Moderated the ISFMF 2026 panel on AI and the future of film, gaming, and media arts. Excellent discussion with Mark Ulano, James Mather, Laia Casanovas, and Alex Serdiuk on authorship, synthetic voices, and emotional authenticity in AI-assisted media.

Two postdoc position in CS famnit.upr.si/en/news/2-post… Applications are centralized - want to work with me? Say so in your application. I’m interested in user modelling grounded in cognition, affect & experience — beyond surface behaviour. DL: March 31, 2026

Because so many people mentioned this. Here is the problem(s) with Grokipedia as it stands, and also just copying Wikipedia: 1) There is no way to know if Grok has changed something subtle, or it's 100% verbatim. There is no message to that effect. Saying "Fact checked by Grok 2 days ago" is the encyclopedic equivalent of "trust me bro". Bottom of the page says "The content is adapted from Wikipedia". HOW "adapted" is it? Is it a 100% copy? 99%, 90%? What's changed and how? 2) Grokipedia ends up as a mix of copying Wikipedia and completely rewriting everything. It's like a box of chocolates, you never know what you're going to get. 3) Grokipedia seems to leave out images. 4) Grokipedia provides zero traceability to changes. This is the huge benefit of Wikipedia. 5) How does someone change Grokipedia if they detect an error or want to add something. Yeah yeah, "that's the point" you say, it stops bad actors from getting their hands on it. Well, it also stops good actors from contributing, which, if you aren't talking about anything sociopolitical, that's a bad thing. Does Wikipedia still remain the main source of human change until Grok crawls it again? Or will Grok independently just update it as it find new references itself? 6) This could be an early issue, but should be noted: some 100% copied pages have references (see "Transistor"), whilst others do not have any references (see "Miller Effect").

Computer Science is not science, and it's not about computers. Got reminded about this gem from MIT the other day

Kicking off our #SIGIR2025 tutorial on Psychological aspects in IR and RS w Elisabeth Lex and @m_schedl

Thank you guys. From the bottom of my heart 😘 🙌🏻 @rogerfederer @DjokerNole @andy_murray

People often claim there are "2 types of people". For this to be true for ANY continuously varying trait, you'd need there to be a bimodal distribution - basically, two separated bumps in the data. Some people claim this is incredibly rare. But is it? We investigated: 🧵

Corporate suicide? Synology requires self-branded drives for some consumer NAS systems, drops full functionality and support for third-party HDDs tomshardware.com/pc-componen…

All in favor, say aye. #NewYorkerCartoons 🖋️ Sam Lau, from 2023 Sign up to get New Yorker cartoons right in your inbox: newyorker.com/newsletter/hum…

Without a doubt, the greatest unexpected plot twist in the history of internet wars has just dropped: youtube.com/watch?v=qFe5LiAC…

lol who did this

Link to the paper (10/10) ⤵️ researchgate.net/publication…

Today is the day! Abstracts for Full Papers for #RecSys2025 are due today by the end of the day (AoE). #MarcoDeGemmis & I look forward to your contributions! recsys.acm.org/recsys25/call… @ACMRecSys

"Text-to-Image Generation for Vocabulary Learning Using the Keyword Method" by @Nu1_Rt at ACM #IUI2025 doi.org/10.1145/3708359.3712… Co-authors: @_mkljun_, Aaron Quigley, @_klen_, @armob, Verena Biener, @luileito, Juri Yoneyama, Alice Toniolo, Angie Miguel, Hirokazu Kato & @AuVoidsky

Best Paper Award at ACM #IUI2025🎉 'Text-to-Image Generation for Vocabulary Learning Using the Keyword Method' doi.org/10.1145/3708359.3712… co-auth: @Nu1_Rt @_mkljun_ Aaron Quigley @_klen_ @armob Verena Biener @luileito Juri Yoneyama Alice Toniolo Angie Miguel Hirokazu Kato @AuVoidsky

Congrats to my colleagues from @hicuplab #iui2025

Very proud of this work led by @r_ungruh in collab w/ @tudelft MS student, that just got accepted for @ACM_TORS special issue on #RecSys for Good, focused on understanding #children potential exposure to #stereotypes as a result of their interactions with recommender systems

I attended Trump’s inauguration yesterday. Here are my thoughts:

Habemus #Dagstuhl Report! #Evaluation Perspectives of #RecSys. doi.org/10.4230/DagRep.14.5.… Edited by @alansaid , @eva_zangerle, and yours truly - and written by many, many brilliant minds in our #recsys community. Too many to mention (pic here dagstuhl.de/24211) @dagstuhl

📣Join my team!📣 I'm hiring a doctoral researcher (PhD) to work on **fairness in recommender systems** - with a focus on the music sector or related media fields. Funded 4-year PhD position (30h/week). Vacancy description: tinyurl.com/4w2ts9pj #job #phd #recsys #fairness