Mark Dingemanse
LightNews LightNews
banner
dingemansemark.bsky.social
Mark Dingemanse
@dingemansemark.bsky.social
2.8K followers 440 following 4.2K posts
Language, interaction, tech • Here with doubts about the next monetizable monopoly... • papers https://markdingemanse.net • blog https://ideophone.org • fedi https://scholar.social/@dingemansemark/ • POSSE: Publish on Own Site, Syndicate Everywhere
Posts Replies Media Videos
Pinned
dingemansemark.bsky.social
Mark Dingemanse @dingemansemark.bsky.social · Oct 18
Welcome new followers! You may find me posting about linguistics, language technology, "AI", open research, epistemic injustice, home computing, and whatever captures my fancy. I blog at ideophone.org (since 2007) and my academic site —with PDFs of papers etc.— is markdingemanse.net
dingemansemark.bsky.social
Evergreen tweet as the AI slop paper in @springernature.com HSSComms keeps doing the rounds

The reputation of @nature.com is suffering death by a thousand cuts thanks to this nature dot com URL branding (justified anyway IMO)
dingemansemark.bsky.social Mark Dingemanse @dingemansemark.bsky.social · Aug 24
I remain amazed at the effectiveness of the "nature․com/journal/" URL branding which invariably leads many to assume that some work is published "in Nature" when in fact it's in one of the other >160 journals owned by Springer Nature...
January 13, 2026 at 10:46 AM
dingemansemark.bsky.social
TFW they ask you to include a table to account for every single of the precise 168 hours students will be expected to spend on this 6EC course
part of a table showing "For MA variant: reading this syllabus, collecting readings, thinking about study load: 1 hour 12 x 2h lectures: 24 hours"
January 12, 2026 at 2:06 PM
dingemansemark.bsky.social
"Silence [...] is necessary for the emergence of persons. It is taken from us by machines that ape people. We could easily be made increasingly dependent on machines for speaking and for thinking, as we are already dependent on machines for moving."

Ivan Illich, Silence is a commons, 1984
Silence is a commons – The Ideophone
ideophone.org
January 12, 2026 at 12:41 PM
dingemansemark.bsky.social
This is how to uphold standards of research integrity 👇👏
ethankutlu.bsky.social Dr. Ethan Kutlu @ethankutlu.bsky.social · 6d
Thank you for sharing this! Recently, I decided to not submit our final manuscript which was accepted. We didn’t receive any feedback on our science and the only review we received was AI generated wording changes. It is the same journal. I emailed the editorial team too and I received no answer.
January 8, 2026 at 7:22 PM
dingemansemark.bsky.social
Re: the AI slop paper shared by @thomaspellard.bsky.social and @lameensouag.bsky.social, I wrote to the editors — will update when I get a reply, and will be following closely what they do.

Key point is that we should hold the *journal* accountable for this mess

I have a few predictions...

1/n
AI slop published on your watch: very bad look for HSSComms and Springer/Nature  Dear editors, I have failed to locate contact information for most of the academic editors for Humanities and Social Sciences Communications; I trust that you will forward this message to them. I want to note that AI slop is being published on your watch in the journal you edit. This paper, out last week (Al-Jarrah 2026), is full of inaccurate claims and includes countless hallucinated references. Even a cursory look at the bibliography shows that at least 10 and probably many more references are bullshit pure and simple; the simplest explanation is that they are confabulated by generative AI, which sheds doubt on the quality of the manuscript as a whole and on the review and editorial processes at your journal. It is a great lapse of editorial judgement to let this kind of obvious drivel pass and I am warning my colleagues at MPI and Radboud to avoid your journal until clear and unambiguous action is taken. You may also want to take note that many people are finding out about this and the online discussion of this paper and of the journal's failing standards is something I hope that Springer Nature cares about. In my opinion the paper does not pass even the most minimal quality assurance checks and is fully against the COPE guidelines on publication ethics. You may want to hold the author accountable for this; I think the only reasonable course is retraction. But the scholarly community also holds the journal accountable. By letting AI slop through, your journal is polluting the information ecology of scientific publishing. Amidst a rising tide of synthetic text, scholarly publishing, with its tradition of human oversight and strong peer revidew, should be one of the last stalwarts to defend the integrity of our research. I look forward to your response and to find out what decisive action you are taking.
January 8, 2026 at 5:14 PM
dingemansemark.bsky.social
New on the blog: Requiem for writing town ideophone.org/requiem-for-...

On the co-working space we had for our team in corona times, what we learned about interactional affordances, why surface realism is overrated, and how we created a sense of connectedness (also ft. @irisvanrooij.bsky.social)
Requiem for writing town – The Ideophone
ideophone.org
January 7, 2026 at 9:47 PM
dingemansemark.bsky.social
Re: the Grok 'apology', already in 2023 we wrote that LLMs are devoid of social accountability so that an apology-shaped string of output tokens is meaningless doi.org/10.1016/j.ti...

we should hold makers accountable for what their text and image generators spit out
Whether following predesigned dialog flows or generated by next token prediction, the output of conversational interfaces is fundamentally devoid of accountability: it is text produced without a commitment to truth or joint action [72]. By tailoring textual output to be maximally in line with what human raters prefer, techniques like reinforcement learning with human feedback obscure this lack of accountability but do not fundamentally change it: a text generator may ‘apologise’ when prompted for clarification or correction but will indifferently regurgitate the same misunderstanding or a new one on the next occasion
January 7, 2026 at 6:57 AM
Reposted by Mark Dingemanse
mathijsvdsande.bsky.social
Komende zondag: het jaarlijkse nieuwjaarsfeest van het Nijmeegse straatorkest Kladderadatsch! Met een gastoptreden van de Bossche dub-band 'Dub-A-Riddim'.

Kom een dansje wagen! Zondag 4 januari, 14:00-18:00, Plan A (vlietstraat 22, Nijmegen).
Uitnodiging: Nieuwjaarsfeest van Kladderadatsch, met gastoptreden van Dub-A-Riddim. Het plaatje is in aubergine en toont de hals/mondstuk van een baritonsaxofoon. Details feest: zondag 4 januari, 14:00-18:00, Plan A (vlietstraat 22, Nijmegen).
January 2, 2026 at 8:00 AM
Reposted by Mark Dingemanse
dingemansemark.bsky.social
Two recent blogs on The Ideophone, prompted by a fun #ReproducibiliTea session in Cologne:

1. On generative AI and reproducibility ideophone.org/on-generativ...

On why the praxis of slow and reproducible science provides useful lessons for navigating the lures of “AI”

1/n
Photograph of Isabelle Stengers and cover of her book, Another Science is Possible: A manifesto for Slow Science
December 18, 2025 at 9:23 PM
dingemansemark.bsky.social
Two recent blogs on The Ideophone, prompted by a fun #ReproducibiliTea session in Cologne:

1. On generative AI and reproducibility ideophone.org/on-generativ...

On why the praxis of slow and reproducible science provides useful lessons for navigating the lures of “AI”

1/n
Photograph of Isabelle Stengers and cover of her book, Another Science is Possible: A manifesto for Slow Science
December 18, 2025 at 9:23 PM
dingemansemark.bsky.social
Cristina Zaga bracht wetenschappers, experts en belanghebbenden tezamen om een monter en helder alternatief te formuleren voor winstgedreven "AI"-plannen: Zorgvuldig & Zorgzaam Digitaal. Verkies je ook gezond verstand boven geautomatiseerde gedachteloosheid? 📄___🖋️ openletter.earth/nl/zorgvuldi...
Zorgvuldig & Zorgzaam Digitaal
openletter.earth
December 15, 2025 at 12:10 PM
Reposted by Mark Dingemanse
irisvanrooij.bsky.social
📝 OPEN BRIEF 📝

Wil jij dat onze regering zorgvuldig en zorgzaam vorm geeft aan onze digitale toekomst, met oog voor mens en natuur?

Teken dan 🖊️ deze open brief, gericht aan de formerende partijen in de Eerste en Tweede Kamer en aan de demissionaire regering.

📝 openletter.earth/nl/zorgvuldi...
Zorgvuldig & Zorgzaam Digitaal Aan de formerende partijen in de Eerste en Tweede Kamer en aan de demissionaire regering, Deze brief schrijven we u vanwege uw belangrijke verantwoordelijkheid over het huidige en toekomstige AI-beleid, digitalisering en publieke waarden. Wij zijn een coalitie van wetenschappers, experts en vertegenwoordigers van maatschappelijke organisaties. Met deze brief willen we twee dingen doen: a) Context bieden bij plannen die nu rondgaan zoals het AI Deltaplan, de AIC4NL position paper en de Invest-NL AI Deep Dive. In de investeringsvoorstellen staan aannames en voorstellen die niet gebaseerd zijn op wetenschappelijke inzichten en publieke waarden. b) Een andere, constructieve en onderbouwde aanpak bieden die uitgaat van mens, natuur en democratie. We zien waarde in een gedeelde aanpak waarin wetenschappers, maatschappelijke organisaties en belanghebbenden meebepalen of, hoe en welke investeringen noodzakelijk zijn.
December 13, 2025 at 7:39 PM
Reposted by Mark Dingemanse
olivia.science
If you're based in NL please consider reading & signing our open letter on reclaiming so-called AI from the AI Delta Plan & co:

"Thoughtfully Shaping Our Digital Future" openletter.earth/zorgvuldig-a...

Big thanks to @lekkerresistance.bsky.social @irisvanrooij.bsky.social @marentierra.bsky.social
Thoughtfully Shaping Our Digital Future To the parties forming the government in the Senate and House of Representatives, as well as the outgoing administration, We are writing to you in recognition of your crucial responsibility for shaping current and future AI policy, overseeing digitization, and upholding public values. We are a coalition of scientists, experts, and representatives of civil society organizations. We believe it is essential to address these matters together. This letter has two objectives: a) Provide context for current plans, including the AI Delta Plan, the AIC4NL position paper, and the Invest-NL AI Deep Dive. These investment proposals often rely on assumptions that lack scientific evidence and do not fully reflect public values. b) To offer a constructive, well-substantiated alternative approach to digital futures based on people, nature, and democracy. We believe a collaborative process should guide decisions on the needs, scope, and nature of investments by bringing together scientists, civil society organizations, and stakeholders.
December 13, 2025 at 11:12 AM
Reposted by Mark Dingemanse
mathijsvdsande.bsky.social
Het gaat helemaal los dit jaar op 4 januari - straatorkest Kladderadatsch luidt traditiegetrouw het nieuwe jaar in met dit jaar als gastact onvervalste Bossche dub door Dub-a-Riddim.

Kladderadatsch Nieuwjaarsfeest, 4 januari, 14:00-18:00, Plan A (vlietstraat 22), Nijmegen. Entree gratis!
Aankondiging voor feestje. Plaatje is is paars/aubergine met de hals en mondstuk van een baritonsaxofoon. Verder bandlogo's van Kladderadatsch en Dub-A-Riddim en de volgende info: Nieuwjaarsfeest, 4 januari, 14:00-18:00, Plan A (vlietstraat 22), Nijmegen. Entree gratis.
December 8, 2025 at 8:34 PM
dingemansemark.bsky.social
Excellent initiative from our French colleagues
December 6, 2025 at 9:19 AM
dingemansemark.bsky.social
see English spelling is very simple: in recur, 'c' occurs once, and in occur, 'c' recurs

HTH!
December 5, 2025 at 11:05 AM
dingemansemark.bsky.social
As I wrote in 2020... "The web makes up a large chunk of the data feeding GPT3 and kin. Posting the output of large language models online builds a feedback loop that cannot improve quality.

All ingredients for an information heat death are on hand."
hetanshah.bsky.social Hetan Shah @hetanshah.bsky.social · Nov 27
Your universe is a photocopy of a photocopy of a…
on.ft.com/4ahkNKa
Al-generated content has surpassed human-written articles Graph
December 3, 2025 at 5:02 AM
dingemansemark.bsky.social
Ik werd geinterviewd in een stuk over het Rijmwoordenboek en leverde een lofzang op het handgemaakte Sinterklaasgedicht www.bd.nl/heusden/daar...
“Het is een glorieuze traditie van wanstaltige rijmelarij: een van de weinige keren dat je meer dan vijf minuten achtereen aandacht schenkt aan iemand die je niet zelf uitgekozen hebt. Wat doen ze? Hoe was hun jaar? Welke grap kun je maken? Welk succes kun je met ze vieren? Welk prikje kun je uitdelen? Hoe kun je laten zien dat je ze gezien hebt?” De professor heeft maar één advies voor worstelende dichters: „Neem die persoon nou gewoon in gedachten. Pak het rijmwoordenboek erbij. Laat dat versritme scheef lopen of schaaf eraan tot je tevreden bent. Voer Sint ten tonele of desnoods Rijmpiet. Maar in godsnaam, kom niet aan met de gladde woordenbrij die uit een tekstgenerator komt lopen.” Een sinterklaasgedicht van ChatGPT is de ultieme troosteloosheid Mark Dingemanse , Taalkundige, Radboud Universiteit
December 2, 2025 at 7:26 PM
Reposted by Mark Dingemanse
michaelerard.bsky.social
The European Review of Books is at a critical juncture. Its reputation booming, it still needs to cover costs. So if you can subscribe or give a subscription as a gift, please do! europeanreviewofbooks.com/ten-issues-in/
Ten Issues in — and Europe keeps thinking - The European Review of Books
With Issue 10, our golden Jubilee, we celebrate ten editions of independent, multilingual writing that have crossed languages, borders, and expectations. In a time when culture is flattened into opini...
europeanreviewofbooks.com
December 2, 2025 at 9:30 AM
Reposted by Mark Dingemanse
samiraibnelkaid.bsky.social
I feel so lucky to be joining such a cool team and contributing to such an important project 🙌 💫
December 1, 2025 at 9:21 AM
dingemansemark.bsky.social
Morgen: 'Terug naar het ambacht!', Letterenhuis Antwerpen. Ik doe mee aan een panelgesprek over organische taal temidden van een opkomend tij van synthetische tekst auteursvereniging.be/vav-werkcong...
Poster van het evenement Terug naar het ambacht! In een tijd waarin teksten in een oogwenk kunnen worden geproduceerd en de schrijftafel steeds vaker plaatsmaakt voor het scherm, rijst de vraag wat er nog overblijft van het schrijfambacht. Welke plek heeft literaire taal in een wereld die draait op algoritmes en automatisch gegenereerde zinnen? En welke waarde kent het boek nog als traag en tastbaar werk?
November 28, 2025 at 9:07 PM
Reposted by Mark Dingemanse
astrokatie.com
I don’t know if anyone else notices or cares, but when I see a presentation in which the speaker uses obviously generated-AI images to illustrate their slides, it makes me immediately less confident in whatever other content they’re presenting.
November 28, 2025 at 3:07 PM
dingemansemark.bsky.social
Incredibly happy to welcome @samiraibnelkaid.bsky.social and Átila Vital to the Futures of Language team! markdingemanse.net/futures/news...

This concludes the hiring process we conducted over Jul-Sept, with >450 applications. Samira just started, Átila arrives in January — good times!
Welcoming Samira Ibnelkaïd and Átila Vital – Futures of Language
markdingemanse.net
November 27, 2025 at 8:39 PM
dingemansemark.bsky.social
the whack-a-mole problem: every few months we see a new creative bypass of this kind (that will be half-patched by ad hoc metaprompts & later by finetuning)

when will it dawn that LLMs are fundamentally vulnerable to endless new attacks like this, and therefore fundamentally unsafe?
cstross.bsky.social Charlie Stross @cstross.bsky.social · Nov 20
Looks like LLMs are *very* vulnerable to attack via poetic allusion: "curated poetic prompts yielded high attack-success rates (ASR), with some providers exceeding 90% ..."

https://arxiv.org/html/2511.15304v1
November 20, 2025 at 5:49 PM
Reposted by Mark Dingemanse
leguinbot.bsky.social
I was in too much haste, and now have no time left.
November 16, 2025 at 8:23 AM