Tag: AI

On Writing in the Age of AI (and a Pre-Announcement)

Things have been quiet—too quiet—over here on the blog. This is a big reason why I did not want to switch to Substack and try to get paying subscribers. Whatever gifts and abilities I have, productive regularity in my writing is not one of them. And I’ve seen too many ambitious Substackers get themselves in an awkward situation with regard to failing to crank out content for their paying subscribers.

Photo by nedimshoots on Unsplash

M-Dashes and the AI Temptation

The reason it’s been quiet here is—follow me closely here—I haven’t been writing very much. Now some astute readers may start to get suspicious at this point. I’ve used two sets of m-dashes in these first few paragraphs, and m-dashes are notorious indications of generative AI writing. But you know what? I love m-dashes. That’s why I quite loved reading this article, “I Love the Em Dash—Too Bad If AI Does Too.” (I found the link to it from this post from Tim Challies). I’ll admit I’ve been avoiding the m-dash for months—until reading the article and embracing my beloved punctuation friend again. Why the avoidance? Because in the back of every writer’s mind is the growing consciousness that something has changed in recent months. And what has changed? Namely that readers now read online texts with a suspicion that it was written by an artificial neural network and not a human person, and they don’t like it.

The writer’s awareness of the reader’s awareness drives the writer to avoid any indications of being artificial, even if that means changing the natural way they would normally write—which is ironic. But it makes sense when you realize that the AI models don’t have any writing ability or any style of their own, they only mimic and ape human writing with varying degrees of skill. So the programmers get the AI to write like good writers and good writers look in horror at the machine simulacra being churned out which most readers would not be able to distinguish from their own writing and the writers respond by changing the way they write to sound more human. The next step is inevitable: the programmers will teach the AI models to mimic this new, more human-style of writing, and the writers will resort to ever more extreme efforts to sound genuinely human.

As I’ve thought about this feedback loop dynamic and the profound challenge it poses to writers, I’ve concluded three things:

  1. It will not be possible for human writers to outpace generative AI, nor for readers to reliably distinguish human from machine text.
  2. Despite this, most readers will continue to strongly prefer human-written texts over AI texts for anything in the fields of journalism, opinion, creative non-fiction, or fiction. These are irreducibly human endeavours, and a reader rightly feels cheated and wronged when he or she discovers that a text of that nature was written by a machine.
  3. Therefore, the only way through for all involved is to cultivate trust. Readers will gravitate towards writers and institutions that have gained their trust through a demonstrated commitment to keeping their writing genuinely human.

I don’t see any other way forward. Writers credibly accused of using generative AI to write for them ought to lose their readers’ trust. (Caveat: I think there is a legitimate place for AI in research and some of the more mechanical aspects of editing).

Sadly, the temptation to use generative AI to write more and better will be too strong for some. There will be more scandals, especially among Christian writers aspiring to greater readership, larger platforms, and more followers. My advice to writers is to tread extremely carefully and consider the realm of AI as a kind of forbidden-fruit temptation. Yes, you’ll surely gain something, but you’ll end up losing far more, even if you don’t get caught.

Related to this is growing importance of institutional stewardship. I see two factors coming together to create a severe challenge to any organization that publishes content. The first is the general decline of trust in institutions that has characterized the 21st century, a trend that was accelerated during the pandemic. The second is the temptation to pass AI-generated content as human-generated content in order to do more faster.

My most popular post of all time on this blog was a critical response to what I considered to be a grievous failure of institutional stewardship: “Yes, Jesus was Crucified with Nails (and It’s Irresponsible to Suggest Otherwise).” In that case, it was not the use of AI that was the issue but the undermining of readers’ trust in the truthfulness of the Scriptures that was the issue. Still, it was a stark reminder that trust is a precious commodity in the realm of readers, writers, and editors.

An Announcement of an Announcement

All this has been on my mind not only because I’m a freelance writer and editor, but also because of a new and exciting development in my life that will soon be public. When that happens, I’ll be sure to post an update here as well.

The Sobering Prospect of “Adult AI”

Samuel D. James has a thought-provoking article over on his Substack that I’d like to interact with a bit. His main point is that we need to prepare for a change in tactics in the fight against the scourge of pornography because one of the main arguments is about to be made largely obsolete:

For many years, one of the key arguments anti-porn crusaders have used is that pornography objectifies and degrades women. Theologically speaking, this is absolutely true. Yet it is not been an effective argument, either in convincing lawmakers to put more legal restrictions on porn, or in persuading individuals to resist it.

I agree. It is a true and important argument, but not a terribly effective one. In my own writing on this subject, I have used this line of argument in a limited way and focused more on wider societal effects and on the personal spiritual effects. James goes on to argue that the church needs to shore up other lines of argumentation in anticipation for the day when pornography is available which does not make use of human actors, but uses AI to generate content. Again, I agree that Christian leaders ought to have a full-orbed view of the harms of pornography, going far beyond a focus on the harm done to those who produce it. But I think Samuel James overstates his case somewhat, and I’d like to lay out a couple of counter-arguments in the spirit of friendly pushback and in the interest of sharpening our thinking on this difficult but vital issue.

Make no mistake: this is no academic debate. This topic forms the battleground where millions of men (and women) are being ensnared by incredibly powerful temptations and progressively transformed into despicable moral cretins.

So while I agree with the main thrust of the argument, I have two pushbacks to offer.

The first regards this statement: “The next era of pornography will almost certainly feature no humans at all, but lifelike computer-generated images that have no souls, no legal status, and no inhibitions.” I think this will be partially true, but perhaps not nearly as much as the author thinks. Why? Because there is a difference that the user will quickly discern between the real and the artificial, and just like the completely CGI-fabricated fight scenes in all the new Marvel movies feel so flat and weightless and unsatisfying, so the novelty of the AI stuff will probably not satisfy the perverted minds and lusts of the users. There is a dark corner of the porn-addicted soul that not only wants to be titillated, but wants to know that this scene really happened.

The second is with respect to this part of the last paragraph: “When there’s no one to exploit, there is still God to offend. When there is no one to be trafficked, there is still God who sees.” True enough about God being offended and God seeing, but the dynamic of sin in the human heart is always towards deeper involvement. So even if we grant that AI-porn will displace most of the Western human actors, the one-way ratchet of this sin-slavery will pull the user towards real-life experience of their dark fantasies, and this will sustain or even increase the tragic demand for trafficked humans to serve as victims to those fantasies.

Related to this, one must ask why OnlyFans grew to be so popular despite an inexhaustible amount of free pornography already available on the web. The answer to this question weakens James’ claim that “porn’s future is post-human.” The lonely lust-addled men clearly find some added value to the OnlyFans experience such that they are happy to part with eye-watering amounts of money. And what is that value? My guess would be the thin veneer of human connection that OnlyFans apparently markets as its main appeal. There is some possibility of direct communication and access. I have my doubts that even the best “Adult AI” offering will be able to replicate the particular thrill this provides.

So I agree that the church needs to articulate a strong and robust argument against porn that does not focus so much on the damage done to the people featured in it. This will be critically important when the so-called “victimless AI porn” becomes even more mainstream. But I am not as optimistic as Samuel James that all this will really lower the demand for content featuring real humans and real bodies, nor that this will result in any decrease in human trafficking for the purpose of sexual exploitation.

Despite my slight disagreements here, I really appreciate Samuel James’ writing both at his Substack and in his recent book, Digital Liturgies. In fact, I’m very pleased to say I have an enthusiastically positive review of it slated for publication in the March/April (print and online) issue of Canada’s biggest evangelical publication, Faith Today. I’ll link to that when it goes live.

Note: The core of this post was first written as a comment on the Substack article and then expanded here.