AI research

Summary Amazon is buying itself and its customers’ priority access to Anthropic’s foundational models. As Amazon announced in a press release, the online retailer and cloud provider plans to invest up to four billion US dollars in the AI startup Anthropic. Amazon has thus secured an important partner in the field of generative AI. Anthropic …

Amazon invests up to four billion dollars in OpenAI competitor Anthropic Read More »

Summary An experiment shows that language models cannot generalize the simple formula “A is B” to “B is A”. But why is that? If Tom Cruise’s mother is Mary Lee Pfeiffer, who is Mary Lee Pfeiffer’s son? To humans, the answer is obvious. But large language models fail to answer this question, according to a …

Language models know Tom Cruise’s mother, but not her son Read More »

Summary Google Deepmind’s AlphaMissense AI system has created a comprehensive catalog of “missense” gene mutations, providing valuable insights into their potential impact. The AlphaMissense AI model categorizes potentially dangerous missense variants, specific gene mutations that affect the function of human proteins and can potentially lead to diseases such as cystic fibrosis, sickle cell anemia, or …

Deepmind’s AlphaMissense predicts potentially disease-causing gene mutations Read More »

Researchers at Cornell University have developed a tiny quadruped robot powered by combustion actuators fueled by methane and oxygen. The insect-sized robot, published in Science, can jump 59 centimeters straight up and walk while carrying 22 times its own weight. The researchers aim to apply the power generated by the combustion actuators to large-scale, variable-recruitment …

Microscale internal combustion engine powers insect robot Read More »

Summary Researchers demonstrate a new method for creating and editing 3D avatars using Stable Diffusion and a new hybrid 3D representation of digital humans. Emerging techniques in artificial intelligence are enabling the creation of increasingly realistic virtual avatars of humans. Two recent research projects from the Max Planck Institute for Intelligent Systems and others now …

TECA is a generative AI for digital humans Read More »

Summary EfficientViT brings efficient computer vision to high-resolution cameras and could lead to more accurate results in autonomous driving or medicine. MIT researchers have unveiled a new computer vision model that can perform semantic segmentation of high-resolution images much more efficiently than previous models. Semantic segmentation involves tagging and categorizing all pixels in an image. …

EfficientViT brings massive speedup to computer vision Read More »

Google’s PaLM 2 gets better at math when you let the language model take a breather. In a paper, researchers have investigated whether language models such as GPT-4 or PaLM 2 are suitable as optimizers that automatically find solutions to predefined problems, such as recommending movies or solving grade school math problems. The language models …

Even a chatbot has to take a breather from time to time Read More »

Summary Ilya Sutskever, co-founder of OpenAI, explains why unsupervised learning works and how it relates to supervised learning. The core concept is compression – good compressors can become good predictors. In a talk at the Simons Institute for the Theory of Computing at UC Berkeley, Ilya Sutskever, co-founder of OpenAI, presented research that tries to …

OpenAI co-founder explains the secret sauce behind unsupervised learning Read More »

Summary Can mathematical proofs and formal verification help ensure the safety and controllability of advanced AI systems? In their new paper, Max Tegmark and Steve Omohundro outline a possible way to ensure the safe development of advanced artificial intelligence, including general artificial intelligence (AGI). Tegmark is a physicist, cosmologist and AI researcher, and as president …

In their quest for AGI safety, researchers are turning to mathematical proof Read More »

Summary Ilya Sutskever, co-founder of OpenAI, explains why unsupervised learning works and how it relates to supervised learning. The core concept is compression – good compressors can become good predictors. In a talk at the Simons Institute for the Theory of Computing at UC Berkeley, Ilya Sutskever, co-founder of OpenAI, presented research that tries to …

compression may be the key, says OpenAI co-founder Ilya Sutskever Read More »