Re: strongly disagree with OSAID 1.0-rc1
I am in violent agreement, as any DD who understands the issues around
AI and Open Source should be.
AI systems are software and (to quote Bruce Perens, DFSG & OSD author
and OSI founder) the training data *is* the source, so the OSAID is
fundamentally incompatible with the OSD (which he recommends we use as
is, applying it to both code and data). I expect it's only a matter of
time until they start chipping away at the very foundation of Open
Source after a quarter century of stability too.
I'm late to the party after providing early input, trusting the OSI
and presuming sanity would prevail — a mistake I won't make again. I
see them railroading this through on Monday, perhaps for no other
reason than to perpetuate their recently significantly expanded budget
and influence, as a credible threat of converting future Open Source
to Freeware. It poses an existential threat to our pAI-OS project at
the non-profit Kwaai Open Source Lab I volunteer at, so we've been
very active in pushing back past few weeks.
The OSI's chosen "co-design"[1] process, which essentially amounts to
"Do Your Own Research" ("We believe that everyone is an expert based
on their own lived experience, and that we all have unique and
brilliant contributions to bring to a design process.") excludes the
existing Open Source community and exploits the minorities who agreed
to participate in our absence by denying them the very data they need
to study and modify AI systems while dealing with bias.
The voting irregularities I discovered on audit aside[2], they're
driving rather than following the process themselves ("We see the role
of the designer as a facilitator rather than an expert."), and
couldn't so much as describe it on the last town hall before releasing
it and going on holidays through the end of the year. They also
unilaterally closed a proposal to achieve consensus for the 1.0
release[3] without explanation, which would have eliminated the
proprietary ("including for fee" like NYT articles) and closed (think
FB/IG social graph) datasets, while leaving unlicensed public datasets
(like Common Crawl) in the mix, per the status quo in many AI
projects.
I look forward to seeing your GR draft and hope my key's back in the
keyring in time to sign it after a long absence,
- samj
[1] https://jods.mitpress.mit.edu/pub/costanza-chock/release/4
[2] https://discuss.opensource.org/t/we-heard-you-lets-focus-on-substantive-discussion/589/2
[3] https://discuss.opensource.org/t/proposal-to-achieve-consensus-for-the-1-0-release/653
On Mon, 21 Oct 2024 at 20:46, Mo Zhou <lumin@debian.org> wrote:
>
>
> On 10/20/24 12:21, Christian Kastner wrote:
> > I fully concur.
> >
> > Are [3, 4] the only options we have to voice this concern?
>
> Those channels are moderated. And they recently blocked some people
> repeating the importance of the data issue.
> > Although I'm confident that most Debian Project Members would also
> > oppose this and an official statement by the Project would be notable, I
> > don't think there is enough time until October 28th to organize something.
> >
> > (Funnily enough, this *is* something that would be worth a GR, I think).
> I think posting a GR would be a good way to show disagreement and
> express concerns. I'll prepare a draft.
> >> [3] - https://hackmd.io/@opensourceinitiative/osaid-1-0-RC1
> >>
> >> [4] - http://discuss.opensource.org/
>
Reply to: