An update on this thread:
On Fri, Jun 20, 2025 at 10:25 AM Leila Zia <lzia(a)wikimedia.org> wrote:
>
> In terms of next steps, I am accountable to bring together the writers
> of the different pieces of the current draft to complete a more
> detailed write-up of the event, that will include more of the
> collective technical conversations we had at the event. Our plan is to
> circulate the document both with attendees as well as some researchers
> who have been part of conversations about AI and knowledge commons but
> weren’t able to attend the event. We will then publish the document
> from the event on Meta-Wiki. We aim to do this in September 2025.
>
We have made significant progress on this front, however, the reviews are
not completed, yet. The updated timeline is October 2025.
If you have questions or concerns, please let me know.
Best,
Leila
--
Leila Zia
Head of Research
Wikimedia Foundation
Hi Maryana,
On Thu, Feb 29, 2024 at 9:04 AM Maryana Iskander <miskander(a)wikimedia.org>
wrote:
> This message will be translated into other languages on Meta-wiki
> <https://meta.wikimedia.org/wiki/Special:MyLanguage/Wikimedia_Foundation_Chi…>
>
> العربية • español • français • português • Deutsch• 中文
> <https://meta.wikimedia.org/wiki/Special:MyLanguage/Wikimedia_Foundation_Chi…>
>
> You can help with more languages
> <https://meta.wikimedia.org/wiki/Special:MyLanguage/Wikimedia_Foundation_Chi…>
>
> Hi everyone,
>
> Since joining the Foundation I have tried to regularly write to you
> <https://meta.wikimedia.org/wiki/Special:MyLanguage/Wikimedia_Foundation_Chi…>
> here and elsewhere, and I wanted to share a few updates since my last
> letter. In October 2023
> <https://meta.wikimedia.org/wiki/Special:MyLanguage/Wikimedia_Foundation_Com…>,
> I reflected that we were in a period of compounded challenges across the
> world with escalating wars, conflict, and climate reminding us each week
> that global volatility and uncertainty was on the rise. That feels even
> more true now. My instinct then was to ask us to make more time to talk to
> each other and to try and pull closer together. This feels even more needed
> now.
>
> [...]
>
> Finally, our human-led values came up in several conversations about
> Wikimedia’s role in shaping the next generation of artificial intelligence,
> a topic of ongoing discussion in the world
> <https://www.nytimes.com/2023/07/18/magazine/wikipedia-ai-chatgpt.html?unloc…>,
> in our communities <https://meta.wikimedia.org/wiki/Talk:Future_Audiences>
> , and at the Foundation. This is complemented by ongoing discussions
> about the role of AI-generated content on our platform by various project
> communities.
> <https://en.wikipedia.org/wiki/Wikipedia_talk:Large_language_model_policy)>
> A recent effort to contribute to a shared research agenda on AI
> <https://meta.wikimedia.org/wiki/Special:MyLanguage/Artificial_intelligence/…>can
> be found here – including the need for more research to understand human
> motivation to contribute to the knowledge commons – it was created by a
> small group working in the open who rushed to publish a ‘bad first draft’
> that will benefit from more input.
> <https://meta.wikimedia.org/wiki/Special:MyLanguage/Talk:Artificial_intellig…>
>
> Thank you for acknowledging the limitations of this document. As we also
noted in the Signpost
<https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2024-03-02/News_…>
at the time, there was indeed some consternation about the lack of
involvement of the volunteer community:
*"While the announcement appears to be speaking on behalf of 'volunteer
contributors', the 'Wikimedians' involved in drafting the document appears
to have consisted exclusively of Wikimedia Foundation staff (largely from
its Research department), according to the attendee list."*
I have to ask though, are there still plans to solicit wider input on this
draft agenda, or at least incorporate more from the numerous related
discussions on AI that have been happening across the movement over the
last several years? (At
https://meta.wikimedia.org/wiki/Artificial_intelligence , some editors
including myself have been trying to keep a list of relevant links, but
it's surely not complete.) Again, I appreciate that your post here invited
"more input" on the agenda's talk page. But it seems that only a single
topic was added there afterwards, and in any case no content updates
<https://meta.wikimedia.org/w/index.php?title=Artificial_intelligence/Bellag…>
have been made to that "bad first draft" since February 2024.
Relatedly, given that the document states that *"Our hope is that many
researchers across industry, government, and nonprofit organizations will
adopt the final research agenda to help support and guide their own
research"*:
Are there plans to solicit input from such external researchers on the
draft? And once this research agenda is finalized, does the Foundation plan
to bring it to their attention? It doesn't seem to have made such efforts
yet, e.g. I can't find any mention
<https://lists.wikimedia.org/hyperkitty/search?mlist=wiki-research-l%40lists…>
of
it on the Wiki-research-l mailing list (CCing it now).
I thought that maybe this Bellagio document had been a tangential one-off
to make use of an external funding opportunity, and had been abandoned
afterwards. But then I saw that more recently Selena highlighted it in
the "Reflections
on 2025 from the Wikimedia Foundation Executive Team"
<https://diff.wikimedia.org/2025/01/29/reflections-on-2025-from-the-wikimedi…>
(as
the only concrete outcome regarding AI mentioned in this entire overview of
WMF accomplishments "Over the past year").
Regards, Tilman ([[User:HaeB]])
> [...]
>
> Maryana
>
>
> Maryana Iskander, Wikimedia Foundation CEO
>
>
Hi Tilman, Hi wikimedia-l,
For those of you that I haven’t met, I’m Leila Zia and I’m the head of
Research at the Wikimedia Foundation. I’m happy to take your questions
as I’ve been involved in co-organizing the Research event at the
Bellagio Center and I am responsible for some of the next steps
associated with it.
The primary role of the Wikimedia Foundation for this event was the
role of a convener and facilitator to bring together a first group of
experts with different backgrounds and expertise who are invested in
the knowledge commons. The goal of the event then was to support this
group of people to arrive at a draft research agenda for AI and
Knowledge Commons. We explicitly told attendees that this was not
about developing a research agenda and next steps for AI on the
Wikimedia projects, or ideas for AI on behalf of volunteer
contributors; but rather to explore key areas of research in open,
commons-related digital spaces as a whole.
In terms of next steps, I am accountable to bring together the writers
of the different pieces of the current draft to complete a more
detailed write-up of the event, that will include more of the
collective technical conversations we had at the event. Our plan is to
circulate the document both with attendees as well as some researchers
who have been part of conversations about AI and knowledge commons but
weren’t able to attend the event. We will then publish the document
from the event on Meta-Wiki. We aim to do this in September 2025. Once
the document is on Meta-Wiki, we can create a space for pointers from
other conversations from different knowledge commons communities to be
gathered in.
I hope this email clarifies some points for you.
And a reminder: if you have questions about research activities of the
Wikimedia Foundation, you are welcome to book a public office hour [1]
with myself or a member of the Research team [2] depending on your
topic of interest.
Best,
Leila
[1] https://www.mediawiki.org/wiki/Wikimedia_Research/Office_hours#Schedule
[2] https://research.wikimedia.org/
--
Leila Zia
Head of Research
Wikimedia Foundation
On Sat, Jun 14, 2025 at 10:09 AM Tilman Bayer <haebwiki(a)gmail.com> wrote:
>
> Hi Maryana,
>
> On Thu, Feb 29, 2024 at 9:04 AM Maryana Iskander <miskander(a)wikimedia.org> wrote:
>>
>> This message will be translated into other languages on Meta-wiki
>>
>> العربية • español • français • português • Deutsch• 中文
>>
>> You can help with more languages
>>
>>
>> Hi everyone,
>>
>> Since joining the Foundation I have tried to regularly write to you here and elsewhere, and I wanted to share a few updates since my last letter. In October 2023, I reflected that we were in a period of compounded challenges across the world with escalating wars, conflict, and climate reminding us each week that global volatility and uncertainty was on the rise. That feels even more true now. My instinct then was to ask us to make more time to talk to each other and to try and pull closer together. This feels even more needed now.
>>
>> [...]
>>
>> Finally, our human-led values came up in several conversations about Wikimedia’s role in shaping the next generation of artificial intelligence, a topic of ongoing discussion in the world, in our communities, and at the Foundation. This is complemented by ongoing discussions about the role of AI-generated content on our platform by various project communities. A recent effort to contribute to a shared research agenda on AI can be found here – including the need for more research to understand human motivation to contribute to the knowledge commons – it was created by a small group working in the open who rushed to publish a ‘bad first draft’ that will benefit from more input.
>>
>>
> Thank you for acknowledging the limitations of this document. As we also noted in the Signpost at the time, there was indeed some consternation about the lack of involvement of the volunteer community:
> "While the announcement appears to be speaking on behalf of 'volunteer contributors', the 'Wikimedians' involved in drafting the document appears to have consisted exclusively of Wikimedia Foundation staff (largely from its Research department), according to the attendee list."
>
> I have to ask though, are there still plans to solicit wider input on this draft agenda, or at least incorporate more from the numerous related discussions on AI that have been happening across the movement over the last several years? (At https://meta.wikimedia.org/wiki/Artificial_intelligence , some editors including myself have been trying to keep a list of relevant links, but it's surely not complete.) Again, I appreciate that your post here invited "more input" on the agenda's talk page. But it seems that only a single topic was added there afterwards, and in any case no content updates have been made to that "bad first draft" since February 2024.
>
> Relatedly, given that the document states that "Our hope is that many researchers across industry, government, and nonprofit organizations will adopt the final research agenda to help support and guide their own research":
> Are there plans to solicit input from such external researchers on the draft? And once this research agenda is finalized, does the Foundation plan to bring it to their attention? It doesn't seem to have made such efforts yet, e.g. I can't find any mention of it on the Wiki-research-l mailing list (CCing it now).
>
> I thought that maybe this Bellagio document had been a tangential one-off to make use of an external funding opportunity, and had been abandoned afterwards. But then I saw that more recently Selena highlighted it in the "Reflections on 2025 from the Wikimedia Foundation Executive Team" (as the only concrete outcome regarding AI mentioned in this entire overview of WMF accomplishments "Over the past year").
>
> Regards, Tilman ([[User:HaeB]])
>
>>
>> [...]
>>
>> Maryana
>>
>>
>> Maryana Iskander, Wikimedia Foundation CEO
>>
>>
> _______________________________________________
> Wikimedia-l mailing list -- wikimedia-l(a)lists.wikimedia.org, 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/wikimedia-l@lists.wikimedia.org…
> To unsubscribe send an email to wikimedia-l-leave(a)lists.wikimedia.org
Seconding Galder in bringing this important research finding to wider
attention (CCing the Wiki-research list as well).
If I may, we also just covered this in the new Signpost with some
additional context and detail that some here might find interesting:
"Wikimedia Foundation reports 8% traffic drop since last year due to 'the
impact of generative AI and social media'"
<https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2025-10-20/In_th…>
.[1]
As mentioned there, the problem of quantifying the impact of generative AI
on Wikipedia readership has already attracted considerable attention by
academic researchers in the almost three years since ChatGPT launched. I
even think it's safe to say that many might regard it as one of the most
important research questions about Wikipedia and AI currently.[2]
However, the academic research publications about this - admittedly
difficult - topic have been a bit of a mixed bag so far
<https://meta.wikimedia.org/wiki/Research:Newsletter/2025/March#So_again,_wh…>.
(Even what I would regard as the best paper among these, which Mako and I
also chose to highlight in our annual "State of Wikimedia Research
2024-2025" talk at Wikimania this year, still has some potential statistical
shortcomings
<https://en.wikipedia.org/wiki/Wikipedia_talk:Wikipedia_Signpost/2025-03-22/…>.
And I am not aware of much peer-reviewed insight that goes beyond
the early-years ChatGPT.)
So it's great to see that the Wikimedia Foundation has now itself jumped
into this breach. It is evidently well positioned to do so. Not just
because it has potentially useful internal data that was not available to
the aforementioned external researchers (say referrer information
<https://wikitech.wikimedia.org/wiki/Data_Platform/Data_Lake/Traffic/Pagevie…>,
or now also the data derived from the unique reader cookies WMF introduced
earlier this year, which enable better but still relatively
privacy-friendly tracking of Wikipedia readers). But also e.g. because of
the foundational work <https://research.wikimedia.org/foundational.html> its
own researchers already did some years ago on "Detecting and Gauging Impact
on Wikipedia Page Views" (IMO a paper worth looking at for anyone
investigating such questions).
Unfortunately though, Marshall Miller's Diff post doesn't contain any
concrete information about the methodology used to causally attribute the
drop to those two factors, nor any links to further details. Can we expect
the underlying statistical analysis to be published soon? It would also be
interesting to know e.g. each factor's share in this 8% drop (is it say 7%
from AI vs. 1% from social media, or vice versa?) Or, to circle back a
little towards Galder's points, whether different Wikipedia languages were
affected differently, etc.
Regards, Tilman
[1[ Alongside a somewhat related story
<https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2025-10-20/In_th…>
about
a certain well-known US entrepreneur trying to replace Wikipedia with a new
AI-based encyclopedia, also mentioning some similar but less well-known
efforts.
[2] For some reason though, this central research question didn't make it
into the list
<https://meta.wikimedia.org/wiki/Artificial_intelligence/Bellagio_2024> of
research directions about the implications of AI for the knowledge commons
that was drafted at the February 2024 Bellagio symposium convened by the
Wikimedia Foundation. But as we heard recently on this mailing list
<https://lists.wikimedia.org/hyperkitty/list/wikimedia-l@lists.wikimedia.org…>,
the WMF research department is still working on getting a fuller
documentation of that event's outcomes ready for publication. (At that
point we will hopefully also see an attempt to solicit wider input on this
draft agenda, from other researchers and the editor community, the latter
appearing particularly relevant as the project's stated purpose is to
convey "vital questions volunteer contributors have raised"). And in any
case, it or some of its colleagues have evidently now tackled this research
question themselves.
On Mon, Oct 20, 2025 at 11:46 PM Galder Gonzalez Larrañaga <
galder158(a)hotmail.com> wrote:
> Dear wikimedians,
> I recommend reading the very interesting diff post by Marshall Miller
> about new user trends on Wikipedia:
> https://diff.wikimedia.org/2025/10/17/new-user-trends-on-wikipedia/. The
> post talks about two main trends affecting our views: AI summaries and
> video consumption by new generations. We don't have a good solution about
> the first one, but I would like to talk about the second one, because in
> the Basque Wikimedians User Group we have been working on this for some
> years now, and I think our experience can be part of the solution.
>
> Five years ago we detected this trend and saw that, while there was a rise
> on educative/informative videos for learning, those videos were mainly done
> in English or hegemonic languages. Students and teachers were using videos
> more than ever, but those videos weren't free nor in Basque. That's why we
> created the platform Ikusgela on wiki and external media channels (Wiki
> platform: https://eu.wikipedia.org/wiki/Atari:Hezkuntza/Ikusgela, other
> links to social media and video platforms there)
>
> After four years creating videos we have nearly 250 free educative videos
> in Basque, many of those subtitled in other languages. We cover topics from
> philosophy to evolution, from literature to basic science. And now we are
> publishing an average of 2 new videos per week. We are now working on two
> new series about migrations/rights and linguistics. The videos have
> received the highest award in the Basque Country for communications and are
> used now in education by teachers in many schools and MOOCs. More than 540
> articles in Wikipedia have a video from Ikusgela available, with many more
> to come this school year.
>
> Making good quality videos is expensive, and we are making this in our own
> with our funding and from competitive grants. However, once the videos are
> made, remaking those in other languages should be cheaper. Building the
> videos together is also cheaper than doing it alone. If your chapter or
> user group is interested on that, let me know and we can make things
> together.
>
> We, Basque Wikimedians, saw how this trend was coming and worked to take
> advantage from it, instead of seeing how it decreases our relevancy. I hope
> efforts from the WMF and other affiliates can go in the direction of
> multimedia soon.
>
> Best,
>
> Galder
>
> _______________________________________________
> Wikimedia-l mailing list -- wikimedia-l(a)lists.wikimedia.org, 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/wikimedia-l@lists.wikimedia.org…
> To unsubscribe send an email to wikimedia-l-leave(a)lists.wikimedia.org
>