With all the talk about "the blackbox problem" and "the right to explanation", I continue to be suprised by how little attention there is to what I have started calling *the faithfulness problem*: the problem of ensuring that generated explanations are faithful to the underlying causal mechanism.

Just making sure that 2025 is definitely over

"When a single company can implement a rights-respecting, consent-based access regime at scale, it’s worth asking why public institutions have failed to do the same."

Big news! 🗞️ I defended my PhD thesis "From Insights to Impact: Actionable Interpretability for Neural Machine Translation" @rug.nl @gronlp.bsky.social I'm grateful to my advisors @arianna-bis.bsky.social @malvinanissim.bsky.social and to everyone who played a role in this journey! 🎉 #PhDone

It's quite amazing how oblivious the people in charge of these services seem to be of the very real concerns people have with being force-fed AI services everywhere.

The debate used to be whether or not neural language models could model the competence of an ordinary language user -- the central topic for linguists. Has the debate now shifted to whether or not language models can model competent linguists?

This situation on a roundabout in Oslo last week, causing major traffic jams, must be a metaphor for something... but I have figured out for what yet. Any suggestions? Photo Endre Helgeland, via www.vg.no/nyheter/i/3p...

Congrats, Scotland, for qualifying for the world cup football! Kudos to this small nation in the United Kingdom of Great Britain and Northern Ireland (population 5.5M)! Congrats, Curaçao, for qualifying too! Kudos to this small nation in the Kingdom of the Netherlands (population 160,000😱😱😱⛈️🤸🏾‍♂️⛹🏾‍♂️🤩❤️...

What do our addictions to sugar, social media, doom scrolling and more have in common? Nicklas Brendborg, in this podcast* and in his book, nicely captures the mechanisms behind them using the concept of 'supernormal stimulus', known from a famous 1947 seagull study *: megaphone.link/NSR7522957702

"The man who co-discovered the double helix, perhaps not surprisingly, regarded DNA as the ultimate puppet master, immeasurably more powerful than the social and other forces that lesser (much lesser) scientists studied. Then his hubris painted him into a corner."

@mikexcohen.bsky.social Nice first post on gender bias in LLMs! mikexcohen.substack.com/p/gender-bia... Now nervously waiting for part 2, to see whether you'll talk your readers through our "adapting Transformer components" work. Or is there something even better? aclanthology.org/2023.blackbo...

Dit Klavertje (dat mijn zoekmachine toont in een zoektocht naar iets heel anders), leest als een halve roman in één zin. Ik ben nu wel benieuwd naar de tweede helft, maar de link is dood (net als de ouders).

Nice post on 'inoculation theory', the idea that we can fight misinformation by training/nudging people ahead of the wave of misinformation. Tom Stafford discusses the current evidence for and against this theory (cf. @profsanderlinden.bsky.social). tomstafford.substack.com/p/helping-pe...

The Majority AI View - Anil Dash Even though AI has been the most-talked-about topic in tech for a few years, we’re in an unusual situation where the most common opinion about AI within the tech industry is barely ever mentioned.

Twenty-four years ago today, our paper “A forkhead-domain gene is mutated in a severe speech and language disorder” was published: www.nature.com/articles/350.... A personal thread about the ups & downs of the journey we took to get to that point....1/n 🗣️🧬🧪

Announcing the first (and perhaps only) Multilingual Minds and Machines Meeting! Come join us in Nijmegen, June 22-23, 2026, if you are interested in computational models of human multilingualism: mmmm2026.github.io

Let's seek shelter under those trees over there! 😱

This is embarrassing reporting, where AI-is-almost-sentient is the "nuanced view", and @mustafasuleymanai.bsky.social the radical. One of @robert-booth.bsky.social's sources: Ufair "a small, undeniably fringe organisation, led by three humans and seven AIs with names such as Aether & Buzz."😱

New blog post: AI Allergy. On my increasing disgust with the AI discourse, even though I still like the technical and philosophical. And how I wish I could be excited about AI again. togelius.blogspot.com/2025/08/ai-a...

Had such a great time presenting our tutorial on Interpretability Techniques for Speech Models at #Interspeech2025! 🔍 For anyone looking for an introduction to the topic, we've now uploaded all materials to the website: interpretingdl.github.io/speech-inter...

The heartbreaking thing about this is that there’s already a proven way to invest lots of money in a knowledge machine that produces unforeseeable results that include fantastically profitable ideas (and some life-saving ones) and generally benefit society and this machine is called a university

Thought provoking work from my Amsterdam/ILLC colleagues

I'll be in Vienna only from tomorrow, but today my star PhD student Marianne is already presenting some of our work: BLIMP-NL, in which we create a large new dataset for syntactic evaluation of Dutch LLMs, and learn a lot about dataset creation, LLM evaluation and grammatical abilities on the way.

Find our paper (w/ Michelle Suijkerbuijk, Zoë Prins, Jelle Zuidema @wzuidema.bsky.social, & Stefan Frank @stefanfrank.bsky.social), and dataset here: 📑 doi.org/10.1162/coli... 🗃️ doi.org/10.34973/tj4...

The @interspeech.bsky.social early registration deadline is coming up in a few days! Want to learn how to analyze the inner workings of speech processing models? 🔍 Check out the programme for our tutorial: interpretingdl.github.io/speech-inter... & sign up through the conference registration form!

Deadlines for PhD and Postdoc vacancies coming up: applications open until Monday June 2!

This Friday at 3pm, Amsterdam's Computational Linguistics Seminar and the CLClab will host Melanie Mitchell, for a special edition of the Amsterdam Lectures in AI and Society. She will speak about "AI's Challenge of Understanding the World". Zoom link/ register: clclab.netlify.app/2025/05/07/a...

TIL that professor Kunihiko Fukushima, inventor of the neocognitron, is still an active researcher in his late 90s. The neocognitron is a convolutional neutral network (CNN); CNNs is turn were the model family with which deep learning revolution in Artificial Intelligence started. A living legend!

✨New paper ✨ Introducing 🌍MultiBLiMP 1.0: A Massively Multilingual Benchmark of Minimal Pairs for Subject-Verb Agreement, covering 101 languages! We present over 125,000 minimal pairs and evaluate 17 LLMs, finding that support is still lacking for many languages. 🧵⬇️

Seems like a crazy way to spend these resources. I'm all for helping American scholars move to Europe, but those that already have an ERC grant waiting for them are pretty well off. The extra 1M€ would be better used to help *other* scholars, and not to make them dependent on those lucky colleagues.

Encouraging to see 5000+ protesters at our local Amsterdam protests against the budget cuts in higher education in the NL! In this time of geopolitical tensions, changing climate, democratic backsliding, & tech revolutions, governments should *invest* in knowledge & the future of the next generation

📢 CALL FOR SATELLITE EVENT PROPOSALS: CCN 2025 @ University of Amsterdam Submit your workshop, tutorial or special session proposal for Aug 11, 2025 to complement the main CCN conference program. Deadline: April 10, 2025 Details in thread 👇 #CCN2025 #CogNeuro

Come present your #NLP work in Leuven!

Let's not polarize unnecessarily. Exploding is harsh; shearing away in a fireball would be fine too.

In the movies, maniacal villains keep media and police busy by setting buildings on fire & sadistically threatening innocent people, while, in fact, robbing the bank Now that we have a maniacal villain setting international relations on fire & making sadistic threats, it's time to connect the dots

Cognitive psych is often traced back to the 1950s. Psychology is often traced to Wundt's lab in 1879. Textbooks speak of "proto-psychologists" like Fechner & Helmholtz in the 1800s... Westerners all. Have you heard about Ibn al-Haytham, whose work predated them by 100s of years? 1/4

Distillation explained, cat edition (via reddit)

Someone at OpenAI should build aan irony detection algorithm For internal use 🙉

Excellent thread (although I think it puts a little too much weight on just one model, deepseek's R1; there were also before last week excellent reasons not to get ahead of ourselves in the race to scale up).

Today is Martin Luther King day in the US. Makes me think back of a shop owner in Queens, who cheerfully sold me this calendar of 1998 (useful again in 2026!), the first time I visited NYC. Incredible to realise that the civil rights movement was as long ago then, as the nineties are for us now.

Can Large Language Models THINK and UNDERSTAND? The answer from cognitive science is, of course, lolwut YES! The more interesting question is CAN TOASTERS LOVE? Intriguingly, the answer is ALSO YES! And they love YOU

This is, or should be, a big deal -- Ann Telnaes, an absolute legend, quitting WaPo due to censorship of a cartoon criticizing billionaires anntelnaes.substack.com/p/why-im-qui...

Saddened to learn about the passing of Felix Hill, a much respected colleague at DeepMind. www.paperlesspost.com/go/7BbrzXXhb... He has himself written about the anxieties that come with working in AI -- hopefully we can take it as a cue to take better care of eachother! medium.com/@felixhill/2...

This is a beautifully produced video explaining the concepts behind 'mechanistic interpretability' if LLMs, that has generated so much excitement in my field in the last two years. Amazing visualizations of the 'residual stream' and 'sparse auto-encoders'. (Commercial break from 7:08 - 9:08) 1/n

Every time a group of linguists manages to publish some cool results in a glam journal*, another group of scholars stands ready to complain about the work and the character of the researchers. Every time. Not cool. *:or, as here, even 5 years later (!) when the journal Science posts on bluesky.

Wat goed van de Vk om hen uit te nodigen, wat goed dat ze er allebei op in zijn gegaan, en wat een goed stuk levert het op. En toch blijft het pijnlijk hoe weinig Levi lijkt te weten over hoe Nederlanders van Marokkaanse afkomst de politieke debatten van de afgelopen 25 jaar hebben ervaren.

Seriously though, I love that we’re learning more about animal communication, but always feel like language is not the best analogue. Music maybe closer?

Today @jumelet.bsky.social is defending 🎉 He's organising a nice workshop today on the intersection of LLMs and linguistics. He just kicked it off with this work with AO @wzuidema.bsky.social on using LLMs to improve our understanding of adjective order