Skip to content

Fields for metadata about AI-generated specimen information #140

@ssjos

Description

@ssjos

What is your idea for a happy hour topic?

AI is increasingly used in digitization workflows, such as for automated specimen label transcription. The transcribed data may or may not be human validated. How should metadata about AI, such as whether AI was used, what model, for what purpose, if or how data was validated, etc, be captured along with specimen data? How do we ensure our future data are correctly tagged with provenance information to prevent false confidence or AI-training loops where AI is trained on AI-generated data? How much metadata is too much?

Who could be involved in leading this happy hour?

Sara Sjosten and Crystal Cortez, CAS

Are there any timing considerations?

June or July 16th preferred for summer, Aug-Oct not good

Next steps

This issue captures a happy hour topic idea. Once you are ready to finalize an idea and put it on the PDWG happy hour schedule, please create a new issue using the Plan a happy hour template. Then, close this issue.

Metadata

Metadata

Assignees

No one assigned

    Labels

    PDWGAll issues receive this label so that they can be aggregated in the GEO OSE GitHub projectideaThis is an idea for a happy hour

    Type

    No type

    Projects

    Status

    To do

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions