Category: Artificial intelligence

  • Yandex Enters Indonesia’s AI Market: Opportunities and Impact

    Yandex Enters Indonesia’s AI Market: Opportunities and Impact

    Founded in 1997, Yandex started its search engine the very same year. Nowadays, Yandex Search leads the search engine industry in Russia, holding a market share exceeding 70 percent. Beyond just search, Yandex offers various services such as online shopping, ridesharing, streaming media, cloud computing, and web mapping, and is also creating a web browser. This giant, which employs over 25,000 people, is now focusing on the AI market in Indonesia.

    Meeting About Expansion

    According to Reuters, Indonesia’s communications minister, Meutya Hafid, had a meeting with Alexander Popovskiy, who heads Yandex’s international search division. They discussed Yandex’s plans to "expand the search engine platform in Indonesia.” However, neither party gave any comments regarding the specifics of the investment, including its financial aspects and timeline.

    AI Investments in Indonesia

    Earlier this year, Nvidia along with the Indonesian telecom company PT Indosat Ooredoo Hutchinson revealed intentions to establish an AI center in Central Java. The investment amount for this venture has been disclosed as $200 million.

    Yandex rolled out the MatrixNet machine learning algorithm back in 2009 and has integrated it into many of its products ever since. In fact, CERN utilizes this technology to process the enormous data generated by the Large Hadron Collider.

  • Meta Uses AI to Predict Ages and Identify Teens on Instagram

    Meta Uses AI to Predict Ages and Identify Teens on Instagram

    A few months back, Meta said it would start looking for teenagers pretending to be adults in early 2025. This announcement came after the company launched Teen Accounts on Instagram in September.

    AI Tool for Identification

    According to a report by Bloomberg, Meta plans to implement an AI tool called an "adult classifier" to spot teens on Instagram. This tool will analyze users’ activity and their profile details to find potential teen accounts.

    Allison Hartnett, who is the director of product management for youth and social impact at Meta, shared with Bloomberg that the tool will categorize users into two groups – either over or under 18 years. The technology can review a person’s profile, their follower list, and the type of content they interact with to determine their age.

    Automatic Migration to Teen Accounts

    Users who are under 18 will automatically switch to teen accounts, no matter what age they list on their profile. Those who are 16 or 17 will have the ability to manage their privacy settings; however, users younger than 16 will need permission from a parent to make any changes.

    Meta is also making it more difficult for users to misrepresent their age on the platform. It will flag any suspicious actions, such as creating accounts with different birth dates but using the same email. Additionally, Meta can access a phone’s unique ID to understand who is behind the profile.

    Age Verification Process

    For teens wanting to alter their age on the platform, they will need to provide official identification. Options include a driver’s license or a video selfie that will be evaluated by a third-party service called Yoti. Meta assures users that their images will be deleted from their servers after the verification process.

    In conclusion, Instagram’s new measures aim to provide a safer environment for its younger users. By implementing these changes, Meta hopes to ensure that the platform is used responsibly.

  • Meta’s Llama Now Available for U.S. National Security Agencies

    Meta’s Llama Now Available for U.S. National Security Agencies

    Meta has decided to offer its open-source LLM (Large Language Model) Llama to U.S. government agencies and various contractors involved in national security initiatives.

    Partnerships for National Security

    The firm revealed its collaboration with a range of companies including Accenture Federal Services, Amazon Web Services, Anduril, Booz Allen, Databricks, Deloitte, IBM, Leidos, Lockheed Martin, Microsoft, Oracle, Palantir, Scale AI, and Snowflake to facilitate this effort.

    Oracle plans to utilize Llama for integrating aircraft maintenance documentation to identify possible issues. Meanwhile, Scale AI is crafting a specialized model based on the LLM to assist missions of national security teams. Lockheed Martin aims to employ Llama to enhance coding processes and data analysis, thereby improving their operational efficiency.

    Hosting and Applications

    Amazon Web Services (AWS) and Microsoft Azure will provide hosting for Llama to be used in various government applications. Additionally, IBM intends to incorporate Llama within its watsnox solution designed for national security agencies. Meta asserts that these advancements will promote “the prosperity and security of the United States.” They are also positioned to help set U.S. open-source benchmarks in the global competition for AI dominance.

    The organization emphasizes that “it is beneficial for both America and the broader democratic community for American open-source models to thrive and outshine those from China and other regions.” Recently, the news outlet Reuters reported that China had developed a tailored AI model based on an older version of the Llama 13B LLM, aimed at potential military uses.

    Meta’s strategy underscores its commitment to maintaining leadership in the critical field of AI technology.

  • Amazon Prime Video Unveils AI-Powered Recaps in X-Ray Update

    Amazon Prime Video Unveils AI-Powered Recaps in X-Ray Update

    Amazon has introduced a new update to the X-Ray feature on Prime Video that incorporates generative AI to provide recaps of series or episodes. According to Amazon, X-Ray Recap can produce "easy-to-read summaries" for the shows you watch.

    Personalized Summaries

    The feature can create summaries for "entire seasons of TV shows, individual episodes, and even segments of episodes, tailored to the exact moment you’re viewing." This personalization enhances the viewing experience by allowing users to catch up quickly.

    Key Story Moments

    X-Ray Recap presents brief text snippets highlighting significant story events, characters, and other pertinent details. Amazon claims that it also understands how much content you have already consumed, ensuring that the summaries remain spoiler-free.

    Advanced Technology

    Recap uses Amazon’s Bedrock AWS service alongside specialized AI models developed through Amazon SageMaker (AWS AI training tools) to analyze video clips and subtitles/dialogue to generate descriptions of key plot points, characters, and events.

    Users can access Recaps from the Prime Video detail page or select it from the X-Ray option while watching. To use this feature, simply tap on the screen during playback or press the up button on the FireTV remote, then click the "View All" option.

  • ADNOC to Launch Autonomous AI in UAE Energy Sector

    ADNOC to Launch Autonomous AI in UAE Energy Sector

    UAE’s state-owned oil company ADNOC (Abu Dhabi National Oil Company) is set to be the first energy firm to use autonomous agentic AI. This initiative is supported by G42, another government-associated AI and cloud computing firm that got a $1.5 billion investment from Microsoft in April.

    Insights from ADNOC’s Leadership

    In an interview with Reuters, ADNOC CEO Sultan Al Jaber mentioned that this strategy aims to "analyze petabytes of data" and will "proactively and autonomously identify operational improvements." Alongside G42 and Microsoft, the project also has backing from AIQ, a company created by ADNOC and G42 in 2020.

    What is Agentic AI?

    Agentic AI refers to an autonomous system that leverages the reasoning skills of LLMs (Large Language Models) to carry out tasks with minimal human involvement. These systems have the ability to evaluate vast data sets and create plans to achieve specific objectives.

    Financial Impact and Environmental Benefits

    In March, ADNOC reported that its use of AI tools generated five hundred million dollars in value throughout 2023, according to JPT75. The company stated it utilized over 30 AI programs to help decrease carbon emissions by one million metric tons from 2022 to 2023.

  • Copilot+ PCs Offer 5x Performance Boost Over 5-Year-Old Models

    Copilot+ PCs Offer 5x Performance Boost Over 5-Year-Old Models

    Microsoft’s Copilot+ PCs are introducing a range of AI features, including Recall, Cocreator, and others. Beyond these intelligent functionalities, they are also said to enhance performance and battery life. According to Microsoft, these Copilot+ PCs represent the fastest and most advanced Windows computers ever made. But how do they compare to older Windows PCs? Microsoft has shared details to clarify this.

    Performance Comparison

    A recent report from Windows Latest indicates that Microsoft conducted a comparison between the Copilot+ PCs and well-known PCs that were popular five years ago. The focus of this assessment was mainly on performance and battery endurance. The Copilot+ PCs utilized in the evaluations are equipped with Snapdragon X Elite 12 core and Snapdragon X Plus 10 core configurations.

    Benchmark Results

    To measure performance, Microsoft used the widely recognized benchmark tool, Cinebench 2024 Multi-Core. The findings reveal that the Snapdragon-powered Copilot+ PCs outperform the most popular Windows PC from 4-5 years ago by 5 times.

    Battery Life Enhancements

    In terms of battery life, the Copilot+ PCs show significant improvements over older Windows models. During local video playback tests, these new devices demonstrated up to 2.3 times longer battery life. All devices during these tests operated on the same Wi-Fi network, with display brightness set to 150 nits.

    When it comes to web browsing, the Copilot+ PCs manage to deliver 3 times more battery life than their predecessors. These findings indicate that while Copilot+ PCs emphasize AI advancements, they also provide enhancements in other important areas. Regarding AI capabilities, Microsoft has yet to introduce features like Recall and Click to Do, but the company has pledged to roll them out by early next year.

  • Microsoft Q1 2024 Revenue Up 16% Driven by Cloud Growth

    Microsoft Q1 2024 Revenue Up 16% Driven by Cloud Growth

    Microsoft Corp. has shared its financial results for the first quarter of 2024, showcasing the company’s growing focus on artificial intelligence (AI) services. For the quarter ending September 30, 2024, Microsoft announced a revenue of $65.6 billion, representing a significant 16% rise compared to the same timeframe last year.

    Financial Performance

    The tech leader’s operating income hit $30.6 billion, showing a 14% increase, while net income climbed to $24.7 billion, which is an 11% growth. A major factor in this success was Microsoft Cloud, which experienced a revenue boost of 22%, bringing in $38.9 billion for the quarter. In particular, the Intelligent Cloud segment contributed $24.1 billion, reflecting a 20% increase.

    CEO’s Remarks

    In a statement to the press, CEO Satya Nadella noted, “AI-driven transformation is changing work, work artifacts, and workflow across every role, function, and business process.” As Microsoft embraces these changes, it’s also aware of the risks that come with AI. The forward-looking comments in the earnings report pointed out possible hurdles, such as competition, cybersecurity dangers, and increased regulatory oversight.

    Other Business Segments

    The report also sheds light on the More Personal Computing segment, which recorded revenues of $13.2 billion, an increase of 17%. Notably, revenue from Xbox content and services skyrocketed by 61%, largely due to the recent Activision acquisition.

    These results reflect trends seen in other tech companies like Google, which noted a 35% rise in cloud revenue.

  • Light-Based Memory Cells: 1000x More Durable Data Storage

    Light-Based Memory Cells: 1000x More Durable Data Storage

    A breakthrough in memory technology has been discovered, featuring a new type of memory cell capable of storing data and executing rapid calculations. This exciting development, reported in Nature Photonics, holds the promise of advancing AI performance while significantly reducing energy consumption in data centers.

    How It Works

    The innovative memory cell utilizes magnetic fields to guide light signals through a circular resonator. This component enhances specific wavelengths of light, directing them either clockwise or counterclockwise to designated output ports. The intensity of the light at each port represents values such as zero and one, or even zero and negative one.

    Advantages Over Traditional Memory

    Unlike conventional memory cells that only store basic binary values, this cutting-edge design is able to capture multiple non-integer values, achieving storage of up to 3.5 bits per cell. Imagine two athletes running on a track in opposite directions; their speeds can vary based on external conditions, and this difference is used to represent positive or negative numbers.

    These encoded values are crucial for artificial neural networks, as they help modify the strength of connections between nodes. This capability is especially beneficial for image recognition tasks, where neural networks analyze visual data in ways that mimic human brain functions.

    Efficiency and Durability

    One of the major advantages of this new technology is its remarkable efficiency. Conventional computers separate storage from calculations, processing data in the CPU and then writing results to memory. In contrast, this new memory cell can perform high-speed calculations directly within the memory, greatly enhancing AI applications that require rapid data handling.

    The research team subjected their design to extensive testing, demonstrating its ability to endure over two billion write-and-erase cycles without degrading performance. This resilience is a thousand times superior to earlier photonic memory technologies. For comparison, standard flash drives typically endure between 10,000 and 100,000 cycles.

    Looking to the future, the team aims to integrate additional cells into computer chips and investigate even more intricate computations. This technology has the potential to lower the energy demands of AI systems, marking a significant move towards more efficient computing solutions.

  • Meta’s Robotic Touch Sensors Detect Forces Smaller Than a Paperclip

    Meta’s Robotic Touch Sensors Detect Forces Smaller Than a Paperclip

    Meta’s AI Research team, known as FAIR, has unveiled several new technologies aimed at enhancing touch perception and dexterity in robotics. They have introduced three key innovations: Meta Sparsh, Meta Digit 360, and Meta Digit Plexus.

    Meta Sparsh: A Game Changer in Touch Sensing

    Meta Sparsh is groundbreaking as the first versatile system for touch sensing that utilizes vision—it operates seamlessly across a variety of sensors and tasks. This technology uses AI models to process touch signals, eliminating the need for manual models that typically require extensive customization for different tasks.

    The Innovative Meta Digit 360

    Next up is the Meta Digit 360, a project developed in partnership with GelSight Inc. This advanced artificial fingertip features more than 18 sensing capabilities, allowing it to detect forces as small as one millinewton. With over eight million tiny sensing units, known as "taxels," it can sense touch from any angle.

    Bringing It All Together with Meta Digit Plexus

    Furthermore, Meta Digit Plexus serves as a unified platform that integrates various touch sensors on robotic hands. It is set to be featured in the Allegro Hand from Wonik Robotics, which is anticipated to launch next year.

    In addition, Meta has rolled out a new benchmark framework called PARTNR, designed to evaluate how effectively robots can plan and work alongside humans. This framework includes 100,000 natural language tasks distributed across 60 virtual homes and over 5,800 distinct objects, all aimed at enhancing the social capabilities of robots.

    These developments are expected to create significant impacts in industries such as healthcare, manufacturing, and prosthetics. The Digit 360 from GelSight Inc. is anticipated to be available next year, with an early-access option through a call for proposals.

  • Disney Launches AI and Augmented Reality Innovation Team

    Disney Launches AI and Augmented Reality Innovation Team

    Founded more than a hundred years ago as the Disney Brothers Cartoon Studio, The Walt Disney Company has evolved into a global powerhouse in the entertainment industry. Employing around 225,000 individuals, Disney reported revenues close to $90 billion and a net income exceeding $2.35 billion in 2023. Recently, the company is taking strides forward by creating a division dedicated to new technologies like artificial intelligence and mixed reality.

    New Leadership

    As per a special report from Reuters, “the media giant explores applications across its film, television, and theme park divisions,” thus it’s not surprising that Jamie Voris will head this new division. Voris previously served as Disney’s Chief Technology Officer, overseeing the app development for the Apple Vision Pro mixed reality headset. This newly formed group is called the “Office of Technology Enablement” and, although it starts with a small leadership team, it is expected to grow to around 100 employees fairly soon, according to a source close to Disney.

    Focus on Emerging Technologies

    Different divisions within Disney have been dabbling in AI and augmented reality for some time, but this new team will concentrate solely on these technologies. One promising area for advancement is enhancing augmented reality experiences in theme parks and even in customers’ homes, thanks to the latest lightweight glasses recently launched by Meta and Snap.

    Market Insights

    In 2023, the AR/VR headset market sold 1.7 million units, with Meta accounting for over 60% of those sales. Competitors include big names like Sony, Apple, and ByteDance, which is the parent company of TikTok.

    The Meta Quest 3S 128 GB, bundled with Batman: Arkham Shadow and a three-month trial of Meta Quest+, is currently Amazon’s top-selling item in the Standalone Virtual Reality Headsets category, moving more than 10,000 units in just the past month. It is priced at $299, which includes free shipping and returns.