I was struck by this passage in an article about the harms of so-called #AI in academia, literature, and other areas (found here https://theconversation.com/ai-generated-text-is-overwhelming-institutions-setting-off-a-no-win-arms-race-with-ai-detectors-274720 while noting I strongly disagree with the author calling LLMs a democratization of access, and I also note the article has other issues like confusing LLMs and machine-learning technology in general):
"[Fiction outlets] may refuse AI-generated work, but this will come at a cost. It’s unlikely that any human editor or technology can sustain an ability to differentiate human from machine writing."
This assumes that fiction submission and #publishing will remain fixed in response to a changing technological and social landscape, passively taking all submissions while launching largely futile countermeasures against LLM-generated entries.
But what if segments of publishing were to transform themselves so that it would CEASE TO MATTER whether entries are generated, while still having a high chance of winnowing out slop without overwhelming editors or forcing them to turn to LLM-driven non-solutions?
One way might be to hold semi-open pitching/brainstorming sessions for magazine issues or anthologies, where, say, the first 50 people in the door are briefly interviewed or submit pitches for what kind of stories they have written or can write for the issue/anthology, while freely discussing with the editor(s) what contribution they can make and how their entry fits into the overall picture of the collection. In this process the fiction issue or anthology can start taking shape through the ideas generated during these discussions and pitches, and the candidates who show a good understanding of their own works/ideas and the collection could be invited to submit their works. If these won't be enough submissions the next batch of e.g. 50, maybe fewer, are invited to pitch, and some of them to submit. Some applicants might be left in reserve in case some of the actual submissions are duds or slop.
In this way writers who wish to submit to an issue or anthology can be pre-screened before submission while having an active voice in editorial decisions and helping set the overall direction of a collection. The process of pitching and discussion itself can filter out empty, banal, or badly-fitting ideas. If LLM-generated pitches and responses can actually pass that process, more power to them, but given the characteristics of LLM-generated text it won't be terribly likely or widespread as long as the editors themselves have original and specific taste and vision. And limiting the number or pool of participants (new authors only, published authors only, members of marginalized communities only, etc.) will prevent editors from being overwhelmed or burning out, while still leaving room for discovering new contributors.
Will this kind of process, just one of many possible adaptations, change the nature of the submission and publishing process? Undoubtedly. Under the particular method outlined above, the submission process will become more like flash events and extended editorial/brainstorming sessions. It could work out great for some outlets while being terrible for others. Still, change is kind of the point and it's clear publishing can't stand static and defensive in the face of a sea change. I hope to see a lot of new initiatives and adaptations going forward.
