I guess this is (much closer to) what Bruce Perens wants instead of the (current) OSAID draft:
> Nor is one necessary, because the original OSD works for AI. You need to treat the training data as source code, and you need to apply the rules to both pieces: the underlying software of the machine learning system, and the training data. This can get complicated because sometimes the training data expands in real time as the system gets more queries.
https://www.linkedin.com/feed/update/urn:li:activity:7254500...
