We strive to inform readers not only about events in the bitcoin industry, but also to talk about what is happening in related technological areas – cybersecurity and in the world of artificial intelligence (AI).
Cryplogger AI collected the most important news from the world of AI over the past week.
- Media: during the year, Yandex will run out of semiconductors needed for servers.
- Chinese researchers have drawn up a roadmap for the development of AI in the “Era of Big Models”.
- Elon Musk thought about creating his own “free” social network.
- Nvidia has introduced an algorithm for turning 2D images into three-dimensional images.
- Boston Dynamics has released the Stretch loader robot for free sale.
- South Korea will test hydrogen-powered self-driving buses in urban areas.
- The most important AI deals of the last week.
Media: “Yandex” will exhaust the supply of semiconductors within a year
Due to US sanctions “Yandex” may exhaust stocks of semiconductors and other components necessary for the operation of servers during the year.
According to Bloomberg, the tech giant is facing a growing shortage of equipment due to US-imposed import restrictions in connection with Russia’s military intervention in Ukraine. Sanctions on dual-use technologies used in the military and civilian sectors have caused the most significant damage to the drone development unit, the newspaper writes.
The lack of server hardware may also affect other areas of Yandex’s activities, including search services, cloud business, music streaming and e-commerce segment.
In addition, the tech giant is facing a brain drain as Russian employees seek to move abroad as the country’s economic isolation deepens. Since the beginning of the war in Ukraine, up to 70,000 IT specialists have left Russia, and in April this figure may increase to 170,000 people.
Chinese researchers have developed a roadmap for the development of AI
100 Chinese explorers developed roadmap for the development of AI in the “Era of Big Models”. It consists of 200 pages and includes a bibliography of 1637 references.
The roadmap identifies three major challenges for modern AI systems that need to be addressed:
- Model intolerance. Since the algorithm is built and trained for a specific task, it cannot be used to solve another problem, which means that you have to start training the neural network again.
- Lots of handmade. Model training, including parameter tuning, requires the participation of a large number of AI engineers.
- Lack of quality data. This may limit the applications of the algorithms in some areas.
The document also considers the elimination of the “main risk of such algorithms” – all their defects extend to subsequent neural networks and can subsequently be transferred to the entire community of basic models.
Elon Musk wants to create his own social network
American businessman Elon Musk “seriously thought” about creating a social network with open source AI algorithms and respect for freedom of speech.
Am giving serious thought to this
— Elon Musk (@elonmusk) March 27, 2022
Earlier, the billionaire asked users if the Twitter algorithm should be open source. Over 82% of the nearly 1.2 million participants responded positively.
Twitter algorithm should be open source
— Elon Musk (@elonmusk) March 24, 2022
Also Musk inquired from their subscribers, whether they believe that the social network adheres to the principles of free speech. More than 70% of users out of about 2 million who participated in the survey answered negatively.
Yandex fired American drone testers
Company “Yandex” fired more than 20 US employees involved in drone testing.
The company said that in early March 2022, the Michigan Department of Transportation suspended the tech giant’s license to operate autonomous vehicles. According to The Verge, the state authorities denied this information.
Yandex confirmed that the management fired six drivers who “secured” unmanned vehicles in Michigan, and 21 employees from the team involved in robot deliveries in Arizona and Ohio.
Nvidia introduced an algorithm for turning 2D images into three-dimensional images
Nvidia company demonstrated the work of the Instant NeRF neural network, which turns a couple of dozen 2D photos into a three-dimensional 3D image.
The NeRF method was developed in 2020 by American researchers in collaboration with Google Research. It matches the colors and light intensity of photographs taken from different angles, and then generates data to connect the shots and render the finished 3D scene. In addition to images, the system needs information about camera angles.
The more objects you need to transfer to 3D, the more 2D photos you need to take, otherwise the end result will be blurry.
According to Nvidia, Instant NeRF is the fastest solution among peers. The model is trained in a few seconds compared to other neural networks, which require several hours.
The company claims that the presented technology can be used to train robots and drones so that they more effectively understand the sizes and shapes of objects in the real world.
Boston Dynamics launches Stretch robot loader
Boston Dynamics will start sale of Stretch loader robots for everyone.
The device is a transportable platform with a manipulator having seven degrees of freedom and equipped with suction cups. The robot is equipped with cameras and sensors to navigate in space, and is designed to move boxes around the warehouse and unload trucks.
According to Boston Dynamics, Stretch is sold out through 2022 due to strong customer demand. However, the company is now accepting orders for the supply of a new batch, scheduled for 2023 and 2024.
Study: Facebook Ad Algorithms Don’t Consider the Context of User Interest Profiles
The researchers found that the models create profiles of interests based on user actions, but do not take into account their context. According to the researchers, if a person posts a message about “how much they dislike cheese,” the AI will only detect interest in cheese and run targeted ads accordingly.
To learn more about the process of creating interest profiles on Facebook, the researchers conducted two experiments. In the first test, they registered 14 new accounts, where they controlled demographic data and tracked a generated list of preferences. The researchers found that 33.22% of the suggested interests were inaccurate or irrelevant.
For the second experiment, the researchers recruited 146 participants from around the world. They collected data from each account and analyzed the AI-generated preference list. As a result, 29.3% of interests did not correspond to the real preferences of users.
Also, the majority of respondents did not even know about the existence of the Facebook advertising preferences manager, the scientists said.
AI helped Google Maps remove over 100 million map edits
In 2021 Google Maps with Machine Learning blocked over 100 million offensive map edits.
According to the company, human operators, along with AI, also removed about 7 million fake profiles, of which only 630,000 were based on user posts. In addition, AI prevented 12 million attempts by fraudsters to impersonate a company and 8 million cases of false claims claiming to be a business account.
In addition, machine learning helped the Google Maps team remove nearly 200 million photos and videos that were of poor quality or violated the rules.
South Korea to start operating hydrogen-powered self-driving buses
April 2022 in South Korea will launch hydrogen bus with the third level of autonomy.
The vehicle was developed by Hyundai Motor and Hyundai Mobis by order and with the support of the government of the country, having allocated $ 17.7 million over two years.
From 2017 to 2021, Korean authorities have invested $119 million to build nine key components for active driver assistance systems, according to Business Korea. This led to the registration of 120 patents related to cameras, radars and lidars.
Until 2027, the South Korean government plans to release systems of the fourth level of autonomy.
Scientists use AI to improve weather forecast
Scientists tied up smartphone data with satellites to improve weather forecast.
Camaliot, funded by the European Space Agency, developed An Android application that, through GPS receivers, collects information with the time and location of satellites. The system notes the time spent on receiving messages and uses this data to determine the location of the smartphone.
The researchers hypothesized that this way they could extract more information about the atmosphere, including how the amount of water vapor affects the propagation of satellite signals through the air.
The most important AI deals of the last week
From March 27 to April 2, 2022, AI startups raised over $253 million. Here are the most interesting transactions.
- Augmented reality glasses manufacturer Nreal collected $60 million
- Yami attracted $50 million to expand the online trading platform for Asian goods and food products.
- Pinecone received $28 million for the development of a specialized database for data scientists.
- Kaiyo attracted $36 million to support an online used furniture marketplace.
- Accounts receivable automation startup Churpy collected $1 million
- sourceful received $20 million to develop a platform for supply chain transparency.
- Zepto attracted $18.8 million to scale the instant payment platform.
- Developers of software for automatic planning of business processes Skan collected $40 million
Also on Cryplogger:
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER