Arjan van Dalfsen
banner
arjanvd.bsky.social
Arjan van Dalfsen
@arjanvd.bsky.social
PhD candidate Dutch disposition towards animals and plants (1550-2000) | Computational Humanities, Cultural History, AI, Historical Literature
Kimi K2-thinking has the ability to make slide decks. Had not seen that with any other LLMs. Aesthetically, they're better than mine (which, admittedly, doesn't say a lot)
November 8, 2025 at 2:19 PM
October 31, 2025 at 11:32 AM
October 14, 2025 at 9:38 AM
Sora2 met wat vroegmoderne casuïstiek
October 3, 2025 at 10:31 AM
Wat mij frusteert aan dat kieskompas is dat de tweedimensionale visualisatie gewoon nergens op slaat
September 30, 2025 at 8:22 AM
Ik lees bij Caesar Gezelle (inderdaad, oomzegger van) over een paard dat het in zijn poten jeukt van "bratte weelde".

Even nazoeken leert: WNT had in 1898 al lemma voor "brat". Betekenis komt aardig overeen met die van van GenZ. Soms heb je helemaal geen nieuw woordenboek nodig.
September 23, 2025 at 8:49 AM
Reposted by Arjan van Dalfsen
L’œuvre des Frères de Limbourg est avant tout un témoignage réaliste de leur époque.

La 1ère photographie.

Les Très Riches Heures du duc de Berry, Septembre, 1402-1416.

Frères de Limbourg.

Le Château de Saumur.

#Saumur #Castle #Anjou #Illumination #Enluminure
#Paint #DeLimbourg #MasterPiece
September 18, 2025 at 9:22 PM
Reposted by Arjan van Dalfsen
Also: we are switching pretraining paradigms right now from fuzzy web crawl corpus with honestly questionable data practices all around, to some form of generalized "distant writing". Don't think anyone is aware of this/talking about this in DH circles and this is certainly a miss.
September 4, 2025 at 11:34 PM
Reposted by Arjan van Dalfsen
Meanwhile, I see deep questions on the AI side that should definitely warrant some humanities input. And I don't mean the usual ethics corner, but: how to design reasoning structure/flow, personality tuning (what is the "I" in the model?), rethinking tokenization. Won't happen for some time.
Keeping “digital humanities” together as a field is going to get so hard in the era of AI. The shear forces are intense: I see lots of humanities profs who were DH-friendly reposting Zitron.

And while E.Z. is a Limbaugh-tier thinker imo, they’re not wrong that AI is disrupting things they value. +
September 4, 2025 at 11:28 PM
Reposted by Arjan van Dalfsen
Seeing a lot of critical responses from medievalists about this BBC 1066 thing. As an early modernist, just to say that I'm happy to confirm that medieval people were indeed covered in dirt all the time and did all sound like they were in the Wurzels.
August 27, 2025 at 6:24 AM
🥲
August 26, 2025 at 8:30 AM
Reposted by Arjan van Dalfsen
Ok small rant: I don't think we're in a generative AI bubble.

Actual market is still small. It seems bigger only because all kinds of things (and liabilities) are labelled "AI" now. Long overdue infrastructure update? For AI. Having to cut jobs? Because of AI. Fundraising? Quick add AI to name.
August 23, 2025 at 9:43 PM
Reposted by Arjan van Dalfsen
OPROEPJE
Ik ben op zoek naar een paar leerlingen in hun 2e of 3e jaar vmbo-BB/KB, die mij zouden willen helpen met het proeflezen van een vragenlijst. Dit duurt in totaal ongeveer 30 minuten en er staat een presentje tegenover! Ken jij of ben jij iemand met interesse?
August 21, 2025 at 12:51 PM
First impression, of 5️⃣
In terms of raw power it's a change in degree not in kind. But for 90% of users it will effectively be huge as they will now get the right model for their prompt.

So my 2 cents: generally a big step, but a bit less for the avant-garde
August 7, 2025 at 10:49 PM
Uit een presentatiereeks voor na de zomer: het leven van St. GPT

#1 St. GPT geeft teksten aan de mensen
July 31, 2025 at 12:45 PM
#59 🤓
Clioguesser: guess the year based on a political map, ranging from 1000 BCE to 2024 CE.

I am 15th on the leaderboard ;)

clioguesser.azurewebsites.net/leaderboard
July 24, 2025 at 9:45 AM
Reposted by Arjan van Dalfsen
Clioguesser: guess the year based on a political map, ranging from 1000 BCE to 2024 CE.

I am 15th on the leaderboard ;)

clioguesser.azurewebsites.net/leaderboard
June 30, 2025 at 8:42 AM
Grok4 is het eerste model dat dit antwoord goed heeft.

Ik ben door mijn testjes heen, het is tijd voor nieuwe.
July 10, 2025 at 9:28 AM
Lmao
July 5, 2025 at 12:14 PM
🤔
June 24, 2025 at 11:39 AM
1. zijn mensen eigenlijk gedesigned om factual questions accurately te answeren? kunnen wij het?
2. hoe kan het dat de ene neurale structuur beter is dan de andere in dat woord voorspellen?
3. als het allemaal gebaseerd is op dingen die al bestaan, waarom boost synthetische data dan performance?
Chatbots — LLMs — do not know facts and are not designed to be able to accurately answer factual questions. They are designed to find and mimic patterns of words, probabilistically. When they’re “right” it’s because correct things are often written down, so those patterns are frequent. That’s all.
June 20, 2025 at 4:55 PM
June 20, 2025 at 1:48 PM
Ben ik de enige die dit een enge serie vond?
nos.nl/artikel/2571...
Paulus de boskabouter-poppen brengen op veiling ruim een ton op
Daarnaast werden originele tekeningen van Paulus de boskabouter geveild. Daar werd 17.000 euro voor neergeteld.
nos.nl
June 18, 2025 at 10:47 AM
putting this in my slides for DHBenelux
June 3, 2025 at 2:05 PM
🫡
June 2, 2025 at 11:39 AM