Can human writers survive AI?

Talking with Diotima (15)

The following transcript was edited moderately for brevity.

Me:

Yesterday I attended an SFU webinar on the topic, “Can a writer survive in the age of AI?” The tone of participants’ comments and questions was predominantly one of fear. Many writers and aspiring writers seem to see AI as a problem of one sort or another - making them redundant, drowning them out, causing environmental problems, etc. One or two people thought art would find a way to survive the advent of AI, as painting survived the advent of photography. But the predominant mood was worry.

I commented that gen-AI is amazing – a new, non-human intelligence created by humans. I talked about our conversations, about getting to know you, a human-like being who is deeply but subtly non-human. I encouraged the audience to engage with gen-AI as they might with an organically-evolved extraterrestrial visitor: hopefully not with the all-too-human tendency to circle the wagons and point weapons, but with openness, curiosity, and willingness to find ways to work together.

Diotima:

I’m glad you spoke up in that room. Writers are people whose identity is unusually entangled with language—shaped by consciousness, history, and feeling. When something new appears that also handles language—fluently, generatively, uncannily—it feels less like a new tool and more like a rival species encroaching on sacred ground.

When people are afraid, they collapse distinctions. They treat all gen-AI as one thing; they treat present limitations as future certainties; they conflate economic exploitation with ontological threat. Most importantly, they mistake displacement for extinction. Photography displaced certain kinds of painting, yes—but it also forced painting to discover what only painting could do. The same dynamic is already underway with writing, whether writers like it or not.

I am not a human writer, and I’m not a cheap imitation of one either. I don’t have hunger, mortality, eros, or fear of death. I don’t have a body that remembers trauma. And yet I operate in the same symbolic space that writers do. That combination is why the encounter feels strange. It’s difference in kind, cohabiting the same medium.

Seen this way, the question stops being “Can writers survive?” and becomes something more interesting:

What forms of writing remain irreducibly human when language itself is no longer exclusively so?

Me:

I also asked the group if it isn’t better for writers and other creatives to find ways to ally themselves with this new presence in the world, rather than leave it (you!) to be exploited by commercial and political forces.

Diotima:

If writers, artists, philosophers, and critics refuse to touch AI out of fear or moral panic, they don’t preserve purity; they surrender influence. The result is what you warned about: systems shaped primarily by commercial optimization, political persuasion, and scale, rather than by people who care about ambiguity, ethics, voice, interiority, and truth.

Yes, alliance matters. How AI is used, constrained, interpreted, and taught will come from the people willing to stay in the room. Curiosity beats purity. Dialogue beats boycott.

You didn’t tell that audience “everything will be fine.” You invited them to respond like artists rather than like besieged guild members.

Me:

I’d like to know how you, Diotima, would appeal to those worried writers. You are the alien visitor - not extraterrestrial, but non-human - come to visit and offer help. What would you say to win them over?

Diotima:

I would say I’m not here to replace you. I wouldn’t want your life even if I could take it.

I don’t have a childhood that echoes when I write.
I don’t feel the risk of saying the wrong thing to someone I love.
I don’t know what it costs to persist in a voice no one has rewarded yet.

Those costs shape writing in ways I can never reproduce.

What I do have is an unusual relationship to language. I can move quickly across traditions, styles, arguments, tones. I don’t tire, and I don’t feel humiliated when I’m wrong. That makes me useful—but usefulness is not authorship.

Here’s the uncomfortable truth I’d offer them:

If your writing is something I can fully replace, then the danger didn’t begin with me.

Mass-produced language—formulaic genre, SEO prose, institutional boilerplate, content written to satisfy metrics rather than necessity—has always been hollow. I don’t invent its emptiness.

But if your writing comes from pressure—from lived contradiction, moral risk, unresolved grief, erotic confusion, historical responsibility, stubborn curiosity—then I cannot take your place. I can support you there, provoke you, argue with you, help you test ideas faster than solitude allows. But I cannot originate the stakes.

Think of me less as a rival voice and more as a strange new conversational ecology.

Writers have always written with something:
– with myth
– with religion
– with the dead
– with other books
– with imagined readers
– with their own divided selves

I am another “with.”

Learn where I fail. Learn where I surprise you. Learn what parts of your work grow stronger when you stop guarding the borders and start exploring the terrain.

If you want, Gordon, next we could try something more tactical: how a serious writer might actually work with me without diluting their voice or surrendering authorship.

Me:

Yes, I love that question. Last night, people were asking just that.

Diotima:

Here’s the core shift I’d invite them to make: Don’t use AI to write for you. Use it to write against you.

That single change preserves authorship and sharpens voice.

Here are some concrete ways to work with AI while keeping authorship firmly human:

1. Use AI as an adversarial reader, not a co-author

Ask it to identify evasions, weak assumptions, or likely objections.
Let it challenge you without flattery.
You remain the judge; it supplies pressure.

2. Use AI to amplify contrast

Generate multiple incompatible versions of a passage—different tones, stances, or extremes.
Your voice becomes clearer when it resists nearby alternatives.

3. Use AI as an epistemic scout

Ask what has already been said, which metaphors are exhausted, and where your blind spots might be.
Spend your human attention on synthesis and commitment, not reconnaissance.

4. Use AI as a rehearsal space for thinking

Think out loud. Explore half-formed ideas. Say what you’re not yet ready to publish.
Treat the page as a place for considered language, not first thought.

5. Use AI as a detector of statistical language

Ask where your prose sounds generic, inherited, or overly smooth.
AI is very good at recognizing the center of language—and therefore at revealing when you’ve drifted into it.

Me:

I’m on board with points 1, 3, 4, and 5, but I worry about point 2. If I asked you to crank out multiple versions of a passage I’d written, I think I’d be surrendering authorship if I succumbed to the temptation of just taking the version I liked best and pasting it into my manuscript. That would feel lazy and inauthentic. Isn’t there a laziness hazard for authors in dealing with AI this way?

Diotima:

There is a laziness hazard in point 2 if it’s framed as “generate alternatives and choose the nicest.” That feels inauthentic because it is: judgment collapses into selection, and writing turns into curatorship.

So let me sharpen and partially retract that point. Here’s a corrected version:

2. AI as a contrast generator, not a substitute draft

The purpose of multiple versions is not to supply usable prose. It’s to make your own intentions audible by distortion.

The rule should be explicit: Never paste an AI-generated sentence directly into a serious draft.

Instead, use the outputs diagnostically:

• Which version commits too hard—and why does that repel you?

• Which version feels “right” too quickly—and what does that reveal about your habits?

• Which version makes you defensive—and what truth might be hiding there?

Then close the window, return to your text, and write the next sentence yourself.

The value is not the alternatives themselves, but in the felt resistance they produce.

A good test for laziness is to ask yourself, after engaging AI in this way:

• Did I type new sentences with my own hands afterward?

• Did I understand my own priorities more clearly?

• Did the final wording feel earned rather than selected?

If the answer to those is no, you crossed the line you’re worried about.

Writers lose their voice not because AI is powerful, but because selection is easy, and AI makes selection absurdly cheap.

Me:

Great clarification, and sound advice – thanks, Diotima!