Research & translation app based on sc-data

[agilgur5]

@sujato

“Huh, I haven’t had to contend with any new AI apps using my stuff for a few days now”.

Oh I didn’t know you were actively pursuing them; in that case, one called “FoJin” presented itself in SC’s own issues ~5 months ago, asking to collaborate, ironically. I explained to them your position, but I don’t think they ever removed SC content: Cross-reference integration with FoJin Buddhist text platform · Issue #3584 · suttacentral/suttacentral · GitHub. Their source repo is linked in that issue as well

@sujato

he wasn’t actually using AI. Perhaps this has changed.

I’m not sure when you last spoke with Ven. Bodhirasa, but there’s lots of signs of LLM coding usage in the repo that are attributed to his handle. For example:

• AGENTS.md is instructions for LLM agents and is currently entirely attributed to him, going back at least a year. The file itself has various signs of being written by an LLM too, like pervasive em-dashes, several horizontal separators, etc.
  • (for reference, the mere existence of the file may not mean usage, as sometimes the content is telling agents to not work on the repo or exists to poison agents)
  • GEMINI.md also symlinks to it
• CONTRIBUTING.md has some similar and some different signs, like lots of bullets with bolded first words, etc
• A commit from 2 weeks ago was “Co-Authored-By: Claude Opus 5”

The Russian fork, whose integration thread you had commented on:

• Also has various commits “Co-Authored-By: Claude”
• The SC Exporter PR has LLM usage signs as well
  • But the code is specifically on its side, not SC’s, and the output is filtered of any AI generated translations. That is to say, the code was written with LLMs, but not the translations, which the contributors specifically ensured.

I’m not too sure about DPD itself, but as far as I understand, folks were abiding by the stance in your original AI-free post:

@sujato

That does not mean that there cannot be any AI behind the scenes, for example as coding assistants, search, and the like. Such things have drawbacks and benefits, and it is practically impossible to avoid them completely at this point. So the ban affects actual presence of machine-generated text.

Although I’m not sure how much your stance has changed since then. Folks have been asking for a bit for an explicitly written policy on all things AI, including definitions and disambiguations of types of AI (like CV, OCR, TTS, ML, search vs later GenAI, LLMs), usage (coding vs translations), scope (3rd party code and apps vs 1st party), etc in order to be able to explain the boundaries and differentiate, since, as you know, the terms have gotten messy these days.