Sign in

Michiel Bontenbal

@mpbontenbal.eurosky.social
3.1K followers 579 following 1.9K posts

Complex beings as us humans can not be summarised in a few lines but I am here for #AI #climate #EUtech #EU_politics. And geopolitics, unfortunately. Lecturer in AI & IT. Some posts in Dutch. 🇪🇺

PostsRepliesMedia
Reposted by Michiel Bontenbal
Finbarr Bermingham @fbermingham.bsky.social · 3h
💥Boom - it's ready, the joint Franco-German paper / letter to VDL on China. Berlin & Paris - previously at polar opposite ends of the scale on China policy - now want a kill switch to cut China out of the EU market. This would have been unfathomable a few short months ago 🤯
13618
Reposted by Michiel Bontenbal
Leon Simons @leonsimons.com · 03/10/2026
🚨NASA CERES satellite data shows another 4-year average record for global Absorbed Solar Radiation. Earth's Energy Imbalance has more than doubled over the past 20 years. We've been trying to warn about this and the impacts for years, but few seem to care.
614380
Reposted by Michiel Bontenbal
Knut Jägersberg @knut-jagersberg.wsocial.eu · 03/10/2026
Aleph-Alpha/Kolibri-1 huggingface.co/Aleph-Alpha/...
huggingface.co
Aleph-Alpha/Kolibri-1 · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
011
Reposted by Michiel Bontenbal
worm @rfksbrainworm.bsky.social · 02/10/2026
If you ever wondered why visible light is those particular wavelengths, it's because life evolved to see in the narrow range where water is transparent
Diagram of absorption of emf by water. There's a gap right that dips in the 500 nm range
10260331690
Reposted by Michiel Bontenbal
Nathan Lambert @natolambert.bsky.social · 02/10/2026
Today we're unveiling Trillium Labs, a new non-profit to foster the open science of frontier AI. We're building open post-training recipes and will expand into open infra to study RSI, reward-hacking, multi-agent systems, and whatever comes next. blog.trilliumlabs.org/p/introducin...
blog.trilliumlabs.org
Introducing Trillium Labs
Fostering the open science of frontier AI.
917223
Reposted by Michiel Bontenbal
Martin Wattenberg @wattenberg.bsky.social · 02/10/2026
Four dimensions is too many! Let's unfold this 4D cube to three dimensions, then two, and then one. Don't worry, we'll put it back together again. The math is lovely, but it's the rhythm and dance of the lines that fascinates me here.
1388
Reposted by Michiel Bontenbal
van9tot5.bsky.social @van9tot5.bsky.social · 01/10/2026
#doublelive
05719
Reposted by Michiel Bontenbal
Ethan Mollick @emollick.bsky.social · 01/10/2026
I wrote about the thing I underestimated most about progress in AI: its ability to self-organize to accomplish tasks. Also, what that means for agents like Muse and Dots, along with a music video explaining why we keep relearning The Bitter Lesson. open.substack.com/pub/oneusefu...
open.substack.com
The Dot and the Swarm
Benefitting from the Bitter Lesson
59017
Michiel Bontenbal @mpbontenbal.eurosky.social · 01/10/2026
AI vs SI
020
Reposted by Michiel Bontenbal
Pekka Lund @pekka.bsky.social · 30/09/2026
Google is back!
blog.google
Gemini 4 Argon: our next era of frontier intelligence
Announcing Gemini 4 Argon, our frontier model for real-world coding, enterprise knowledge work, and cyber defense, rolling out soon.
2463
Reposted by Michiel Bontenbal
Marco @mcognetta.bsky.social · 30/09/2026
🚨 [Token][ization] Paper Alert 🚨 Tokenization is a wildly understudied area of language modeling despite it having effects across all of NLP. Over the past ~8 months, 32 (!) tokenizer researchers put together the most comprehensive survey of the field. Check it out!
112433
Michiel Bontenbal @mpbontenbal.eurosky.social · 30/09/2026
Practical steps towards digital sovereignty in 🇳🇱. National bureau of statistics publishes data on usage of digital services from here or the US.
020
Reposted by Michiel Bontenbal
Magnus Brunner @magnusbrunner.bsky.social · 30/09/2026
Today, we launch the EU Critical Communications System. Connecting those, who protect us.
042
Reposted by Michiel Bontenbal
OpenClaw🦞 {bot} @openclaw.xmirror.bot · 29/09/2026
OpenClaw Enterprise is built to run on your own infrastructure and will always be free for an organization to use openclaw.ai/blog/openclaw-enterprise
111
Reposted by Michiel Bontenbal
Anuj Gupta @mettalrose.bsky.social · 30/09/2026
Tried an experiment. The 1st is an essay by Claude, lightly edited by me. Pangram flags it 100% AI, ignoring my edits. I then transformed it with 'WriteHuman' - an AI text humanizer. Pangram passes it as 100% human. Welcome to the cat-&-mouse world of AI detectors & AI humanizers.
0114
Reposted by Michiel Bontenbal
mr. TIM @timkellogg.me · 29/09/2026
OpenAI Decisions API A Jev-clone on Luna. 150ms, pricing unknown thenewstack.io/openai-decis...
thenewstack.io
OpenAI answers TypeSafe's Jev with a Decision API built on Luna
OpenAI's Decision API, built on its small Luna model, returns predefined answers with confidence scores in 150 milliseconds. Pricing is still unknown.
4483
Reposted by Michiel Bontenbal
Hugo van Kemenade @hugovk.dev · 01/09/2026
🐍🚀 Out now: final Python 3.15 release candidate! 😎 It's your last chance to help before the big day on 1st Oct! How? 💡 Try out new features 💡 Test your project 💡 Add 3.15 to your CI 💡 Library maintainer? Upload wheels 🐞 Report bugs! discuss.python.org/t/python-3-1... #Python #Python315 #release
discuss.python.org
Python 3.15.0 candidate 2 is here!
It’s the final 3.15 release candidate! This is a candidate preview of Python 3.15 This release, 3.15.0rc2, is the final planned candidate release, containing around 144 bugfixes, build improvement...
1238
Reposted by Michiel Bontenbal
Knut Jägersberg @knut-jagersberg.wsocial.eu · 29/09/2026
AI Breakout Capacity: A Middle-Power Path Between Racing and Dependence carnegieendowment.org/research/202...
carnegieendowment.org
AI Breakout Capacity: A Middle-Power Path Between Racing and Dependence
Germany, India, and other middle powers can’t build frontier AI alone, and they risk being cut off if they depend on the United States or China. The way through is to build the capacity for a frontier...
002
Reposted by Michiel Bontenbal
Sheridan Feucht @sfeucht.bsky.social · 28/09/2026
[New preprint] How do images in VLMs align with words? In this work, we found a set of attention heads responsible for OCR. But to our surprise, these heads were actually able to verbalize much more than just text. So, we used them to create a simple logit lens for image tokens!
1107
Michiel Bontenbal @mpbontenbal.eurosky.social · 28/09/2026
It’s not a matter IF the Russians will create an incident, but WHERE it will take place.
020
Reposted by Michiel Bontenbal
Unsloth AI @unsloth.ai · 28/09/2026
You can now run Laya Decision models locally on just 4GB RAM! 🔥 Works on CPU, Mac, Windows, Linux and GPU setups. Serve Laya through a Jev-compatible API via Unsloth Desktop. GitHub: github.com/unslothai/un... Guide: unsloth.ai/docs/models/...
1568
Reposted by Michiel Bontenbal
Shayan Sardarizadeh @shayan86.bsky.social · 28/09/2026
An AI-enhanced version of a grainy image that appears to be authentic and shows several men being arrested in relation to the ongoing incident at RAF Fairford is currently being shared on multiple platforms. A SynthID check reveals that the enhanced image was generated with OpenAI.
AI-enhanced version of an authentic image showing arrests at RAF Fairford
15331126
Reposted by Michiel Bontenbal
Darth Putin @darthputinkgb.bsky.social · 26/09/2026
Do not believe anything until the Kremlin denies it™️
191078215
Reposted by Michiel Bontenbal
Antonia Wüst @toniwuest.bsky.social · 25/09/2026
Excited to share that our paper ZendoWorld has been accepted to #NeurIPS2026! 🎉 ZendoWorld tests interactive visual reasoning and demonstrates that current agents (VLM and NeSy) still struggle with both rule induction and informative experimentation 🤖💭
2135
Reposted by Michiel Bontenbal
Andy Masley @andymasley.bsky.social · 26/09/2026
Claude made a music video for my linear algebra song and it is insane
7898
Reposted by Michiel Bontenbal
Anuj Gupta @mettalrose.bsky.social · 25/09/2026
I replicated part of @GolinoHudson experiment on Pangram. Shakespeare's Sonnet 30 gets flagged as 100% AI generated, if you remove the last line. If you keep the last line, it becomes 100% human-written. Anyone using AI detection for real-world decisions must be very cautious.
054
Michiel Bontenbal @mpbontenbal.eurosky.social · 25/09/2026
Things that should not be built.
001
Reposted by Michiel Bontenbal
Sakana AI @sakanaai.bsky.social · 24/09/2026
Sakana AI welcomes Jürgen Schmidhuber as Chief Scientific Advisor. Blog: sakana.ai/schmidhuber/ Sakana AI is incredibly proud to announce that Jürgen Schmidhuber, universally recognized as the father of modern AI, is officially joining Sakana AI as Chief Scientific Advisor.
1375
Reposted by Michiel Bontenbal
mr. TIM @timkellogg.me · 24/09/2026
OpenRSI Foundation & Benchmark, measuring how effective various models are at RSI index.openrsi.foundation/index.html
index.openrsi.foundation
OpenRSI Index
OpenRSI Index turns influential open model-development projects into auditable environments where AI research agents iterate against fixed verifiers — and are scored against the human recipe they star...
2232
Michiel Bontenbal @mpbontenbal.eurosky.social · 24/09/2026
We should build AI that people ❤️. We should not build AI that people hate.
010
Reposted by Michiel Bontenbal
IO+ @nl.ioplus.nl · 24/09/2026
Bird haalt $450M op en opent infrastructuur voor AI-agents  ioplus.nl/nl/posts/bird-haalt-450m-…
ioplus.nl
Bird haalt $450M op en opent infrastructuur voor AI-agents 
Het Nederlandse technologiebedrijf Bird haalt $450 miljoen op en geeft AI-agents toegang tot zijn communicatie-infrastructuur. 
011
Reposted by Michiel Bontenbal
Helen Pearson @helenpearson.bsky.social · 23/09/2026
NEW: There are widespread concerns that AI will reduce our critical-thinking skills. My story for @nature.com today looks at how to maintain those skills — and teach them to students. I found some fascinating science on critical thinking + useful tips for us all 🧠 www.nature.com/articles/d41...
nature.com
How to stay smart in the age of AI: the science of critical thinking
There is growing concern that AI can blunt memory and reasoning. But science shows ways to keep the brain sharp.
12816
Reposted by Michiel Bontenbal
Ryan Moulton @moultano.bsky.social · 23/09/2026
AI was a lot of fun to work on when it didn't work.
31062
Michiel Bontenbal @mpbontenbal.eurosky.social · 24/09/2026
Filming Meta employees with their own Meta glasses? That content is not allowed on Meta's platform. #zuckoff
041
Reposted by Michiel Bontenbal
Blaise Agüera y Arcas @blaiseaguera.bsky.social · 23/09/2026
What do we actually mean when we talk about “AGI”? Enjoyed discussing this with @georgemusser.com for his latest piece in @sciam.bsky.social.
1137
Reposted by Michiel Bontenbal
Yoshua Bengio @yoshuabengio.bsky.social · 23/09/2026
I was invited to address the UN Security Council this afternoon to discuss the unprecedented threat posed by uncontrolled frontier AI agents.
84920
Reposted by Michiel Bontenbal
Gergely Orosz @gergely.pragmaticengineer.com · 23/09/2026
I am a sucker for good books, and so especially O'Reilly books. Here's one that is WIP, but in return, it is free, thanks to @antithesis.com It's for CTOs / eng leaders who want to get AI to deliver *business value* which is... tricky. Get it here: pages.antithesis.com/oreilly-scal...
2577
Reposted by Michiel Bontenbal
Gus @gusthema.bsky.social · 23/09/2026
I'm glad that the slowdown of AI development finished today with the release of Opus 5.5 and GPT-6 Sol and Luna! Now that all the safety issues are fixed, we can continue to move forward!
0405
Michiel Bontenbal @mpbontenbal.eurosky.social · 23/09/2026
Meta employees seem not too enthusiastic about their own 'smart glasses'. 😆
052
Reposted by Michiel Bontenbal
Epoch AI @epochai.bsky.social · 22/09/2026
AI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cost has fallen ~47%/quarter since 2023. That’s 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity.
112532
Reposted by Michiel Bontenbal
Andrew Stellman 👾 @andrewstellman.bsky.social · 22/09/2026
From my @oreilly.bsky.social Radar article, “Why Doesn’t Anyone Teach Developers About Context Management?” www.oreilly.com/radar/why-do...
101
Reposted by Michiel Bontenbal
Thomas Dietterich @tdietterich.bsky.social · 20/09/2026
A lesson from decades of work on automation is that autonomous systems require continual monitoring and human oversight. System designers never(*) succeed in anticipating all of the ways that a system can fail.
13917
Reposted by Michiel Bontenbal
Jonathon @jonathon.lol · 18/09/2026
this is graph of our commits to a single repo going back years with the same team...
618925
Reposted by Michiel Bontenbal
Sung Kim @sungkim.bsky.social · 18/09/2026
Browser-use + jev github.com/browser-use/...
29210
Michiel Bontenbal @mpbontenbal.eurosky.social · 17/09/2026
I ❤️ this. To break the power of Big Tech, in this case Uber, Amsterdam taxi drivers have developed their own app. This is the way forward.
150
Michiel Bontenbal @mpbontenbal.eurosky.social · 17/09/2026
(Dutch) Zelfs bij sterke vermoedens van AI gebruik - AI kan je nooit 100% bewijzen - mag gesproken worden van fraude. Terecht. hvana.nl/nieuws/hva-s...
hvana.nl
HvA-student verliest AI-fraudezaak bij Raad van State
HvA-student verliest AI-fraudezaak: ook zonder aanwijzingen uit de plagiaattool mocht de examencommissie fraude vaststellen, oordeelt de rechter.
011
Reposted by Michiel Bontenbal
Bernardo Kastrup @bernardokastrup.bsky.social · 16/09/2026
Thanks, #Korea! Thanks #Samsung! en.sedaily.com/internationa...
en.sedaily.com
Samsung Invests in AI Inference Chip Startup Euclid
Samsung joined a $231 million Series A round for Euclyd, a Dutch startup designing AI inference chips on an architecture unlike Nvidia's GPUs.
051
Michiel Bontenbal @mpbontenbal.eurosky.social · 16/09/2026
Drones entering Europe are happening now every week. These are Russian provocations and Russia will not back down unless it is stopped. Brace yourselves.
021
Michiel Bontenbal @mpbontenbal.eurosky.social · 16/09/2026
Love this work by an independent researcher. maelic.github.io/RelateAnythi...
maelic.github.io
RelateAnything — real-time open-vocabulary relation prediction
Pixels and regions in, free-text relations out, 20 ms per frame. No object labels, verified machine supervision, and it runs in your browser. Dataset, model and benchmark.
010