The problem is that the “different voices and perspectives” that OSI is bringing “into the open source community” behind the scenes are not addressing the issues reported for the latest drafts and RC1 of the “Open Source AI” definition.
And given that OSI’s “inclusive approach” to the “co-design process” included Meta and enabled Meta employees to cancel other voices to obtain the exclusion of training data, to any external observer the whole process look severely biased towards interests that have nothing to do with open source.
And maybe I’m connecting the dots in the wrong way, but guess what?
The interests that OSI is trying to defend so desperately are totally aligned with those of their most generous sponsors right now: Google, Microsoft and GitHub (aka Open AI owner), Intel (@Mer employer) and Meta itself (who blatantly exploited the “co-design process”).
And yet such Big Techs’ arguments in support of the currently flawed definition stay behind the scene.
Why?
Don’t get me wrong, if OSI is so fond of Big Tech arguments, I’m sure they are pretty solid and valuable.
That’s why we are asking in public how they address the issues we reported.
Because to an outside observer, secret arguments, however valid, are indistinguishable from missing arguments.