We Don’t Use AI

Choice of Games is the world’s largest publishing house for interactive novels. We don’t allow authors to publish work generated by AI or large language models (LLMs).

  • We pay our authors to write words themselves
  • We pay our artists to make art themselves
  • We review everything submitted to us

We have three publishing labels: Choice of Games, Heart’s Choice, and Hosted Games. We don’t allow AI on any of our publishing labels.

Nobody knows who owns AI-generated text

Due to legal uncertainty around the copyright of AI-generated content, all submissions, including all artwork, must be created by a person or group of people who have the legal rights to publish it, and we have to provide them credit in the Credits.

AI generates slop by design

LLMs are trained to predict what the “next word” would be a sentence. Their training objective requires the LLM to keep surprise to an absolute minimum.

Unsurprising, predictable output is the defining characteristic of AI slop.

This will not improve with more processing power or more training data. For the same reason you can’t find funnier jokes by polling a larger and larger number of people, the architecture of LLMs is just going to give you increasingly unfunny, unsurprising text. The more resources AIs consume, the less surprising they’ll be.

Don’t let AI-based grammar checkers rewrite your prose

Grammarly and other AI-based grammar checkers don’t just flag issues; they offer to “fix” them for you, too. Often, their “fixes” are wrong.

But even when they’re working at their very best, Grammarly is a tool to homogenize text, to make your words and ideas sound like everybody else’s. Those aren’t words worth paying for, and they’re not yours to sell.

When exact word choices really matter, we must choose the words ourselves.