Samual Hammond walks you through the institutional economics of an intelligence explosion

Breakthrough brain implant and digital avatar allow stroke survivor to speak with facial expressions for first time in 18 years.

Caglar Gulcehre, Tom Le Paine, Srivatsan Srinivasan, Ksenia Konyushkova, Lotte Weerts, Abhishek Sharma, Aditya Siddhant, Alex Ahern, Miaosen Wang, Chenjie Gu, Wolfgang Macherey, Arnaud Doucet, Orhan Firat and Nando de Freitas propose a simple algorithm for aligning LLMs with human preferences inspired by growing batch reinfor...

Shengyu Zhang, Linfeng Dong, Xiaoya Li, Sen Zhang, Xiaofei Sun, Shuhe Wang, Jiwei Li, Runyi Hu, Tianwei Zhang, Fei Wu and Guoyin Wang survey research works in the quickly advancing field of instruction tuning (IT), a crucial technique to enhance the capabilities and controllability of large language models (LLMs)

Maciej Besta, Nils Blach, Ales Kubicek, Robert Gerstenberger, Lukas Gianinazzi, Joanna Gajda, Tomasz Lehmann, Michał Podstawski, Hubert Niewiadomski, Piotr Nyczyk and Torsten Hoefler introduce Graph of Thoughts (GoT): a prompting framework that advances prompting capabilities in large language models (LLMs) beyond those offer...

Jessica Dai discusses the current trajectory of AI 'alignment' research and its focus on building a product that people will pay for, rather than solving grand challenges like human extinction. She highlights the financial incentives that shape alignment work and calls into question whether current approaches are well-suited ...

Ziniu Hu, Ahmet Iscen, Chen Sun, Kai-Wei Chang, Yizhou Sun, David A. Ross, Cordelia Schmid and Alireza Fathi from Google Research (Perception Team) introduce AVIS, Autonomous Visual Information Seeking with Large Language Models, a novel method that achieves state-of-the-art results on visual information seeking tasks.

Quantamagazine - How hard is it to prove that problems are hard to solve? Meta-complexity theorists have been asking questions like this for decades. A string of recent results has started to deliver answers.

Vitalik Buterin, co-founder of Ethereum, shares his views on Twitter's fact-checking tool

Jie Huang, Wei Ping, Peng Xu, Mohammad Shoeybi, Kevin Chen-Chuan Chang and Bryan Catanzaro propose RAVEN, a model that combines retrieval-augmented masked language modeling and prefix language modeling and introduce Fusion-in-Context Learning to enhance the few-shot performance by enabling the model to leverage more in-contex...

Hao Ouyang, Qiuyu Wang, Yuxi Xiao, Qingyan Bai, Juntao Zhang, Kecheng Zheng, Xiaowei Zhou, Qifeng Chen and Yujun Shen present the content deformation field CoDeF as a new type of video representation, which consists of a canonical content field aggregating the static contents in the entire video and a temporal deformation fie...

Robert Knight, Ludovic Bellier, Anaïs Llorens, Déborah Marciano, Aysegul Gunduz, Gerwin Schalk and Peter Brunner conducted the study to reconstruct a recognizable song from brain recordings

Quantamagazine - Giant black holes were supposed to be bit players in the early cosmic story. But recent James Webb Space Telescope observations are finding an unexpected abundance of the beasts.

VIDEO | The Simons Institute for the Theory of Computing released the recordings of a 5-day workshop on the ongoing revolution in transformers and large language models

Ariel N. Lee, Cole J. Hunter and Nataniel Ruiz present Platypus a family of fine-tuned and merged Large Language Models (LLMs) that achieves the strongest performance in the Open Source space (at the time of writing)

Leo Kozachkov, Ksenia V. Kastanenka, and Dmitry Krotov show that neuron–astrocyte networks can naturally perform the core computation of a Transformer, a particularly successful type of AI architecture

SEC Proposes New Requirements to Address Risks to Investors From Conflicts of Interest Associated With the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisers

GoogleAi propose a comprehensive benchmark to extract structured data and automate business workflows

With 3D Gaussian Splatting, Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler and George Drettakis introduce three key elements that allow us to achieve state-of-the-art visual quality while maintaining competitive training times and importantly allow high-quality real-time (≥ 100 fps) novel-view synthesis at 1080p resolution

Finbarr Timbers discusses several papers on Mixture of Experts (MoE) models, which have been receiving attention due to rumors about their use in GPT-4. MoE models are a form of model parallelism where input tokens select combinations of parameters for each input.

Big Think - Ethan Siegel addresses the common accusation of "groupthink" in science to go through the history of the geocentric and heliocentric models

TechCrunch - The credit facility, which was led by existing investors Blackstone and Magnetar Capital with participation from Coatue, DigitalBridge Credit and funds and accounts managed by Pimco and Carlyle, comes only weeks after CoreWeave’s announcement that it plans to build a $1.6 billion data center in Plano, Texas.

Joshua Harrison, Ehsan Toreini and Maryam Mehrnezhad present a practical implementation of a state-of-the-art deep learning model in order to classify laptop keystrokes, using a smartphone integrated microphone.

Quantamagazine - Inside of a black hole, the two theoretical pillars of 20th-century physics appear to clash. Now a group of young physicists think they have resolved the conflict by appealing to the central pillar of the new century — the physics of quantum information.

Zhewei Yao, Reza Yazdani Aminabadi, Olatunji Ruwase, Samyam Rajbhandari, Xiaoxia Wu, Ammar Ahmad Awan, Jeff Rasley, Minjia Zhang, Conglong Li, Connor Holmes, Zhongzhu Zhou, Michael Wyatt, Molly Smith, Lev Kurilenko, Heyang Qin, Masahiro Tanaka, Shuai Che, Shuaiwen Leon Song and Yuxiong He introduce DeepSpeed-Chat, a novel sys...