marcelsalathe.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

engineeringprompts.substack.com/p/weekend-re...

submitted 5 days ago • 0 comments

Three years of war in Ukraine - © Chappatte in Le Temps, Geneva 👉 www.chappatte.com/en/images/th...

submitted 5 days ago • 0 comments

Super News aus der SPK-N! Zusammen mit @marcelsalathe.bsky.social haben wir vor ein paar Monaten in der NZZ zusammengefasst, was es für ein sicheres 🇨🇭e-collecting braucht: www.nzz.ch/meinung/e-co...

submitted 6 days ago • 1 comment

Die Schweiz hat weder ein Ausgaben- noch ein Schuldenproblem. Aber wenn sich die Finanzpolitik des Bundesrates durchsetzt, bekommt unser Land ein Zukunftsproblem. In der heutigen Südostschweiz habe ich für eine andere Finanzpolitik argumentiert👇

submitted 8 days ago • 2 comments

For this next episode of #InsideAI, @marcelsalathe.bsky.social talks with Nicolas Dessaigne, from Y Combinator. Gain insights on startups, keys to founders’ success and why we might be on the cusp of the next consumer AI revolution. 🎙️ Listen here on Spotify: open.spotify.com/episode/0dD9...

submitted 9 days ago • 1 comment

We are excited! Tomorrow February 11, we open the doors on AMLD EPFL 2025. Join us in Lausanne for 4 action-packed days. Get your ticket! https://buff.ly/3UYv8Cr 🗓️ February 11-13: Three-day Conference +400 expert speakers, keynotes, 46 tracks, 21 exhibitors 🗓️ February 14: Workshop Day

submitted 17 days ago • 0 comments

Only 2 days until AMLD EPFL 2025 opens its doors! Don't miss this incredible opportunity to network, collaborate and innovate. Delve into the latest AI and machine learning breakthroughs and be part of shaping the future with AI. Join us February 11-14. Get your ticket today. https://buff.ly/3UYv8Cr

submitted 18 days ago • 0 comments

I'm not posting a lot here, but if you're interested in my weekly update on AI, I have a low-volume newsletter: engineeringprompts.substack.com

submitted 20 days ago • 0 comments

🎙️In the latest episode of #InsideAI, @marcelsalathe.bsky.social talks to Daniel Huttenlocher from MIT. They explore how AI impacts education, jobs and geopolitics. Tune in on... 👉 Spotify: open.spotify.com/episode/1scr... 👉 Apple: podcasts.apple.com/ch/podcast/c...

submitted 20 days ago • 0 comments

⭐ Only 1 week until #AMLDEPFL2025! Join us to hear from 400+ inspiring speakers. Attend comprehensive tracks, visit the vibrant exhibition, participate in hands-on workshops, & experience unmatched networking. February 11-14 ⭐ Get your ticket 👉https://buff.ly/40WitTP.

submitted 23 days ago • 0 comments

Got your ticket for AMLD EPFL 2025? Our exhibition will be buzzing with a fantastic lineup of exhibitors. This is an amazing opportunity to talk one-on-one with leading companies and innovative startups. Get your ticket today! https://buff.ly/3UYv8Cr February 11-14. SwissTech Convention Center

submitted 24 days ago • 0 comments

Ep 2 of #InsideAI is out! 🎙️🎉 We sat with Google Fellow Urs Hölzle. From optimizing data centers to AI efficiency, AGI, autonomous vehicles & Europe’s AI outlook—don’t miss this insightful conversation. On Spotify: open.spotify.com/episode/5Tu9... or Apple Podcast: podcasts.apple.com/us/podcast/c...

submitted 29 days ago • 0 comments

Interested in running VS Code with a local LLM? github.com/ggml-org/lla...

submitted 34 days ago • 0 comments

Join us in shaping the future with AI @ #AMLDEPFL2025 and hear from an amazing lineup of over 400 expert speakers including keynote speakers: Nicolas Dessaigne, Mark Greaves, Mackenzie Mathis & Sepp Hochreiter. 🗓️ February 11-14 🎟 Get your ticket here https://buff.ly/3UYv8Cr

submitted 36 days ago • 0 comments

The #AI risk executive order in the US is revoked. Status today: 🇺🇸US: heavy on tech, light on regulation 🇪🇺Europe: light on tech, heavy on regulation www.reuters.com/technology/a...

submitted 36 days ago • 0 comments

40% seems low. Great to have the data, but let’s not attach some kind of implicit “failure” sentiment to it. Isn’t it just perfect if most people are in academia for a limited period of training? And a few find it in their calling to stay? www.nature.com/articles/d41...

submitted 37 days ago • 0 comments

The latest from my team in Nature Medicine. We comprehensively mapped the benefits and risks of GLP-1 receptor agonists (GLP-1) (Ozempic, Wegovy, Mounjaro, Zepbound). By @Biostayan, Taeyoung Choi and me A 🧵 www.nature.com/articles/s41...

submitted 37 days ago • 17 comments

US #AI restrictions - excellent SemiAnalysis deep dive. Worth repeating: limits are on COMPUTE, not chip numbers. Want a more powerful chip? Your allocation goes DOWN. The geopolitical aftershocks of this regulation will be massive. Absolute bunkers. semianalysis.com/2025/01/15/2...

submitted 38 days ago • 0 comments

In a very large dataset of people with diabetes treated with GLP-1 drugs cf other medications, there were benefits seen for reduced clotting disorders, seizures, neurocognitive disorders (including Alzheimer's), by @zalaly.bsky.social and colleagues www.nature.com/articles/s41...

submitted 38 days ago • 8 comments

I had a chance to sit down & discuss with JP Fricker, EPFL alumnus and cofounder of AI chip unicorn Cerebras Systems - the company behind the world's largest AI chip. It's the first episode of the new @epfl-ai-center.bsky.social podcast, "Inside AI". Many more to come! Link to pod in comment

submitted 40 days ago • 2 comments

👏 Huge congrats to @epfl-ai-center.bsky.social colleagues on developing an AI tool to design proteins that bind only when specific drugs are present, enabling precise control of therapeutic proteins. A game-changer for medicine! 🤩 OA paper: www.nature.com/articles/s41...

submitted 41 days ago • 0 comments

🐍 Important AI breakthrough: Using deep learning to design proteins that neutralize deadly snake venom toxins, protecting animals even after exposure. Combined with rapid snake ID via smartphones, this could save thousands of lives from snakebites each year! 🧬📱 Paper: www.nature.com/articles/s41...

submitted 41 days ago • 0 comments

Does ChatGPT use 10x more energy than a standard Google search? TLDR: Probably, though absolute numbers are much lower, and it really doesn’t matter much. More details: engineeringprompts.substack.com/p/does-chatg...

submitted 41 days ago • 0 comments

Elon’s extreme hatred of his former friend Sam Harris seems to have begun after Elon made a bet with Sam that Covid would never top 35,000 cases in the US, and was proven spectacularly wrong. samharris.substack.com/p/the-troubl...

submitted 42 days ago • 43 comments

We are delighted to welcome back longstanding partner Google to #AMLDEPFL2025. Join us & connect with industry leaders, academics & innovative startups, who are shaping the future with AI. 🗓️ February 11-14, 2025 🇨🇭SwissTech Convention Center EPFL, Lausanne 🎟 Get tickets https://buff.ly/3UYv8Cr

submitted 49 days ago • 0 comments

🚀 Join us for Workshop Day at AMLD EPFL 2025!🚀 Participate in 16 hands-on workshops on February 14, 2025, at the SwissTech Convention Center, Lausanne. Details & Schedule: https://buff.ly/4gDyfrS Get your ticket today https://buff.ly/3UYv8Cr

submitted 45 days ago • 0 comments

Quite astonishing how many otherwise smart people fall for the “AI will make us all stupid” trap. You literally have thousands of years of evidence against that bet. You'd better have overwhelming evidence that this time is different.

submitted 46 days ago • 0 comments

Just saw a post that said “only liars would be against fact checkers” You can’t be much more naive than that. Only a fool would claim that his social network was the truth.

submitted 48 days ago • 0 comments

The illnesses & deaths in Panzi region, DRC, were caused by a combo of regular respiratory bugs + malaria, in people weakened by severe malnutrition. Not a new disease but definitely a local tragedy. www.who.int/emergencies/...

submitted 62 days ago • 4 comments

super interesting to me that "AI = the worst chatbots = theft = ecocide = right-coded" has become an article of faith for a decent swath of users of Bluesky like, I totally get why, but it's also a case study in how something can become politicized bsky.app/profile/ajh....

submitted 62 days ago • 44 comments

submitted 66 days ago • 0 comments

I’ll say it first. Last Christmas by Wham is the best Xmas song ever written. Period.

submitted 67 days ago • 0 comments

I’m looking forward to the time when, thanks to #AI, it will be impossible to drive a car into a crowd of humans. 😞

submitted 68 days ago • 0 comments

Submit your caption

submitted 68 days ago • 1 comment

This may be the most remarkable paragraph I’ve read about AI this year. It shows a level of self-awareness (in the technical sense) that’s just mind boggling.

submitted 70 days ago • 1 comment

Academic writing is getting harder to read - the humanities most of all https://www.economist.com/science-and-technology/2024/12/18/academic-writing-is-getting-harder-to-read-the-humanities-most-of-all

submitted 70 days ago • 2 comments

Using an objective biomarker for wine consumption in people with cardiovascular risk supported CV benefit of light to moderate consumption academic.oup.com/eurheartj/ad...

submitted 71 days ago • 10 comments

Context windows in #AI models are increasing massively, but this study suggests anything beyond 10,000 tokens and you're asking for trouble. arxiv.org/abs/2406.10149

submitted 70 days ago • 0 comments

Good news on a H5N1 vaccine efficacy in a preclinical model in case it is needed science.org/doi/10.1126/...

submitted 70 days ago • 14 comments

🚀 Join us for Workshop Day at AMLD EPFL 2025!🚀 Participate in 16 hands-on workshops on February 14, 2025 - SwissTech Convention Center, EPFL, Lausanne. Schedule: https://buff.ly/4gDyfrS Get your ticket today https://buff.ly/3UYv8Cr Don’t miss this chance to elevate your skills in #AI and #ML

submitted 73 days ago • 0 comments

Yesterday, I experienced something unsettling: an AI that refused to admit it was wrong about a simple piano piece. It kept insisting I was the one mistaken - about music I've played for 15 years. Here's what happened: https://engineeringprompts.substack.com/p/sorry-human-youre-wrong

submitted 74 days ago • 2 comments

Confidently wrong: No model so far was able to answer this correctly. Not o1 pro, not Gemini advanced, not Claude Opus. The "better" the model, the more confident it was in its wrong answer. At least Mistral and Claude Sonnet were able to say they didn't know.

submitted 75 days ago • 1 comment