Apr 11, 2024
AI race heats up as OpenAI, Google and Mistral release new models
Posted by Rx Sobolewski in category: robotics/AI
Launches within 12 hours of one another, and more activity expected in industry over summer.
Launches within 12 hours of one another, and more activity expected in industry over summer.
From U tubingen and cambridge U
Wu’s Method can Boost Symbolic AI to Rival Silver Medalists and AlphaGeometry to Outperform Gold Medalists at IMO Geometry https://arxiv.org/abs/2404.
- Wu’s…
Continue reading “AlphaGeometry: An Olympiad-level AI system for geometry” »
Emo the robot has been developed by researchers at Columbia University to both read and mimic human’s non-verbal cues, like eye contact and smiling.
Fraunhofer researchers developed an easy-to-operate, unmanned watercraft that autonomously surveys bodies of water both above and below the surface and produces corresponding 3D maps.
The unmanned watercraft uses its GPS, acceleration and angular rate sensors, and a Doppler velocity log (DVL) sensor to incrementally feel its way along the bottom of the body of water. In combination with mapping software, laser scanners, and cameras enable the device to reconstruct high-precision 3D models of the surroundings above water. A multi-beam sonar integrated into the sensor system is used for underwater mapping and creating a complete 3D model of the bed.
“Our navigation system is semi-automatic in that the user only needs to specify the area to be mapped. The surveying process itself is fully automatic, and data evaluation is carried out with just a few clicks of the mouse. We developed the software modules required for the mapping and autonomous piloting,” explains Dr. Janko Petereit, a scientist at Fraunhofer IOSB.
Continue reading “Autonomous watercraft for underwater and surface mapping of rivers, lakes” »
In a revolutionary scientific endeavor, researchers are using 5,000 miniature robots perched atop a mountaintop telescope to peer an astonishing 11 billion years into the past. This cutting-edge instrument, known as the Dark Energy Spectroscopic Instrument (DESI), is capturing light from distant objects in space, allowing scientists from the Lawrence Berkeley National Laboratory to map our cosmos as it was in its infancy and trace its evolution to the present day.
Why is this so important? Understanding how our universe has evolved is intrinsically linked to predicting its ultimate fate and unraveling one of the biggest mysteries in physics: dark energy. This enigmatic force is causing our universe to expand at an ever-increasing rate, and DESI is providing us with unprecedented insights into its effects over the past 11 billion years.
Continue reading “5,000 Tiny Robots Unveil Secrets Of Universe’s Dark Energy” »
Machine learning techniques may appear ill-suited for application in fields that prioritize rigor and deep understanding; however, they have recently found unexpected uses in theoretical physics and pure mathematics. In this Perspective, Gukov, Halverson and Ruehle have discussed rigorous applications of machine learning to theoretical physics and pure mathematics.
Here’s why he believes the landmark law he helped to shepherd through will change the AI sector for the better.
CEO Sam Altman famously has no equity in OpenAI, but startup bets like Reddit, Stripe and Helion have made him a billionaire anyway, a Forbes investigation found.
Microsoft has unveiled a new London hub for its recently launched consumer AI division, fronted by an AI scientist poached from Inflection AI.
From mcgill university and facebook CIFAR #AI
LLM2Vec.
Large language models are secretly powerful text encoders.
Continue reading “LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders” »