Eine Antwort warum wir #highspeed #Internet in 🇩🇪 brauchen ... Wieviel Zeit geht verloren weil wir 30+ min brauchen um unsere Entwicklungstools zu laden? .... will mir garnicht vorstellen wie das in der #Bahn oder in ländlichen Gegenden funktionieren soll. #breitbandausbau

#rainbow|s are one of my favorite things when it comes to the physics of light. So naturally this one at Feigumfossen in #Norway is very special! 🌈🇸🇯

No More GIL! the Python team has officially accepted the proposal. Congrats Sam Gross on his multi-year brilliant effort to remove the GIL, and a heartfelt thanks to the Python Steering Council and Core team for a thoughtful plan to make this a reality. discuss.python.org/t/a-steering-c…

♥️I love this revival of old tech! A punch card as a hotel key 🗝️ So retro and ingenuous 🧠

🧠-time: Luhmann in his lecture on Systems Theory essentially describes autoregression as the only way to perceive the world: "What I am thinking just now, [...], what I perceive, is that which constitutes the starting point for the conceivability of further perceptions." 🤯

Our paper "Curve Your Enthusiasm: Concurvity Regularization in Differentiable Generalized Additive Models" was accepted at #NeurIPS2023 🥳 Great Merantix Momentum teamwork with Julien Siems, Konstantin Ditschuneit, Winfried Ripken, Alma Lindborg, Max Schambach, and Johannes Otterbach!

#siftedsummit

⚗️ Science: Simple things interacting at scale lead to emergent complexity. 🚂 Engineering: To handle complexity at scale simplicity is the name of the game. I feel blessed to work in an industry where you need to leverage both aspects to be successful!

The power of systems thinking is rooted in good abstractions. They don't give you solutions. Rather, they let you ask questions in the right language, giving you 80% of the solution. Success lies in thinking about good abstractions most of the time, execution is fast afterwards.

Healthy computation graphs give me an outsized amount of happiness 🤗. There is something magical about understanding and managing the complexities of large scale distributed systems ... 🪄

The recent wave of "AGI arrives in <X> years" by tech pioneers appears more driven by a fight over the right to define AGI, rather than factual statements. Claiming a moving goal post is more about legitimizing yourself rather than advancing understanding ... really annoying ...

AGI is always 3-5 years away

Probably an unpopular opinion these days ... but: #opensource #AI is not equivalent to #opensource #software and #opensource weights are not a product ... Imagine any pharma or drug company just releasing their stuff without any responsible testing ...

OpenAI is now entering the same realm as other social media companies. Amplification of certain news sources (in this case the Fox News of Germany) without accountability to present nuanced and balanced opinions will repeat the same mistakes again. Sad to see ...

This is pretty much what I've been saying. Long-term insists are missing the discount factor in their risk assessment. If you include this the most relevant challenges are on a decade time scale not infinite horizon...

Humans largely learn from human-generated data. Machines currently learn from human-generated data. It makes sense for machines to learn from machine-generated data.

Hybrid architectures FTW!

I made this observation before ... It feels large scale modern AI systems are going the same way as experimental high energy physics. So be prepared to soon crack >1000 authors on papers regularly.

“resolve conflict with curiosity” ... powerful wisdom from Will Larson. Thank you for this crisp formulation!

Stranden am Münchener Flughafen nach mehrstündiger Flugverspätung und ausweichen auf den ICE nach Berlin erinnert mich an dieses Glanzstück bayerischer Politik Stilblüten ... 🤣