I became aware of P.D. Edgar’s
online literary magazine,
re•mediate, after attending the ELOonline (Un)Linked virtual conference last July. I was
immediately struck by re•mediate’s thoughtfulness.
Feeling a pedagogical kinship in his project, I got in touch with P.D. and we
began a delightful, generative correspondence. I also had been wanting to
respond to Mark
Goldstein’s essay about AI generated poetry, published here in 2023, for some time.
But though I was eager to challenge many of the arguments and assumptions in
his piece, I was reluctant to take an adversarial position, which tends to have
the effect of both camps becoming more entrenched. Instead, I asked P.D. if
he’d be willing to formalize and publish our conversation on Periodicities, He enthusiastically agreed.
The first half of our conversation focuses on P.D.’s use of AI in his own poetry, including a discussion of his
poem, “I make a patchwork of my childhood,” which was published in Issue 3 of the AI Literary
Review.
In the second half, we talk about re•mediate,
and his role as editor/curator. We also address the fraught challenge of
environmentally responsible use of AI models like ChatGPT.
Katherine Parrish: Tell me about your web magazine, re•mediate. Why did you start it?
P.D. Edgar: In the past several years, many waves of new(ish) writing practices
have come and stayed online. There were some social-media based forms like
Twitterature and Instapoetry, and all of the webnovels and fanfictions and
digital folklore on text-based sites, and then when I was in graduate school we
got whiffs of Web3: first, the blockchain and NFT
art,and then the
public-facing chat-based large language model ChatGPT, which was specifically
being marketed as a writing tool. I became immediately interested in what sort
of things could or couldn't be done with it at the time, and I presented at the
Digitorium conference, hosted by the Alabama Digital Humanities Center, on how
some poetry forms came almost naturally to ChatGPT even in the early days (some
didn't).
But each time those things came out, I watched some folks condemn that
exploration, and some folks build community around it. NFT poetry got the VerseVerse, there was a short-lived attempt to do
poetry in the Metaverse [the Facebook parent company’s virtual-reality space],
and fanfic and Instapoetry are still going strong. I wondered what it might be
like to bring a literary sensibility to the AI tools, and with my background
with Black Warrior Review (art and design),
Kenyon Review (digital archive
associate), and my college newspaper (I was news, opinions, copy, and
editor-in-chief at different stages), I wanted to be an early (but critical!)
adopter and editor of computer-assisted work writ large—I wanted to put
"AI" in conversation with all of the other ways our experiences of
life and of writing are mediated by
our computers.
Up until that point, no one I knew was doing that, but little did I
know AI
Literary Review and Ensemble
Park were shaping up at the exact same time,
which proves to me that there was an obvious need to ask, "With these
tools, what is responsibly
possible?" I was just starting my PhD program, and doing a summer
fellowship planning the online ELO
2024, and it seemed
like I'd have the advising/institutional support to do this in the Texts & Technology program at
the University of Central Florida, compared to my previous (great, but more
traditional) institutions.
KP: How do you feel it’s going? And what are your hopes for its future?
PD: Honestly, I feel kind of honored and shocked every submission period with
the variety and volume of work we get in the queue. It's not 6,000 poems a
period, like it can be at an historic magazine, which is a good thing. I don't
know what I would do if it were! But luckily re•mediate's mission and vision are specific enough that it's
really easy to turn down good work that could definitely find a home at another
magazine.
Being plugged in with the Electronic Literature Organization from the
beginning has been a huge part of there being a community interested in the
work, and interested in being a part
of the magazine—Monica Storss and I met over a Second Tuesday Salon Zoom, and I
was delighted to bring her on to edit re•mediate
issue 3 [Next
Realities].
I had done a little of my own virtual-reality poetry at one point, but I
didn't have the breadth of vision Monica has for extended-reality storytelling,
and it's been wonderful to share the reins with her, and a really lovely
collaboration. It makes me excited for the future of the magazine, because as
long as folks submit work that applies to re•mediate's
intentions as a publication, I'll be happy to put together seven or eight
pieces every four months for our readers, and for future researchers who might
wonder what the heck was going on
with literature and AI in 2025.
As a digital humanist, I have to hold this project as open-handedly as
possible. I was telling another collaborator over the phone today that if this
were to go away tomorrow, I would be happy to sunset the project (as long as
it's archived well). I've been taking a directed reading course on
literary-periodical/little-magazine history this summer, and it's been really
refreshing to know that the poverty of the magazine, its ephemerality, the kind
of stubborn willingness of little-mag editors to take risks, are all baked into
the definition of the term—that in some ways it's the journals and magazines we
think of as literary institutions that are the anomalies.
KP: Tell me about the process notes you require for each submission.
PD: That’s something that AI Literary Review, as well as re•mediate and another magazine, Ensemble Park, all require. I think it
comes out of Taper’s requirement of making the code
open-source: the process of becoming a poet and developing your taste is kind
of each person’s to undergo, but the process of coding or leveraging
computational tools is a barrier-to-entry that we like to leave unlocked for
anyone to play with.
I require process notes partially for that archival process (knowing
how these pieces came into being) and partially because of that mission to
inform/educate folks on the methods and ethics of working with
"AI"—and to demystify that nebulous, loaded term. Most of my
contributors are homebrewing a large language model, doing found-language work,
or writing around the topic of life online, not just "prompting chatGPT to
write them a x about y." However, the most interesting instances of that so far was the screenplay by Daniel
Sherman in issue 3, which incorporates it almost as a character, and Dan
Power's satire made by prompting a legal
chatbot to summarize purchasing the moon.
Neither of those are copy-pastes of AI, and I haven't got much of that
in the queue, either. If I did, I'd look at the process note to get a sense for
the vision of the submitter. Because, in the interest of critique, sometimes it
is interesting to let AI speak for
itself. From Black Warrior Review, I
inherited a sensibility that says in most
cases, the writer's work is complete when they submit, so I don't overhaul
stories or poems before they go out unless there's a word-count issue on a
piece I really value (we take prose under 4,000 words). Ensemble Park goes into a lot more detail and has a more rigorous
process-note editing process, which I really value as well.
I talked to Dan Power of AI Literary Review about this in preparation for a panel we were on
together at ELO 25 in Toronto. A discerning submitter will probably notice the
slight differences between the submission guidelines of his magazine, mine, and
Ensemble Park, not just the language
choices (whether we call these works of “human–computer interaction,”
“AI-generated,” or “computer-assisted”) but also about the degrees of
authorship in a piece. The other existing schools of thought around authorship,
like the automatic writers, Oulipo, found poets, combinatorial writers, centos,
renga, and so on, raise this question. I wonder if there’s a matrix or multiple
Venn diagrams to be made?
These questions are often at the heart of the little two-axis plots I
include in re•mediate in each
Editor’s Note. I don’t think there’s going to be a single field on which these
questions can be plotted, per se, because the conversations we’re having on AI
are multi-dimensional: the above graphs are neutral on the environment,
politics, and kind of definitionally loose. Where does “found” poetry land
here, for example?
KP: I love those two-axis plots. It feels like a more useful way to talk
about the relationship between the human author and computer/digital tools than
a taxonomy. The feminist in me is wary of taxonomies, and the need to name and
label. Of course, we need a language to describe these operations in order to
have meaningful conversations. I find it instructive. I’ve learned a lot about
what goes into publishing a physical text by studying the various stages and
participants in the creation and publication of a text that involved computer
generation. But I whole-heartedly agree with you that the matrix doesn’t take
into account environment, politics, gender, sexuality, class, etc.
PD: I'm so glad for what you
said about taxonomies—I think we're right to be wary of them, and I hope that
the graphs that I include with each Editor's Note aren't taken by readers to be
definitive. For me, they're gestures at the varying dimensions available to us
in literary criticism, like the Franklin
spectacles in National
Treasure. Every time I put
out a call for submissions at re•mediate,
I make a point of saying that we invite criticism about computer-assisted creative writing. As of
yet few have taken up the charge. I hope that, whether it arrives in my
inbox or not, someone will take re•mediate
to task for the work that I've published so far, will engage with it
creatively or critically. One of my favorite things about collaborating with folks,
whether it's with Monica Storss for the guest-edited issue, with you on this
interview, with contributors, or with the editors of AI Literary Review and Ensemble
Park on the upcoming ELO panel, is seeing this project through others'
eyes!
KP: At ELO, we mused about a way to estimate the emissions of a piece
of AI art/writing, based on the process note, and find a way to offset this.
PD: It was a fun thought experiment between me and you and Kyle Booten, of Ensemble Park, the idea of a kind of
carbon credit being the submission fee for an issue of re•mediate, or maybe instead of a dollar amount, someone could
offset their use by not driving for a week, or bumping their air conditioning
down a few degrees, or fasting from meat for a month. For me, that line of
thinking is really interesting, and I think the world would benefit from an
open-source "emission estimator" for the use of AI similar to the Microcovid Project. But I also go a step further, to the
provocation of how much data about your life you include in that calculation. I
live in Florida, in a Duke
Energy zone, and while
there's a lot of solar here, there's also a huge energy cost to the
deforestation of the state, the air conditioning, the amount I have to drive to
do anything in Orlando, and the
provenance of my food.
If we were to make emissions offsets or climate credits of some kind
the submission fee, for some of our submitters, that would make submissions
free, right? It seems that way on its face, unless you were to count the meals
you ate the day before you wrote the piece (which powered the work by
extension), or the amount of time you spent on the computer putting together
the submission, and so on. Maybe the initial version of the AI Art Emissions
Estimator is just tuned to the production process itself, but as an essay like
Wendell Berry's "Why
I Am Not Going to Buy a Computer" suggests, personal computers or
laptops were in the hotseat environmentally that AI is now 30 years ago—in
which case the paper/emissions required to switch to mail submissions and print
publishing would be implicated too. I grew up without air conditioning,
drinking from the tap, so it's not so hard to keep my apartment comparatively
warm and not buy bottled water. But I drink coffee, an imported good, every
day. That certainly powers re•mediate,
especially in April, July, and November.
Maybe it's because I've been reading e. e. cummings, Carl Sandburg,
the book of Jeremiah, and Kierkegaard
this summer, but at this moment I think that the humanists who are best
equipped to deal with the world we're in and the one to come are the ones who
can look clearly at all of the ways our fault-findings self-implicate on labor,
ecology, and humanity. As a Floridian, my concern about AI's impact on the
environment must reflect back on my
state's insistence on air conditioning while we disincentivize home solar
installation and sell a third of state land to developers. My desire for
environmental justice can't just be "my entire technological life as is, but I abstain from AI"—I have to
interrogate how vertically integrated the excess is.
To be honest, I don't use AI much in my day-to-day, and I ideate as
much as I can on paper because I want to prompt in as targeted a way as
possible and build a coding literacy for myself. I can only bring that
magnifying glass to bear on my own existence, though, because I fundamentally
believe that my humanity is not grounded in whether a computer can do what I do
or not, or in whether this magazine lives long, or whether I have 100 percent
certainty that I'm living the absolute most ethically I can in this exact
moment. To spend more time in critical remove from my own body, my own life,
and my own art, rather than in them,
is maybe a more inhuman thing. I want to know by engagement, by reading,
by listening, by living, by making—in some
ways, "living the questions" has disillusioned me from certain uses
of AI quicker than if I fretted over feathers on ethical scales.
P.D. Edgar (MFA, MA) is a Ph.D. student in Texts
& Technology at the University of Central Florida, where he studies poetry
culture, social media, and electronic literature. PD grew up between Managua,
Nicaragua, and Central Florida. As an experiment and extension of his research,
he started re•mediate, a lit mag for
computer-assisted creative writing, in 2024. He previously worked as Art &
Design Editor of Black Warrior Review
and as Digital Archive Associate for Kenyon
Review; his creative work is available at Ghost Proposal, EKPhrastic Review, SAND Journal, and AI Literary
Review.
Katherine Parrish is a researcher, teacher and apologist
for experimental poetry. In the early 2000s, she began experimenting with
digital poetry and poetics, presenting her findings at the inaugural E-Poetry
festival, at SUNY Buffalo, the New Media Poetries conference 2002, where she
was an invited speaker, and in various publications,including Object, The
Cybertext Yearbook, and Canadian Notes and Quarterly. She recently presented at
the Electronic Literature Organization Conference (York University, July
2025.) Katherine has taught in the high
school English classroom in the Toronto District School Board for the past 25
years.
