In discussing the metrics we have been working on for AI use and consent (#66 and #68 ), we ended up taking so many different perspectives on this that it was deemed worth breaking them down into a taxonomy.
I was assigned to first-pass this taxonomy for the AI consent metric as a parallel to the Use taxonomy (#76) - allowing us to reframe the current metric in a smaller scope to get it published and iterate.
- Consent Metric - Established consent (location/binary)
- As a contributor, I want to know whether the community has established rules around AI and where I can find them
- As a contributor, I want to be able to easily find and navigate the applicable AI policies that may exist at multiple levels (i.e. project policy, community policy/CHAOSS, governing body policy/LF, state/national policy, etc)
- As a community architect, I want to know how many of the communities I oversee have implemented policies around AI
- Consent Metric - Supervision level/ keypolicy features
- As a contributor, I want to know where a community’s policies stand/what key features it has at a high level (i.e. no AI allowed, “human takes responsibility/is accountable”, disclosure required, limited unsupervised use allowed, fully unsupervised use allowed, etc)
- As a user of open source, I want to be able to evaluate the potential risk of choosing one project as a dependency over another based on how well their policy position surrounding the unsupervised use of AI aligns with my own
- Consent Metric - Actions/tool types allowed
- As a contributor, I want to know which tools/models im allowed to use in/introduce to a community (like a notetaker bot, etc)
- As a community architect/company who uses lots of open source, I’d like to be able to quickly determine which of the communities I depend on would allow a specific tool I want to use
- Seeking more User stories
- Consent Metric - AI use allowed
- as a contributor I want to be able to quickly learn what ways a particular community allows AI use so I can know whether my contributions are a good fit for the community (vs a self-maintained fork)
- User story “as a maintainer I worry about”
In discussing the metrics we have been working on for AI use and consent (#66 and #68 ), we ended up taking so many different perspectives on this that it was deemed worth breaking them down into a taxonomy.
I was assigned to first-pass this taxonomy for the AI consent metric as a parallel to the Use taxonomy (#76) - allowing us to reframe the current metric in a smaller scope to get it published and iterate.