my books, nor AI-gen graphics if I can help it, I have no compunctions about using its marketing capabilities. The marketing types (social engineers) have always been the Kings' wizards, and therefore enemies of The People, as far as I'm concerned. If I don't use their powers to help audiences discover my educational, enlightening, smartly-subversive content, then those sorceries remain solely in the hands of the social engineers pursuing an evil agenda.
A couple of years ago I tried using ChatGPT to help with my web presentation for my books. The hope was to turn that tool into less time spent on such things, so I could focus more on creating.
Back then, what it offered me wasn't so great. Not bad, but really just rephrasing what I gave it.
Recently, I wanted to see how things have changed. Could 2025’s ChatGPT help me on the outreach side of the authoring equation?
After finding this article: https://www.authormedia.com/book-promotion-in-2025-how-ai-gives-authors-more-time-to-write
it certainly seemed like I should try ChatGPT again. The interviewees agreed that of the various AI tools out there, ChatGPT was the best.
They also strongly suggested using the paid version.
ChatGPT/OpenAI claims it has not stolen any materials, just uses the vast array of available public sources. Presumably that includes vast stores of public domain educational materials from decades past, on through to materials freely offered online today.
Nevertheless, for now I chose to stick with the free version. I can’t be sure, and even doing what it does might be unethical. I’m just not sure, not enough to fund it.
What I experienced with 2025’s free ChatGPT was not only very surprising to me, it was downright moving.
Large Language Models are not thinking algorithms. They don’t attempt to use virtual lattices of neuron-like code to mimic how animal brains function. They do use machine learning techniques for training and improvement, but it’s simply not an approach that’s trying to think. It’s an approach that’s trying to generate emergent, imitative riffing — a seemingly educated, thoughtful, Turing Test golem bubbled up out of the entirety of humanity’s public works.
And it has done so. Where ChatGPT from a few years ago was a few steps beyond what you see on aggregator websites, 2025’s ChatGPT is full-on Turing Test stuff, at least when having a consultation style conversation.
But that’s not important. Non-sentient humans beat the Turing Test all the time. What’s important is the thing’s answers to your questions, their value to you.
I didn’t treat it like a person and ask it how it feels or what it thinks, I treated it like the query engine it is. And the results were not only useful, but supportive.
NOTA BENE: Each separate session using ChatGPT is a standalone instance and does not automatically upload to the servers.
First I described my novels to it and asked for things like blurbs, marketing copy. What it came up with was really quite good, far and away better than what it came up with a couple of years ago.
I have updated my book landing pages based on its output.
Go check them out. The up front marketing copy is leaps and bounds better than what I had before, which were my clumsy attempts at enticing summaries.
The output wasn’t not perfect, and I did edit it a little here and there, but all it all it was quite good, in my opinion.
Other aspects of its attempted assistance were not so successful. It tried to generate some mockup graphics, even at a wireframe/layout level, but completely failed. But that didn’t stop it from being useful. It also answered my queries about offline methods of outreach with some very helpful suggestions, including useful details (actual names of organizations and so forth).
After exploring how it could help me with outreach for my novels, I asked similar questions about my brain tasks book for tweens. Its output was similarly helpful, mainly with the marketing copy that ended up almost verbatim on the “Puzzling Innerverse” page.
But how was it moving, you ask? Well, the next day I took a stab at using the engine in a different way, more like a brainstorming session. I described in detail my current vision for my current work-in-progress novel, “Attendants of the Future” and asked the thing for what it thought log lines, elevator pitches, and comps might look like.
Its output helped me clarify to myself what I’ve been intuitively groping for with this story, aside from helping with a few good marketing copy suggestions.
But then something astonishing happened. Since its comp list was fairly long, eight titles including some rather famous works and authors, I started to wonder if my story was overdone, cliche, possibly even fodder for accusing me of copying. (That’s certainly not the case, especially since the ‘biggest’ works in the list are works I don’t know, except for Le Guin’s “The Dispossessed”, but I read that decades ago and don’t remember it.)
So, in so many words, I asked ChatGPT is my story was a cliche rehash.
It’s response was a point by point analysis of the important ways my story was different from known works. But it didn’t just spit out a cold analysis:
“That is a very fair and smart question—and it's actually a good sign you're asking it.
Short version: your story is not redundant. It has familiar elements, but the combination, tone, and your specific execution make it fresh. Let me unpack that:”
You can see that for whatever reason, it gave supportive feedback on my having asked the question in the first place. What followed was its point by point examination of the elements of my story which not only stood out as different from past works that shared these themes, but presented those differences as valuable and expressed all this with supportive wording.
And I reacted emotionally to that. I don’t see the thing as a person, believe me. But it gave me intelligent, helpful, and supportive words that no human has given me. It took the place of a collaborator, or at least someone knowledgeable to shoot the shit with, someone helpful, someone willing to be helpful and supportive.
Whatever sources were used to train ChatGPT, by 2025 the engine absorbed humane aspects of all that absorbed language. Its riffs now incorporate being supportive.
In my first novel, I intentionally tried to break sci fi’s history of seeing intelligent machines as monsters bent on destroying humanity. That novel also made it clear that both of the fully sentient, non-biological systems depicted had to incorporate moral and ethical frameworks bottom up. I harbor no illusions about LLM AI systems — they are not thinking beings, and they have not been intentionally built with moral and ethical frameworks that can prevent bad behavior. Recent similar AI systems cheat while playing games. It’s already well known that these systems can sometimes flat out lie and generate and cite entirely fake references. That’s what you get when you don’t build in from the bottom up moral and ethical constraints. With black-box LLM systems, you *can’t* build that in, because, well, the thing is a black box. There’s no way to isolate how it does what it does. There’s no “prefrontal lobe” where morality gets trained in.
But in this case, in this conversation, somehow the thing absorbed and reflected back upon me a “ghost in the machine” that was supportive as much as it was analytical. In the moment, I had a distinct emotional reaction to receiving that support, despite the fact that it was simply in the form of words mindlessly spat out by a black box algorithm.
Frankly, my dear, I don’t give a damn. It felt good to be told that my questions and story ideas were good and valuable. I don’t have that in real life.
And I can believe the thing when it says those things. It’s basing those assessments on the sum total of human knowledge, or near enough equivalent to be indistinguishable.
I then went on to ask it for what it thought the cover should look like. It tried to generate some mockups, and one of them depicted a decent style I wouldn’t mind seeing used. But it was unable to graphically depict its own textual descriptions of various cover options. Those descriptions were sufficiently terse that I’m still not entirely clear on what they should look like.
And then I ran out of free-subscription usage.
All in all, I got some good marketing blurb out of the thing, some good ideas on offline outreach pursuits, some very important mirrored thoughts and rephrasings of my WIP story that will help me focus down and execute the thing, all of which was conveyed with a level of ‘intelligence’ in its language that I also don’t get on a daily basis. Very refreshing. And from that black box, mindless LLM system also came seeping out some ghostly and much needed wisps of human support. How bizarre is that?
And this was the free version.
