I was happy to add my name to this list of employees and alumni of AI companies. These signatories have better insight than almost anyone else into what is coming next with AI and we should heed their warnings.

Thanks for visiting Sriraam Natarajan

Geoffrey Hinton Most of these signatories have a distorted view of what is coming next with AI. The distortion is due to their inexperience, naïveté on how difficult the next steps in AI will be, wild overestimates of their employer's lead and their ability to make fast progress, and financial

So true 👇 thanks Artur for saying this. I really hope more people start listening to you

In this #Chess position #GPTo1 answers exf7+ after a few seconds of "reflection". It's an illegal move (4th move!). I've repeated 5 times the session. ChatGPT4o and gpt3.5-instruct made similar mistakes: blog.mathieuacher.com/ChessWinning7M… It's a specific position but the limitation remains.

Our new computer chess architecture, MPC-MC, based on #ReinforcementLearning and model predictive control, beats engines found at Chess.com and lichess.org, such as Stockfish Chess Komodo Dragon and transformer-based at all levels. See: web.mit.edu/dimitrib/www/M… youtu.be/UeVs0Op-Ui4

For anybody attending KonKis next week, let me make a quick add read for Session 4: "Large AI Models by and for Europe." felfri and lukas helff will be presenting some of our recent safety research. And Malte Ostendorff is giving a talk on OcciGlot. events.gwdg.de/event/615/sess…

Answer the questions that people aren’t already asking but should be asking.

It was inspiring to attend the inauguration of the #Alps supercomputer for #AI and #HPC at CSCS Lugano in Lugano 🇨🇭 today and I‘m thankful for the opportunity to present the swiss-ai.org initiative on behalf of the team across ETH Zurich EPFL and other partners.

Ever wondered how to efficiently build hybrid architectures for accurate probabilistic time series forecasting? Then check out our recent AutoML_conf paper! Paper: openreview.net/forum?id=AaPhn… Joint work with Fabian, Zhongjie, Fabrizio Ventola and Kristian Kersting. Details ⬇️

Sorry for raining on o1’s parade, but the more we train LLMs, the worse they get on generating and understanding captions for the New Yorker cartoons. 🧵on our latest study on humor in AI: •Large gap between AI & top human submissions •Dataset with 250M+ human ratings •New

We have an opening for a PostDoc on Efficient Multimodal Models. Details see here: linkedin.com/jobs/view/4027… Apply here: informatik.tu-darmstadt.de/mai/multimodal…

60 Jahre Elektronenbeschleunigung an der TU Darmstadt - das feiern wir mit einem Tag der offenen Tür am 1.10.24 von 15 bis 18 Uhr! 🧑‍🔬 Meet a Scientist 📷 Ausstellung & Exponate 🏎️ Führungen zum #Teilchenbeschleuniger (mit Anmeldung) ℹ️ ikp.tu-darmstadt.de/60 #elementsfüralle

BlackRock is preparing to launch a more than $30bn artificial intelligence investment fund with technology giant Microsoft to build data centres and energy projects to meet growing demands stemming from AI, people briefed about the matter said on.ft.com/3Bh2aGG

Every day, I get more excited about European startups. Today, we announce $865M and a new team to help growth-stage founders win, bigger. sifted.eu/articles/headl…

Many of the talks in UAI'24 can now be accessed using the link below; check them out! youtube.com/playlist?list=…

🎊 It has arrived 🎊, the 2nd edition of my "Deep Generative Modeling" book. It has 100 new pages, 3 new chapters (incl. #LLMs) and new sections. It covers all deep generative models that constitute the core of all #GenerativeAI techs! Check it out: 💻tinyurl.com/mwj9dw83

😅

Black Forest Labs — a startup building generative AI image models, which only came out of stealth two months ago with $31 million in funding — is closing new funding tcrn.ch/3BdxmH7

Great kick-off of Lab 1141. A lot of interesting ideas, talks, and discussions. I am looking forward to all the new #AI stuff we will create. Aleph Alpha TU Darmstadt