Change Graphite's AI contribution policy to a full ban on AI contributions.

[bmakin-chapman]

I propose that Graphite should categorically reject any and all AI contributions. My understanding is that the team hopes that Graphite will be able to be incorporated directly into other projects, and the current AI contribution policy effectively limits users' ability to do so.

Explanation

Graphite currently has very limited acceptable usage of AI, according to its AI contribution policies:

Non-agent AI tools may assist with debugging and tab-completion of single lines of code you would have otherwise written yourself. This does not require disclosure.

AI chat tools (not agents) may help you generate small (sub-40 line) snippets of code that you manually copy and paste, provided that you carefully review every line to ensure it is consistent with how you would have written it yourself. This requires disclosure.

My understanding is that the team behind Graphite hopes that Graphite will be able to be incorporated directly into other projects. I personally am a professional game developer and believe that Graphite can become a powerful tool for designing user interfaces for my games. With Graphite's permissive license, I could allow my art team to design these UIs in Graphite with full use of the node editor and incorporate Graphene into my game rather than re-implementing any interactivity by hand for each new Graphite project.

However, Graphite's allowance of AI contributions may cause issues for commercial sales of games. Take, for example, the popular video game storefront Steam. Steam requires disclosure of AI content.. AI content remains controversial among video game developers and players, and a developer like myself may be hesitant to incorporate Graphene in a project that is otherwise untouched by AI content. Even putting aside my own ideological concerns, I would have to mark by game as containing AI generated content which may dissuade potential players.

If a developer incorporates Graphene and does not label the game as containing AI generated content for whatever reason, that developer risks being reported for undisclosed AI generated content.

The only remaining option for a developer in that situation is to re-implement the functionality of either their specific Graphite projects or of Graphene in general, which seems like an unproductive use of time when Graphene itself should be available for use.

I have focused on the Graphene node engine for this post, but the same reasoning applies to all of Graphite. In this interview, Keavon Chambers and Dennis Kobert mention the possibility of reusing the Graphite editor for another application. (At 7:20 They pitch repurposing it for a digital audio workstation.)

Seeing as Graphite already has such a limited allowance for AI contributions, I believe that it would be of very little additional burden to ban AI contributions entirely. The restrictive AI contribution policy suggest to me that the developers are already hesitant about AI code, so I urge the team to take this additional step.

[Keavon]

Thanks for posting your thoughts. Regarding your concerns about Steam's policy, that does not apply to AI-generated code, only to assets. Your game engine and other supply chain components already include, and will continue to increasingly be built from, AI generated code, as will the entire rest of the software ecosystem in the years to come, whether we like it or not.

The reasoning for our currently posted policy you've quoted is due to a shortage of maintainer bandwidth, where an external contributor using an AI agent would add yet another layer of lossy communication to the process. As maintainers, we do not have the resources to code review someone else's AI output without knowing the level of due diligence and skillful operation involved in that process. Most PRs we have received in the past low-quality and the availability of AI agents has lowered the floor for unskilled beginners to spam us with unusable PRs. So far, the policy has been an effective deterrent to the quantity of those low-quality submissions that were previously becoming a significant time sink. As the landscape of tools evolve and our reviewer bandwidth changes, we may also revisit the policy and loosen it or optimize its focus to ensure it is still having its intended effect of filtering out lazy and low-effort work.

[bmakin-chapman]

Whatever policy your team ultimately decides on, thank you for taking the time to respond.