On Mon, May 4, 2026 at 8:42 AM Jan Ainali via Wikimedia-l <
[email protected]> wrote:

> Thank you Luis for your thoughtful reply. Your historical examples are
> particularly illustrating because they show compromises with a long term
> goal of "breaking free". They are stories of building new tools rather than
> locking themselves into the existing technologies. In our context that may
> mean that we will build upon and borrow where we can, rather than being
> just a consumer of the Big Products of the day.
>

Sure, and/or using our weight to ensure that the Big Products bend to our
will and not the other way around.

If we have any red lines in this compromise it surely is at a privacy
> preserving aspect.
>

I agree. Also, by focusing on a specific aspect (that is not covered at all
by the OSI definition!) we have a more interesting and useful and concrete
discussion.

For example, I am working on an LLM-backed tool that helps editors with
citation data. It uses a proxy to not expose editor IPs or user IDs to LLM
vendors, though I suspect it means CloudFlare has a little bit of
traceability—and you’re right that I should look more deeply at that.

The example with the Wikipedia library is interesting because even if we
> are using works that are non-free as sources and the ethics of many
> publishers deserves to be discussed, arguably these authors at least
> knowingly (albeit begrudgedly) have some sort of deal with them. It's also
> not unreasonable to think that many of the authors are happy being cited
> and that a citation on our projects bring some, but perhaps mostly a low,
> value to them. That's a stark contrast to the ethics of how the "Frontier
> models" are being made.
>

Two things:
(1) “arguably” is very arguable for academics, who must “publish or perish”
with the big journals and so don’t really have a lot of choice in the terms
their publishers apply.

(2) are you suggesting we should stop using Google Book Search and Internet
Archive? Authors gave the same amount of consent to those book scanning
projects as to the big LLMs.

>
_______________________________________________
Wikimedia-l mailing list -- [email protected], guidelines at: 
https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines and 
https://meta.wikimedia.org/wiki/Wikimedia-l
Public archives at 
https://lists.wikimedia.org/hyperkitty/list/[email protected]/message/6DIZ5CFFXJSL3LD3E5UUPC2VN6M3IC7R/
To unsubscribe send an email to [email protected]

Reply via email to