SETH W. JAMES

Nebula Reaction: So-called “AI” and LLMs have No Place in Fiction

January 9, 2026

By Seth W. James

The Nebula Awards (and the Hugo and the Locus and the . . .) rarely attract my attention except for a few minutes the day after they’re announced, just to see if anyone I know took a prize.  But with the recent controversy involving the Science Fiction and Fantasy Writers Association (SFWA) changing—and then changing back—the qualifying criteria for the Nebula, I thought I would put down a few thoughts.

The crux of the matter is, as so many things are these days, the use of so-called Artificial Intelligence and Large Language Models (AI and LLMs, respectively).  The SFWA board had released updated qualification criteria for works submitted for consideration for the Nebula Awards to include those works created with the use of AI/LLMs—and, not surprisingly, artists everywhere responded with entirely justified outrage.  To be clear, there is not now, nor has there ever been, Artificial Intelligence: AI remains what it has always been, a matter of science fiction.  What is disingenuously marketed as AI is not intelligent and is, in fact, based on technology that has been around since the 1970s.  LLMs, while less disingenuously named than AI, are largely the same thing, simple predictive models executed on a large scale through the use of datacenters (which, incidentally, are helping to worsen climate change, increase fresh-water scarcity, and raise energy prices).  Both are, in essence, plagiarism devices because their databases were created by scraping content from creators, usually without notice, consent, or compensation.

A great deal has been written about the evils and incompetence of so-called AI, so I don’t feel that I need to reiterate it at length here.  What I did want to address, however, was the SFWA board’s changes to the criteria, both in substance and in process.

In substance, it completely baffles me that the people sitting on the board, who I assumed were fellow artists, would condone the use of AI/LLMs in the “creation” of a work passed off as original art.  As the Anthropic (or misanthropic, if we were genuine about it) case plays out, as some tiny, token compensation is paid to authors whose works were stollen, how could the SFWA board possible think that it would be okay to legitimize the plagiarism devices created through Anthropic’s (and other company’s) theft?  To put it simply, the AI companies stole our work, broke them down into their component parts, and then cobbled them together to sell as if a new product.  Why would any artist agree that such products constitute art, let alone that they should be considered alongside actual, human-written works for the Nebula Awards?  Naturally, I have to wonder about the board members’ current and future works, if this seemed acceptable to them.  That being said, it is commendable that they very quickly changed course and put the qualification criteria back to what they should have been.

In process, though, this all could have been easily avoided if the board had not taken unilateral action.  For something as important as AI/LLMs and the ongoing assault on artists’ autonomy and ownership of their work, any changes to the SFWA’s policies should have been put to a vote by the entire organization.  Yes, of course, the board should be able to change certain policies without the entire SFWA having to vote on it; something like when voting begins or ends or the RSVP for the in-person ceremonies are administrative matters that we expect the board to handle on their own.  Policies that affect the definitions of work, from the Nebula categories to qualifying venues/markets to authorship criteria, fundamentally define the SFWA, in its purpose and identity.  Such policies are statements about the organization’s beliefs and should originate strictly and exclusively from the membership’s consensus.  A better grasp of the scope of their decisions is needed from the board in future.

Lastly, I think one of the causes of this mistake is the continued absence of the SFWA members’ website.  After the new board was elected, one if their priorities was to overhaul and update the members’ website.  It has been about a year, if I remember correctly, and the website is still down.  Previously, the members’ website forum was the place where we connected and, at times, debated the issues of the day.  I remember controversies about the previous board’s actions in regard to volunteers generating a great deal of discussion, as well as other topics such as TOR’s saturation in recent award cycles, new categories to receive awards, and also news around small presses that had folded and what would happen to their backlists.  Without the members’ website, we have lost that virtual townhall and the discussions that had taken place there.  Yes, there is a Discord server for the SFWA, but it is highly problematic: for one, access to the Discord is not automatic, members need to discover its existence themselves and then contact someone to get an invite; Discord is also crawling with state actors, is rumored to be hacked constantly by criminals, and is likely selling every word typed into it to AI companies, to populate their datacenters.  So, no, Discord is not a suitable alternative and certainly not a replacement.

In closing, I would just like to say to the board that though I applaud their rapid response to artists’ outrage at the inclusion of genAI theft content in the Nebulas, it is a decision that should not have happened in the first place.  Please, in future, put such important matters to a vote by the whole SFWA.  And, to facilitate this and other worthy discussions, bring the members’ website back online as soon as possible.