Category: Artificial intelligence

  • Baidu AI Cloud Unveils 5 New ERNIE Models for Businesses

    Baidu AI Cloud Unveils 5 New ERNIE Models for Businesses

    Baidu AI Cloud is making strides in the AI industry through a significant upgrade to its Qianfan platform, aimed at enabling businesses of all sizes to harness advanced AI technology more easily and affordably than ever before.

    Enhanced ERNIE Models Offer Cost-Effective Solutions

    The focal point of the update is the introduction of five new ERNIE models, which stand out for their smaller footprints compared to traditional large language models (LLMs). These innovative models provide businesses with a cost advantage, eliminating the need for substantial computing power investments. The versatility of the new ERNIE models ensures they excel across various applications, catering to the specific needs of different businesses.

    ![Image Credits: Reuters](Image link here)

    Diverse Lineup of ERNIE Models on Qianfan ModelBuilder Platform

    The Qianfan ModelBuilder platform now features a diverse range of ERNIE models tailored to different requirements:

    • ERNIE Speed: Known for its context comprehension capabilities, ERNIE Speed delivers highly accurate and coherent content, handling complex scenarios effortlessly. It can analyze context windows of up to 128,000 tokens and competes favorably with top-tier ERNIE models in specific domains.
    • ERNIE Lite: An upgraded version of ERNIE Bot Turbo, ERNIE Lite enhances performance by 20% for tasks like sentiment analysis and natural reasoning. Notably, it also reduces inference costs by 53%, making it ideal for deployment on low-powered AI accelerator cards.
    • ERNIE Tiny: Despite its compact size, ERNIE Tiny excels in applications prioritizing low latency and cost-effectiveness, such as retrieval and recommendation tasks. In conversational recommendation tests, ERNIE Tiny outperformed ERNIE 3.5 by enhancing dialogue interactions by 3.5% and cutting costs by 32%.

    Specialized ERNIE Models and Improved Qianfan AppBuilder Toolkit

    In addition to general-purpose models, Baidu AI Cloud has introduced two specialty models:

    • ERNIE Character: Tailored for roleplaying scenarios, ERNIE Character breathes life into non-playable characters in video games and intelligent chatbots for customer service.
    • ERNIE Functions: Catering to businesses requiring seamless integration of external tools or internal functionalities, ERNIE Functions simplifies the process by providing pre-trained models for developing intelligent assistants.

    The Qianfan AppBuilder toolkit has been enhanced with 55 new tools aimed at simplifying the development of AI-native applications. Offering a wide array of features, from core AI capabilities to advanced functionalities like Retrieval Augmented Generation (RAG) and Generative Business Intelligence (GBI), the toolkit streamlines the application development process.

    Qianfan has already empowered 80,000 businesses, witnessing a remarkable 97% surge in daily usage quarter-over-quarter since December 2023, solidifying its standing as a leading AI model development platform.

    Tags: , ,
  • Nvidia’s AI Empire Hit Hard by Samsung’s Mach-1

    Nvidia’s AI Empire Hit Hard by Samsung’s Mach-1

    Artificial intelligence technologies are expanding their presence across various products and services, with smartphones emerging as the latest frontier for AI integration. While AI features have predominantly been found in high-end devices due to their demanding performance requirements, the idea of incorporating expensive and powerful processors in every phone for AI tasks is currently not feasible. To tackle this issue, a leading Korean tech company has devised a solution in the form of a dedicated AI accelerator chip, aiming to make AI capabilities more accessible.

    Samsung Introduces Mach-1 for AI Domination

    Samsung recently unveiled its Galaxy AI in conjunction with the Galaxy S24 series, showcasing advanced features such as a ChatGPT-like language model and a range of artificial intelligence functions. However, the high performance demands of AI have restricted this feature to select models, even causing powerful processors like the Snapdragon 8 Gen 3 to potentially struggle with upcoming AI applications. In response, Samsung is forging ahead with the development of its own dedicated AI accelerator chip known as Mach-1.

    The Future of AI Integration

    Reports suggest that Samsung is setting its sights on a 2025 release for the Mach-1 chip, marking a significant move towards enhancing AI integration within smartphones. While initial models may not immediately offer the necessary power efficiency for seamless smartphone incorporation, the prospect of a dedicated AI chip being integrated into the Galaxy S25 cannot be ruled out.

    Competition and Market Dynamics

    This development signifies an important initial step, with future iterations potentially leading to substantial advancements. Samsung not only aims to leverage the chip internally but also intends to market it to other companies, positioning itself as a direct competitor to Nvidia. Given Nvidia's notable success in the AI chip realm over the past few years, Samsung, a key player in the tech industry, is understandably seeking to enter this lucrative market to foster a competitive landscape.

    The dominance of Nvidia in AI technology is evident through collaborations with major companies like Microsoft, Meta, OpenAI, Amazon, and Google, resulting in significant revenue generation. While Samsung has been working on similar processors, the actual launch of a concrete product is yet to materialize. The absence of substantial competition could potentially create an imbalanced market, making Samsung's foray into this sector crucial for ensuring a healthy competitive environment.

    Tags: , ,
  • Google AI Models Fighting High-Mortality Diseases in India: TB, Breast Cancer & More

    Google AI Models Fighting High-Mortality Diseases in India: TB, Breast Cancer & More

    Google AI has partnered with Apollo Radiology International in an effort to introduce AI-driven healthcare solutions to various communities across India. The primary focus of this collaboration is to address prevalent high-mortality diseases such as tuberculosis, lung cancer, and breast cancer, where the scarcity of specialists poses a challenge to early detection.

    Leveraging AI for Healthcare Solutions

    Apollo Radiology in India has begun utilizing AI models to conduct 3 million complimentary screenings for TB, lung cancer, and breast cancer. In the case of TB detection, a significant obstacle arises from the lack of radiologists available to interpret chest X-rays, which are crucial for screening purposes. Google’s AI system plays a vital role in the early identification of tuberculosis by scrutinizing these scans. Moreover, AI aids in enhancing the accessibility of lung and breast cancer screenings while also assisting in identifying incidental nodules in lung scans and mammograms, enabling radiologists to prioritize cases effectively.

    Improving Access to Timely Care

    Over the next ten years, Apollo Radiology International aims to leverage these AI models to conduct 3 million screenings for TB, lung cancer, and breast cancer at no cost. This ambitious initiative has the potential to significantly enhance access to timely healthcare for hundreds of thousands of individuals in India.

    Bridging Healthcare Gaps with AI

    Google.org is also extending its support to ARMMAN, a mobile service that delivers tailored preventive care messages to expecting and new mothers. Through the implementation of Google AI technology, health workers can predict women who are at risk of discontinuing healthcare programs, enabling them to take proactive measures in following up with these individuals.

    This partnership exemplifies how AI is closing the healthcare disparity gap in regions with limited resources, facilitating earlier diagnoses and ultimately leading to improved health outcomes. If the initial trials of this technology prove successful, there is potential for its widespread deployment in numerous regions, signaling a bright future where AI plays a pivotal role in transforming healthcare accessibility and efficiency.

    Google AI Models Fighting High-Mortality Diseases in India: TB, Breast Cancer & More
    Tags: , ,
  • Apple’s GenAI Plans: Gemini AI Possibly Arriving on iPhones

    Apple’s GenAI Plans: Gemini AI Possibly Arriving on iPhones

    Generative AI has made significant advancements over the past couple of years, with its integration into consumer electronics such as smartphones and laptops becoming more prevalent. Tech giants like Google, Microsoft, and Samsung have embraced this technology trend, positioning themselves at the forefront of innovation. In contrast, Apple seems to be falling behind in this race, prompting the Cupertino-based company to reevaluate its AI strategies.

    Apple's Potential Collaboration with Google for AI Integration

    Recent reports from Bloomberg suggest that Apple is actively engaging in discussions with Google to incorporate its Gemini AI engines into iPhones. Furthermore, Apple has also been exploring collaborations with OpenAI's ChatGPT. This move indicates that Apple's internal AI development may not be on par with that of its competitors, such as Google. Internal tests on an Apple chatbot named Apple GPT and a large language model known as Ajax have hinted at the company's AI technology lagging behind its rivals.

    Strategic Shift in Apple's Approach to AI

    Apple's CEO, Tim Cook, had previously emphasized the company's interest in the GenAI sector during an earnings call. However, faced with the realization of their lagging position, Apple seems inclined towards partnering with external entities to leverage their AI models on iPhones. The recent acquisition of DarwinAI by Apple further underscores this strategic shift in their AI development approach.

    Implications of a Potential Apple-Google Partnership

    For Google, collaborating with Apple on integrating Gemini AI into iPhones could be a significant breakthrough, especially amidst recent controversies. While the specifics of branding and implementation of the AI model in Apple's devices remain uncertain, users can anticipate functionalities such as image generation, essay writing, and other tasks facilitated through an external provider or cloud service. Such a partnership would not only enhance the reach of Gemini AI but also leverage Apple's massive user base of over two billion iPhone users worldwide.

    Should the speculated deal between Apple and Google materialize, it has the potential to disrupt the AI industry landscape. Bloomberg suggests that any official announcement regarding this partnership is unlikely before the annual Worldwide Developers Conference (WWDC) scheduled for June. This strategic collaboration could mark a significant turning point in the evolution of AI technologies within consumer electronics.

    Tags: ,
  • Samsung Unveils 2024 Micro LED TV Range: Prices from 650K to 1.25M Yuan

    Samsung Unveils 2024 Micro LED TV Range: Prices from 650K to 1.25M Yuan

    Yesterday, Samsung unveiled its latest range of TVs – Micro LED, Neo QLED, and OLED – during a product launch event in China. This launch, much like others by Samsung, centered on AI technologies. The 2024 Micro LED lineup offers TVs in four sizes: 76/89/101/114 inches, with this year's new additions being the 76-inch and 114-inch variants.

    Micro LED Panels Enhanced with AI Features

    The TVs in this year's lineup boast Micro LED panels constructed from sapphire and inorganic self-luminous pixels. They incorporate "Micro contrast technology" to enhance sharpness by eliminating blurry boundaries, along with "Micro color" providing precision control with 65,536 levels.

    Advanced Audio and Video Capabilities

    These TVs feature OTS Pro audio and video tracking technology, supporting a built-in 6.2.2 channel setup with Dolby Atmos for a 3D full surround sound experience. The inclusion of a "Q Symphony" function enables synchronized sound between the TV and sound bars, powered by the "NQ8 AI Gen 3" processor for upscaling and AI enhancements.

    Samsung's Focus on AI Integration

    Samsung is looking to usher in an "AI Screen Era" with these innovations, offering features such as 8K AI Upscaling Pro and AI Motion Enhancer for an improved user experience. During the launch, the company emphasized creating a more intuitive and convenient device experience.

    Choi Seung-sik, President of Samsung Electronics Greater China, highlighted the company's commitment to integrating AI across various connected technologies. He stated, "Samsung is integrating AI into our connected technologies, from Galaxy AI in mobile devices to disruptive display technologies and smart home appliances."

    Pricing:

    • 76 inches: RMB 650,000 (~US$90,299)
    • 89 inches: RMB 750,000 (~US$104,191)
    • 101 inches: RMB 900,000 (~US$125,030)
    • 114 inches: RMB 1.25 million (~US$173,650)
    Tags: , ,
  • Apple’s MM1 Large Language Model Blurs Image-Text Lines

    Apple’s MM1 Large Language Model Blurs Image-Text Lines

    Apple's latest advancement in the realm of artificial intelligence comes in the form of the innovative "MM1" multi-modal large language model. This groundbreaking model, discussed in the paper "MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training," demonstrates remarkable proficiency in tasks involving both image recognition and natural language processing.

    Impressive Model Capabilities

    The MM1 model is offered in three different sizes: 3 billion, 7 billion, and 30 billion parameters. Through extensive experimentation, researchers have identified crucial factors influencing the model's performance. Surprisingly, aspects such as image resolution and the quantity of image tags have a more substantial impact compared to visual language connectors. Moreover, the choice of pre-training datasets significantly influences the efficacy of the model.

    Innovative Architecture and Methodology

    Apple's research team meticulously crafted the MM1 model using a unique "Mixture of Experts" architecture and a "Top-2 Gating" technique. This method not only excelled in pre-training assessments but also translated to impressive performance on established multi-modal benchmarks. Even after refining the model for specific tasks, MM1 continued to exhibit competitive capabilities.

    Competitive Performance and Future Prospects

    Upon testing, it was found that the MM1-3B-Chat and MM1-7B-Chat variants outshine many other models of similar sizes available in the market. These models particularly excel in tasks like VQAv2, TextVQA, and ScienceQA. Despite its strengths, MM1 still falls slightly behind Google's Gemini and OpenAI's GPT-4V models in overall performance. Nevertheless, Apple's MM1 signifies a notable advancement in the field of artificial intelligence, positioning the company as a key player in AI innovation. Apple's recent acquisition of DarwinAI further underscores their commitment to advancing AI technologies.

    Tags: , ,
  • Apple Acquires DarwinAI: Expect AI-Powered Features Ahead

    Apple Acquires DarwinAI: Expect AI-Powered Features Ahead

    Apple recently took a significant step towards enhancing its artificial intelligence (AI) capabilities by acquiring the Canadian startup DarwinAI, as reported by Bloomberg. This move could have significant implications for the tech giant, which boasts a trillion-dollar valuation.

    Founded in 2017, DarwinAI specialized in two crucial areas within the AI sector: model inspection and machine learning techniques for creating streamlined AI systems. Public records reveal that over a span of five years, DarwinAI secured a total of $15.5 million in funding, with major collaborators such as Lockheed Martin and Intel.

    Acquisition and Transition

    Upon the acquisition, DarwinAI's online presence swiftly disappeared, with the deactivation of their website and social media accounts. Numerous former employees of DarwinAI have reportedly shifted to Apple's AI division, including prominent AI researcher Alexander Wong, who has taken on a directorial role.

    Future Plans and Speculation

    While Apple has acknowledged the acquisition, no specific details have been provided regarding the company's intentions or future strategies. This move may signify a broader initiative by Apple to embed AI functionalities throughout its product range, potentially including the forthcoming iOS 18 and macOS 15 operating systems. It is evident that Apple is striving to narrow the gap with competitors like Microsoft and Google, possibly expanding into generative AI and integrating it with core applications such as Siri and Shortcuts. Speculation suggests that Apple is actively exploring large language models and securing new patents for cutting-edge designs, hinting at the development of more sophisticated Apple products in the future.

    Tags: , ,
  • Nvidia to Reveal New AI Chip at GTC Conference

    Nvidia to Reveal New AI Chip at GTC Conference

    Nvidia's Anticipated Showcase at GTC Conference

    Nvidia is gearing up to headline the forthcoming GTC conference, marking the industry's return to in-person events post-pandemic. All eyes are on the tech giant as they gear up to reveal their latest innovations, with a pivotal question hanging over the heads of both customers and investors: can Nvidia maintain its dominance in the AI chip market?

    The B100 Chip: Setting a New Price Benchmark

    The upcoming B100 chip is anticipated to come with a price tag exceeding $20,000, surpassing that of its predecessor. Nvidia's recent surge has been nothing short of remarkable, propelling their market cap to $2 trillion and positioning them closely behind Apple for the number two spot. Analysts foresee a doubling of revenue by 2024, driven by the soaring demand for their top-tier AI chips.

    Future-Proofing Success with the B100

    The focus isn't solely on current achievements but also on fortifying future success. The GTC event is projected to unveil the B100, Nvidia's next-generation AI processor. This formidable chip is poised to serve as the core of their upcoming AI systems, raising the performance bar even higher.

    Challenges and Opportunities on the Horizon

    Already facing a demand that outstrips supply, Nvidia's current chips entail prolonged wait times for developers. The expected price hike for the B100 beyond the $20,000 mark raises concerns among analysts about Nvidia's stock valuation. While the company has experienced a significant uptick in recent years, some analysts worry that overly optimistic future earnings forecasts could potentially impact the company's standing.

    Apart from hardware, Nvidia's software strategy presents another area of interest. Their CUDA platform entices developers with tools tailored for Nvidia chips, fostering developer allegiance. The forthcoming CUDA updates are poised to further solidify this loyalty. Nvidia's venture into cloud services is also a development to monitor closely. The geopolitical dimension, particularly concerning China, introduces an additional layer of complexity. US-imposed restrictions on access to Nvidia's cutting-edge chips have compelled China to develop its own AI hardware. While Chinese competitors offer chips on par with Nvidia's A100, they have yet to match the H100. Analysts anticipate that the B100 will widen this gap even further.

    As the GTC conference approaches, it is evident that this event will be pivotal for Nvidia. Stay updated with the latest Nvidia insights on Gizmochina as the conference unfolds!

    Tags: , ,
  • OpenAI to Release Sora to Public by 2024: Could Be Available in Just Months

    OpenAI to Release Sora to Public by 2024: Could Be Available in Just Months

    OpenAI recently introduced its innovative text-to-video tool, Sora, which has the capability to produce lifelike 1080p videos. At present, the tool is being tested by a select group of filmmakers and creators to identify and rectify any potential weaknesses before making it available to the public. According to OpenAI's chief technology officer Mira Murati in an interview with the Wall Street Journal, the plan is to release Sora to the public before 2025, with the possibility of this happening in just a few months.

    Advancements in Video Generation

    OpenAI's objective with Sora is to provide creators with a versatile tool for editing and content creation, aiming to cater to the creative needs of users.

    Safety and Future Prospects

    In contrast to a video featuring Will Smith consuming spaghetti that was revealed a year ago, content generated by Sora is described as "hyperrealistic" by the OpenAI CTO, with the exception of some anomalies in hand and finger representations. The company is adamant about ensuring the safety of the tool before its public release, with plans to watermark the Sora-generated videos, similar to other text-to-image tools.

    Impact on Creators and Data Usage

    CTO Murati emphasized that the text-to-video generation model is intended to complement creators' work rather than replace them, serving as a tool to enhance creativity. Regarding the training data for Sora, she mentioned that publicly available and licensed data, possibly sourced from platforms like YouTube, Facebook, and Instagram, was utilized.

    Moreover, Sora's generated clips do not include audio currently, but there are intentions to incorporate audio in the future, possibly in an enhanced version of Sora under a different name. Just like DALL-E, users may have to pay to access and use the model, with OpenAI focusing on creating an economically viable model that involves input from contributors in the film industry and beyond.

    Tags: , ,
  • Samsung to release One UI 6.1 update for older Galaxy devices by March.

    Samsung to release One UI 6.1 update for older Galaxy devices by March.

    The Samsung Galaxy S24 series embodies the quintessence of a top-tier smartphone in 2024, offering a plethora of cutting-edge features. However, what truly stands out are the innovative Galaxy AI capabilities integrated into Samsung's latest Android 14 interface, One UI 6.1.

    Galaxy AI Features Unveiled

    Samsung's unveiling of the Galaxy AI features has captivated tech enthusiasts worldwide. These features, initially introduced with One UI 6.1, are now set to extend to older Galaxy smartphones and tablets come March. The official announcement from Samsung confirms the commencement of One UI 6.1 rollout to legacy Galaxy devices by the end of March 2024.

    One UI 6.1 Update Details

    A comprehensive list of Samsung Galaxy devices slated to receive the One UI 6.1 update has been disclosed by the company. Users can anticipate the update on a range of devices, including the Galaxy S23 series, Galaxy Z Flip 5, Galaxy Z Fold 5, and the Galaxy Tab S9 lineup.

    Galaxy AI Feature Highlights

    The forthcoming update heralds a new era for Galaxy device users, with the introduction of remarkable Galaxy AI features. Notable inclusions encompass Circle to Search, Generative Edit, Live Translate for voice calls, and Note Assist within the Samsung Notes app. These features aim to enhance user experience by offering functionalities like object search, image editing, real-time translation, and text summarization with formatting options.

    Tags: , ,