Tilman Bayer
@tilmanbayer.bsky.social
AI, data, Wikipedia, co-maintainer of @wikiresearch.bsky.social
5.1 Thinking argues that the regulators make the "online platform" classification on the level of "services" (recital 15), after which "Online-platform-specific rights [...] aren’t limited to public content" on that service chatgpt.com/share/691ba8...
November 17, 2025 at 11:19 PM
5.1 Thinking argues that the regulators make the "online platform" classification on the level of "services" (recital 15), after which "Online-platform-specific rights [...] aren’t limited to public content" on that service chatgpt.com/share/691ba8...
What if this is about the denominator rather than the numerator - say because NYT, Reuters, WaPo are being shared disproportionally more often by posters who care more about quality than about engagement?
November 16, 2025 at 9:25 PM
What if this is about the denominator rather than the numerator - say because NYT, Reuters, WaPo are being shared disproportionally more often by posters who care more about quality than about engagement?
November 16, 2025 at 11:58 AM
What if this is about the denominator rather than the numerator - say because NYT, Reuters, WaPo are being shared disproportionally more often by posters who care more about quality than about engagement?
David Rosenthal (a long-time HAMR skeptic) still seems not convinced, at least as of earlier this year: blog.dshr.org/2025/06/the-...
The State Of Storage
The Register is running a series on The State Of Storage . Below the fold I flag some articles worth reading.
blog.dshr.org
November 16, 2025 at 11:28 AM
David Rosenthal (a long-time HAMR skeptic) still seems not convinced, at least as of earlier this year: blog.dshr.org/2025/06/the-...
Not according to en.wikipedia.org/wiki/SS_Edmu... , which cites many changes that were implemented as result of investigations by USCG and NTSB - most of which did not involve NOAA, e.g. mandating use of depth finders, installing a positioning system and radio beacons, etc.
SS Edmund Fitzgerald - Wikipedia
en.wikipedia.org
November 11, 2025 at 8:54 PM
Not according to en.wikipedia.org/wiki/SS_Edmu... , which cites many changes that were implemented as result of investigations by USCG and NTSB - most of which did not involve NOAA, e.g. mandating use of depth finders, installing a positioning system and radio beacons, etc.
I'm sorry, but this analysis of "the sourcing used by Grokipedia" is essentially bogus - it's based on major misconceptions of how such AI systems work. I tried to explain at en.wikipedia.org/wiki/Wikiped...
Wikipedia talk:Wikipedia Signpost/2025-11-10/Opinion - Wikipedia
en.wikipedia.org
November 11, 2025 at 8:52 AM
I'm sorry, but this analysis of "the sourcing used by Grokipedia" is essentially bogus - it's based on major misconceptions of how such AI systems work. I tried to explain at en.wikipedia.org/wiki/Wikiped...
Then again, clinicians' use of speech recognition goes back to the 20th century (pmc.ncbi.nlm.nih.gov/articles/PMC... ), so it's quite possible that this TikToker's "My provider used AI to take notes" j'accuse could already have applied to the midwife who assisted when she was born 😉
Speech recognition for clinical documentation from 1990 to 2018: a systematic review
The study sought to review recent literature regarding use of speech recognition (SR) technology for clinical documentation and to understand the impact of SR on document accuracy, provider efficiency, institutional cost, and more. We searched 10 ...
pmc.ncbi.nlm.nih.gov
November 10, 2025 at 4:06 AM
Then again, clinicians' use of speech recognition goes back to the 20th century (pmc.ncbi.nlm.nih.gov/articles/PMC... ), so it's quite possible that this TikToker's "My provider used AI to take notes" j'accuse could already have applied to the midwife who assisted when she was born 😉
(And it also doesn't mention that psychotherapy notes receive special protection under HIPAA: www.hhs.gov/hipaa/for-pr... - so are we sure that HIPAA lets "Data Brokers Make Money Off" these too?)
November 10, 2025 at 4:06 AM
(And it also doesn't mention that psychotherapy notes receive special protection under HIPAA: www.hhs.gov/hipaa/for-pr... - so are we sure that HIPAA lets "Data Brokers Make Money Off" these too?)
Oh, I agree re anonymization and data brokers in general.
However, I thought you were warning about AI-specific privacy problems. The article you linked is from 2016 and does not mention AI at all.
However, I thought you were warning about AI-specific privacy problems. The article you linked is from 2016 and does not mention AI at all.
November 10, 2025 at 4:06 AM
Oh, I agree re anonymization and data brokers in general.
However, I thought you were warning about AI-specific privacy problems. The article you linked is from 2016 and does not mention AI at all.
However, I thought you were warning about AI-specific privacy problems. The article you linked is from 2016 and does not mention AI at all.
Re privacy: So you're asserting that the already widely used scribes by e.g. Abridge, Nuance DAX (Microsoft), Suki, Nabla, Ambience/Epic all violate HIPAA? dmclawllc.com/2025/08/12/u...
Or are you proposing new, stricter guardrails beyond HIPAA?
Or are you proposing new, stricter guardrails beyond HIPAA?
Using AI Scribes and Legal Compliance: What Providers Need to Know - DMC Law, LLC
As the use of artificial intelligence-powered scribe tools becomes more common in clinical settings, providers must carefully consider patient consent, including the possible applicability of state au...
dmclawllc.com
November 10, 2025 at 2:01 AM
Re privacy: So you're asserting that the already widely used scribes by e.g. Abridge, Nuance DAX (Microsoft), Suki, Nabla, Ambience/Epic all violate HIPAA? dmclawllc.com/2025/08/12/u...
Or are you proposing new, stricter guardrails beyond HIPAA?
Or are you proposing new, stricter guardrails beyond HIPAA?
... which means that nobody can be held accountable for this quote 😁
November 3, 2025 at 7:21 AM
... which means that nobody can be held accountable for this quote 😁
Musk had talked about "rewrit[ing] Wikipedia" last month already (as we also mentioned twice in the Signpost en.wikipedia.org/wiki/Wikiped... ). So, not sure where this "called it" comes from.
Besides, it's also the approach taken previously by Conservapedia, Citizendium (initially), Ruwiki etc.
Besides, it's also the approach taken previously by Conservapedia, Citizendium (initially), Ruwiki etc.
Wikipedia:Wikipedia Signpost/2025-10-20/In the media - Wikipedia
en.wikipedia.org
October 28, 2025 at 9:14 PM
Musk had talked about "rewrit[ing] Wikipedia" last month already (as we also mentioned twice in the Signpost en.wikipedia.org/wiki/Wikiped... ). So, not sure where this "called it" comes from.
Besides, it's also the approach taken previously by Conservapedia, Citizendium (initially), Ruwiki etc.
Besides, it's also the approach taken previously by Conservapedia, Citizendium (initially), Ruwiki etc.
Funny that they removed "harassment campaign" from the first sentence, but left it in the URL grokipedia.com/page/Gamerga...
Gamergate
Gamergate was a grassroots online campaign launched in August 2014 to demand accountability and ethical reforms in video game journalism, ignited by a detailed blog post from programmer Eron Gjoni acc...
grokipedia.com
October 28, 2025 at 6:02 AM
Funny that they removed "harassment campaign" from the first sentence, but left it in the URL grokipedia.com/page/Gamerga...
This seems to confound using Wikipedia for training LLMs with the direct, attributed republication or adaption of Wikipedia content, which has been going on for almost a quarter century: en.wikipedia.org/wiki/Wikiped... ("Thousands of 'mirror sites' exist that republish content from Wikipedia")
Wikipedia - Wikipedia
en.wikipedia.org
October 28, 2025 at 5:26 AM
This seems to confound using Wikipedia for training LLMs with the direct, attributed republication or adaption of Wikipedia content, which has been going on for almost a quarter century: en.wikipedia.org/wiki/Wikiped... ("Thousands of 'mirror sites' exist that republish content from Wikipedia")
All good (I wasn't there, although I did spend a bit of time helping to cover it in our little community newsletter en.wikipedia.org/wiki/Wikiped... ). Hope you're well too!
Wikipedia:Wikipedia Signpost/2025-10-20/In the media - Wikipedia
en.wikipedia.org
October 24, 2025 at 7:55 AM
All good (I wasn't there, although I did spend a bit of time helping to cover it in our little community newsletter en.wikipedia.org/wiki/Wikiped... ). Hope you're well too!
In a quick scan of the paper, I don't see anything about the funding or organizing of conferences, workshops and the like. Would you consider these possible avenues of influence in this area too?
October 24, 2025 at 1:27 AM
In a quick scan of the paper, I don't see anything about the funding or organizing of conferences, workshops and the like. Would you consider these possible avenues of influence in this area too?
The author is on here too 😉 bsky.app/profile/mano...
In a new blog post, I argue we must be more critical in evaluating research on “algorithmic amplification.” Otherwise, we risk rediscovering that algorithms do not exist in a vacuum. We're better off doing other things!
doomscrollingbabel.manoel.xyz/p/does-tikto...
doomscrollingbabel.manoel.xyz/p/does-tikto...
October 24, 2025 at 1:08 AM
The author is on here too 😉 bsky.app/profile/mano...
I'm aware of the timing. It's more about your sources seemingly including none of the several Wikimedians at the event who apparently had at least some knowledge of the issue.
If you or another journalist had broken that story, that might also have relieved SFR of the burden of breaking the ANPDP.
If you or another journalist had broken that story, that might also have relieved SFR of the burden of breaking the ANPDP.
October 23, 2025 at 11:32 PM
I'm aware of the timing. It's more about your sources seemingly including none of the several Wikimedians at the event who apparently had at least some knowledge of the issue.
If you or another journalist had broken that story, that might also have relieved SFR of the burden of breaking the ANPDP.
If you or another journalist had broken that story, that might also have relieved SFR of the burden of breaking the ANPDP.
Like many Wikimedians, I generally appreciate how your reporting is better informed than others thanks to your connections with (parts of) the Wikipedia community. Can't help noticing though that here you missed a major angle of the story that was revealed yesterday: en.wikipedia.org/wiki/Wikiped...
Wikipedia:Arbitration Committee/Noticeboard - Wikipedia
en.wikipedia.org
October 23, 2025 at 7:11 PM
Like many Wikimedians, I generally appreciate how your reporting is better informed than others thanks to your connections with (parts of) the Wikipedia community. Can't help noticing though that here you missed a major angle of the story that was revealed yesterday: en.wikipedia.org/wiki/Wikiped...
As a longtime Wikimedian, I still remember you being seen as an ally of the free knowledge movement; you even used to be on the WMF advisory board.
Very sad to see you have joined the anti-fair-use bandwagon and engage in full-throated advocacy for the business interests of the copyright industry.
Very sad to see you have joined the anti-fair-use bandwagon and engage in full-throated advocacy for the business interests of the copyright industry.
October 23, 2025 at 7:01 PM
As a longtime Wikimedian, I still remember you being seen as an ally of the free knowledge movement; you even used to be on the WMF advisory board.
Very sad to see you have joined the anti-fair-use bandwagon and engage in full-throated advocacy for the business interests of the copyright industry.
Very sad to see you have joined the anti-fair-use bandwagon and engage in full-throated advocacy for the business interests of the copyright industry.
Helping researchers secure their future @pessimistsarc.bsky.social spot
October 20, 2025 at 2:19 PM
Helping researchers secure their future @pessimistsarc.bsky.social spot
They fixed this one too in the current version (but the Wayback Machine confirms your screenshot: web.archive.org/web/20251016... )
web.archive.org
October 19, 2025 at 1:58 AM
They fixed this one too in the current version (but the Wayback Machine confirms your screenshot: web.archive.org/web/20251016... )
In case anyone wants to check further, the earliest snapshot on the Internet Archive is web.archive.org/web/20251016... (unlike arXiv, agidefinition.ai doesn't preserve a revision history...)
web.archive.org
October 19, 2025 at 1:54 AM
In case anyone wants to check further, the earliest snapshot on the Internet Archive is web.archive.org/web/20251016... (unlike arXiv, agidefinition.ai doesn't preserve a revision history...)
Don't forget banner donation revenues, which might even be more affected than volunteer edits per the (quite preliminary) findings from one of the aforementioned papers: meta.wikimedia.org/wiki/Researc...
Could create some interesting and so far unexamined WMF-community dynamics and possible COIs
Could create some interesting and so far unexamined WMF-community dynamics and possible COIs
Research:Newsletter/2025/March - Meta-Wiki
meta.wikimedia.org
October 19, 2025 at 12:19 AM
Don't forget banner donation revenues, which might even be more affected than volunteer edits per the (quite preliminary) findings from one of the aforementioned papers: meta.wikimedia.org/wiki/Researc...
Could create some interesting and so far unexamined WMF-community dynamics and possible COIs
Could create some interesting and so far unexamined WMF-community dynamics and possible COIs
Not very surprising. But I'm quite curious how WMF determined causality: "We believe that these declines reflect the impact of generative AI and social media".
This can be quite difficult; I covered several papers that tried to do that (for earlier timespans) at
meta.wikimedia.org/wiki/Researc...?
This can be quite difficult; I covered several papers that tried to do that (for earlier timespans) at
meta.wikimedia.org/wiki/Researc...?
Research:Newsletter/2025/March - Meta-Wiki
meta.wikimedia.org
October 19, 2025 at 12:11 AM
Not very surprising. But I'm quite curious how WMF determined causality: "We believe that these declines reflect the impact of generative AI and social media".
This can be quite difficult; I covered several papers that tried to do that (for earlier timespans) at
meta.wikimedia.org/wiki/Researc...?
This can be quite difficult; I covered several papers that tried to do that (for earlier timespans) at
meta.wikimedia.org/wiki/Researc...?