Maarten Paulusse

Tired of AI hype posts? You might like my sober assessment of whether AI-generated summaries are suitable for studying and research. Spoiler alert, they are not. The text is primarily aimed at students and researchers, but has much broader relevance. So share freely!: www.tue.nl/en/our-unive...

"Dutch research institutions will adopt a new Code of Conduct for Research Integrity with effect from 1 January 2027." www.nwo.nl/en/news/revi...

Third, automating science is not really possible in the way described. In the same way a machine cannot eat food for us because we'd starve: scientific understanding is for us to experience and share with others as a psychological state. It's a process and a cognitive state not an output.

Was reminded of this by someone's questions to me about how to argue against the (common) claim that LLMs can "summarize", so reboosting as a gentle reminder for you all too

we're in a time where "this email chain could have been a simple meeting" is a bizarre new truth that clashes wildly with just a few years ago where academics were whinging about meetings

How come Wikipedia is doing better than academic publishing? Btw love that icon: 🚫🤖 🔗 en.wikipedia.org/wiki/Pragmat...

“If we only see the mind as a computer, AI might seem advantageous to learning — this is why many dream of building AI tutors. But already such efforts are failing, because our minds are more than just input-output devices.” www.theverge.com/ai-artificia...

Dear @erc.europa.eu , we DO already have common standards & frameworks. 'AI' likes to pose itself as 'special' but we do not need to reinvent the wheel. We already have established codes of conduct for research integrity. We just need to apply them. See below + more resources in thread🧵

Tech CEOs are promising us AI that will soon be as capable as humans (AGI). Some computer scientists are not convinced. Iris van Rooij has warned about the “intractability” of human-like computer systems for decades. We are now seeing signs LLMs are indeed already unscalable at levels far below AGI.

Entirely as expected (see link.springer.com/article/10.1...) --- “Epoch AI … published a graph last year … It shows the exponentially increasing costs of serving more “tokens”—the words users type to chatbots—with several public AI models.” article: archive.ph/IrNBI (h/t @maartenp.bsky.social)

I'm attending a digital humanities event in Montreal. There was a keynote on AI and something about the talk made me wonder if it had been written by Claude. I said as much in the Q&A. As I posed the question, the speaker shifted, looking slightly uncomfortable. What he said next shocked the room +

At the end of the term I asked my college creative wriing students to submit anonymous thoughts on AI. No real surprises: Mood ranges from resignation to despair, capitulation from embittered erosion of standards to total, feelings of betrayal from deep to furious. 1/

This is funny, because it confirms that LLM-powered chatbots are easily jailbroken (guardrails are a sham) and it makes Anthropic admit that despite their scaremongering hype marketing, Mythos is not some super hacker. It can spot simple, known exploits like any other model.

@emilymbender.bsky.social @alexhanna.bsky.social Please consider giving this the Mystery AI Hype Theater 3000 treatment…

I guess I'll keep shouting this message from the rooftops until it sticks. Lumping disparate technologies under the ✨"AI"✨ label only inhibits clear discussion and reasoning about what each does. >>

Al sinds september heb ik kanker. De ziekte is nu weg, maar om terugkeer te voorkomen of vertragen is een medicijn nodig dat voor mij niet wordt vergoed. Gelukkig zijn mijn vrienden een crowdfunding gestart. Alle donaties zijn welkom en delen helpt ook! www.gofundme.com/f/help-floor...

I desire more in life than to be the “don’t summarise with AI”-guy on bluesky 😅 but reports like this keep coming in, so I’ll repeat ad nauseam, don’t summarise with ‘AI’! www.tue.nl/en/our-unive...

this graduation speech moment is notable, and her amazed shock at having failed to read the room feels instructive. when you're inside the bubble, you think everybody else is. but everybody isn't.

Meet the academics refusing to use generative AI Researchers say they have their reasons for avoiding AI tools — and they’re sick of arguing about it. www.nature.com/articles/d41... h/t @maartenp.bsky.social

The world is now so full of ridiculous things that I struggle to deal with it all. But this is not an 'us' problem. The world really is idiotic. I made a list of things that are impossible to believe. Seeing it in writing might help you deal better with the situation. berthub.eu/articles/pos...

The entanglement of Silicon Valley with politics and other public domains such as education should worry us all, especially in the age of GenAI.

This set is a celebration of electronic music, community, creativity and life. Amazing: youtu.be/gfF8jzBVWvM?...

Wanneer weet je dat de periode die je beschrijft de geschiedenis zal veranderen en wanneer weet je zeker dat iets een voetnoot zal blijken te zijn? Na 7,5 jaar stop ik als politiek verslaggever. Lees hier mijn afscheidsstuk: www.nrc.nl/nieuws/2026/...

in the end where claude gives him an answer on moratoriums that he doesn’t like, so he pushes back and it gives him the answer he wants is a pretty pure example of the reinforcement bias and sycophancy we talk about with chatbots, but you get the impression he convinced claude he was actually right

Google is now screwing with the 10 blue links in traditional search and rewriting headlines - including ours - to be the worst kind of slop. This sucks so bad www.theverge.com/tech/896490/...

Arnold's editorial team is modeling critical thinking and I am here for it.

Heel mooi essay van @chrisjulien.bsky.social dit: www.nrc.nl/nieuws/2026/...

Is this the future we want? nymag.com/intelligence...