Voice Assistant in Cosmos, processor for AR chocks and other news from the world of AI

We strive to inform readers not only about events in the Bitcoin industry, but also to talk about what is happening in related technological spheres-cybersecurity in the world of artificial intelligence (AI).

Forklog Ai has collected the most important news from the world of AI over the past week.

• NASA tests and technologies as part of the lunar mission.
• Scientists decided the centuries -old differential equation, eliminating the computational restrictions of “liquid” artificial intelligence.
• The neural network was taught to decode brain waves and understand what a person sees.
• Microsoft and Nvidia will build a supercomputer.
• Qualcomm introduced a processor for the next generation AR-shafts.
• Robops taught to move around any area.
• The most important transactions from the field of AI in the last week.

ALEXA AI assistant was launched on the moon as part of the NASA mission

The USA space agency NASA has successfully launched the Moon Mission Artemis I, within the framework of which II-technologies will test.

On board the Orion unmanned ship, a Callisto console on the basis of the iPad with the installed assistant Alexa and the client of the Webex video name platform was placed on the board.

The mission will last 25 days 11 hours 36 minutes. During this time, the Orion ship should overcome about 2.1 million km.

Scientists are approaching the opening of fast and effective neural networks

MIT researchers decided the differential equation of centuries -old prescription to eliminate computing limitations of “liquid” artificial intelligence.

In 2021, scientists developed AI-Algorithm, capable of learning and adapt to new information during work, and not just at the initial stage of training. These “liquid” neural networks are suitable for use in urgent tasks like monitoring a pacemaker, weather forecasting or autonomous navigation of Robokar.

However, such models are expensive in computation as the number of neurons and synapses increases. Algorithms also require various computer programs to perform complex mathematical operations underlying them.

Now MIT said that the solution of the differential equation of 1907, which was behind the interaction of two neurons through the synapses, will open up a new type of fast and effective ALGRITMES.

The neural network was taught to analyze the brain waves and decode vision

A group of researchers created the Mind-Vis AI system, which can decode brain waves and display what a person sees.

https://gagarin.news/news/here-comes-the-new-heist-horizon-bridge-lost-100-million/

The algorithm interprets “mnemonic invariants” formed in the head of people at the sight of objects. They are patterns that are stored in different memories of the same subject.

To decode waves, scientists involved a model of self -controlled learning on the same brain activity of several people. They also added pre-trained diffusion II generator of images in the text and transformer neural network for these imaginary representations.

After a small thin setup for 1,500 pairs “Picture – FMRTs”, the system was able to decode what a person sees.

Mind-Vis is distributed with open source. Libra of models are provided on request.

Microsoft and Nvidia will build a supercomputer

Microsoft and NVIDIA have concluded many years of cooperation to develop a massive AI-Supercompeuter.

Partners will post a system in Azure cloud environment. The machine will be equipped with network technology Quantum-2 400GB/S Infiniband and H100 graphic processors.

As part of cooperation, NVIDIA plans to use copies of Azure virtual machines to study achievements in the field of generative AI and self -controlled algorithms capable of creating text, code, image, video or audio.

Microsoft intends to optimize the DeepSpeed ​​deep learning library for a partner equipment. The company seeks to reduce the expendable capacity and memory during the workloads of training AI.

Qualcomm introduced a processor for the next generation AR-shafts

Qualcomm announced the Snapdragon AR2 Gen 1 platform for “thin and light” augmented reality glasses.

The system is a multi-chip design performed according to the 4-nm process. Snapdragon AR2 Gen 1 supports up to nine cameras simultaneously to understand the user and the environment using the “Visual Analytics Engine”, which is entrusted with detecting objects.

The accompaniment, located in another place of glasses, uses AI accelerator to track the gaze and fulfill other tasks of computer vision. The third module provides a connection to networks like WiFi 7, smartphones and PC.

The company said that the multi -chip design of the system provides 2.5 times higher AI with half the energy consumption compared to the reference solution on the basis of XR2.

Earlier, Qualcomm introduced the flagship processor Snapdragon 8 Gen 2. The novelty was equipped with a dedicated AI-Chip, which is 4.35 faster than its predecessor.

Waymo launched robotax in the center of Phoenix

Waymo will allow everyone to test robotaxi without a driver in Phoenix (USA).

Jaguar I-Pace drones will be able to transport customers who loaded the company’s application in the established area of ​​service around the clock and seven days a week.

There is no information about the number of robotaxi involved.

AI-helper appeared in notion to write content

Notion launched testing of AI assistant to write notes and other content.

The company said that the platform users may allow or prohibit their work in the training dataset Dreamup. The authors also have the opportunity to abandon the use of their paintings for training third-party and models.

Deviantart called on other platforms to accept this approach so that artists could share images with the audience, maintaining control over the “inhuman use”.

Robops taught to move around any area

Researchers from the University of Carnegie – Mellon and the University of California at Berkeley trained Robops to move in almost any location.

The system consists of 16 CS-2 servers built on a large-sized WAFER-Scale Engine 2, with 13.5 million nuclei. For comparison: the largest supercomputer in the world Frontier has 8.7 million nuclei. The cluster is controlled by 284 third -generation GEN AMD EPYC processors.

According to the company, the cost of creating Andromeda amounted to $ 35 million, unlike $ 600 million spent on Frontier development.

The most important transactions from the field of AI in the last week

From November 13 to 19, 2022, startups in the field of artificial intelligence attracted over $ 431.7 million. We give the most interesting transactions.

• Astera Labs received $ 150 million for the development of chips for connecting II systems in the cloud.
• Attabotics attracted $ 71.7 million to expand robotic structures for vertical warehouses.
• The creators of the technology for automation of the processes of debt collection of Symed raised $ 42 million.
• Spot AI attracted $ 40 million to develop a smart video surveillance system.
• Speak received $ 27 million to expand the AI ​​platform to learn languages.
• The creators of soft robotic captures Soft Robotics raised $ 26 million.
• The manufacturer of smart robbers Pickle Robot Company attracted $ 26 million.
• OWL Labs received $ 25 million for scaling AI cameras for meetings.
• Terzo raised $ 16 million to improve the software for processing contracts.
• The creators of built -in tools for teaching the correct pronunciation of Namecoach names have attracted $ 8 million.

Also on FORKLOG:

• EU published a new edition of the bill on AI.
• Italy prohibited the use of persons recognition technology.
• Waymo turned robotax into mobile weather stations.
• Tesla autopilot was accused of a death accident.
• Intel introduced the DIPFEIC detector.
• Boston Dynamics sued the manufacturer of “killer robots”.
• Microsoft and Lockheed will help the Pentagon introduce advanced technologies.

Subscribe to FORKLOG news at Telegram: Forklog AI – all news from the world of AI!