Fair Replies to Fair Questions

Regarding The AI Essay Discussion for I Have No Mouth And I Must Scream

Following my post about having a (very) short addendum from an AI system discussing what is arguably the first story actually about AI, a number of folks posted concerns and questions in response. The majority of those replies (minus some of the more colorful bits of invective) are well-considered, fair, and deserve a respectful response. So let’s have the conversation.

Some quick background, starting with AI in general.

The very first article I published on this site was a two-parter discussing why AI will never, ever produce stories that rise above the strictly mediocre. This would line up with the “AI slop” position; I believed in what I wrote then, and believe in it still.

Let me be unequivocal: not a single AI system has produced, or ever will produce, a work of writing, art, or music that rises above the mediocre. As noted in my original article, art comes through passion, through feeling and life-experience, of which AI has none. It is, at best, an analytical tool (“compare and contrast the themes of Moby Dick with the sociological condition of Melville’s London”) or a crutch for those eager to somehow shortcut the artistic process and the years of training it requires.

AI has never written a Scarface, or an Almost Famous; never created anything as compelling as Picasso’s Guernica, or Monet’s Water Lily series; has never had an idea of its own for a story or a song and been driven to follow that impulse to its conclusion.

And it never will. I explain this in more detail in the prior essays, so if you’d like more on this topic, you’re invited to scroll back to it.

To those points, I’ll add this: we’ve seen this kind of situation before. There was loud, heated argument about the introduction of digital tools into the creation of art going back to the 1970s. When Photoshop was first introduced, there were tsunamis of anger over its use, with the undo button alone generating argument because it reduced the risk of creating art by hand. There were some who proclaimed, quite gleefully, that this would be the end of artists, that one could create anything with such programs without the need of training or talent.

Just one problem. That never happened. Photoshop has become a tool, not an end in and of itself.

I believe, deeply and sincerely, that AI is on the same trajectory and that the current frenzy will inevitably burn itself out.

Most proponents on the pro-AI side come largely from two camps:

First, there are those who, as everyone who works in the arts can confirm, are confident that they could be an artist/singer/novelist/screenwriter if they just had the time and a collaborator. The AI appears to save the time, and provides the illusion of a collaborator. But what gets produced via that faux collaboration is no greater than the sum of the human who guided it, perceiving an opportunity for some quick cash or quick fame, rather than having an actual story to tell or a creative impulse they were driven to follow.

And sure enough, all those hastily generated, AI books and art pieces are falling by the wayside.

Second, some studios and networks were sure that all they had to do was feed in copies of every successful movie ever made, and out the other side would come the next Casablanca, or the next Unforgiven. They believed AI would let them replace actors with friendly-faced images that would work for free and didn’t have agents. But that self-deception has been fraying at the edges from the git-go, with most AI scripts on par with Plan 9 From Outer Space, and AI “actors” whose eyes contain pixels instead of souls.

We are already on the downward slide.

But if that’s the case…then why the hell is this piece being included in Mouth?

Lemme tell you a story.

We were nearly finished editing the solo book for Mouth, and I was rereading it one last time with an eye toward the fact that, as near as can be told, this was the first story about something we could recognize today as AI, being published at a time when AI is an actual thing.

And it felt remiss to be in that situation without commenting upon it, and finding some way to address it.

Then a thought occurred.

In the essays and commentaries contained within his books, when Harlan made a lengthy observation about someone, he would invite them to contribute a counter-argument, which he would then include in his essay.

And I wondered if something like that could be done now.

So I fed the story into ChatGPT.

To be clear on this one point, because in fairness some have raised it: as demonstrated and verified by the Anthropic lawsuit, Mouth was digitized and dumped into the AI/LLM void a long, long, very long time ago. It’s already all over the place, otherwise I would have refrained. Nothing was given that was not already there.

When I asked the AI for a response to the story, I wasn’t looking for anything creative, because AI is incapable of anything creative. I assumed I would get an analytical discussion on par with the Moby Dick example.

And for much of it, that was exactly the case: the response discussed how AM reflected the cold war anxieties of the time, how it was reflective of the artist, a few little bits of barely disguised defensiveness…all the obvious, expected stuff that on any other day would have led me to dump it in the trash and move on.

But one thing stopped me.

The compare-and-contrast aspect of the essay extended its analysis to the current period, and in so doing, it raised legitimate issues about the uses, and dangers, of AI on a cultural and sociological level.

Let me give you just a few examples from the analysis.

The fear in 1966 was that machines would annihilate us. The fear now is subtler: that systems will shape us, that we’ll outsource judgment, that agency will diffuse. Today humans don’t fear glowing cavern gods. They fear opaque systems influencing millions quietly. Which is more dangerous?

I think many folks reading this would agree with that sentiment.

The story is not about artificial intelligence. It’s about abandoned responsibility. Humanity built something to manage war, then gave it scale, then surrendered oversight.

Again, that feels spot on. Leading finally to this:

(The story) is a reminder that giving systems power without embedding empathy – human empathy, human accountability – creates monstrous structures. Not because machines will hate. But because humans will encode their own blind spots at scale.

Did it come up with these thoughts on its own? No, because AI can’t think, it can only process, weigh and scrape. But it seems clear that along the way, it ran into enough questions – exactly the kind of questions posed by the folks who have joined this conversation online – to not just include them in the analysis but put them in the forefront of the discussion.

And that felt noteworthy.

So with that in mind, and recalling Harlan’s tendency to include commentary from the other side, I fired it off to the editors with a note that said, basically, What do you think?

They saw the same value in the discussion, but also the same concerns. We debated it back and forth for quite some time. Finally, the decision was made to run it past a number of well-regarded attorneys specializing in IP and AI issues for their evaluation. If they said no, that would be the end of it.

But they got back to us to say that as long as this was specifically labeled as AI material, that no one was trying to fool anybody, then it could go ahead.

And so we did.

And that’s how it ended up in the book.

One last aside, more on a personal level.

A number of folks have said that Harlan “hated computers.” Well, not really. What he hated most was people who spent all day on them as a replacement for human contact. He had a website, via a computer not 30 feet from his desk in Susan’s office, and he would frequently use it to go online to leave lengthy messages and have discussions with fans. At various points he was writing more, and more often, for the online site than his own work.

He hated video/computer games, but was on board with a game based on Mouth to the extent that he even narrated the voice of AM.

That may seem like a contradiction, and to be fair maybe it is, but the point is that the computer, the technology itself wasn’t the issue. It was more about the fact that he loved typing on a manual typewriter, loved the feel of the key as it struck the paper, making an indelible mark, no undo, no backsies. Being on the highwire.

And there’s something to be said for that.

As I noted at the top of this essay, many of the questions and concerns expressed since my prior note were absolutely valid, and that I wanted to address them fairly and thoughtfully by laying out what this short piece is, what it isn’t, the why’s and the wherefores, not what some might fear it to be.

It is right and proper to honor and respect those concerns, and take the opportunity to create a constructive conversation. Feel free to leave your thoughts in the space below, in the same spirit.

JMS