News

9月18

  • Lenovo Group to manufacture AI servers and set up AI R&D labs in India

    10:35

    Lenovo Group on Tuesday announced that it will start manufacturing artificial intelligence (AI) servers at its plant in southern India and open an AI server-focused research and development lab at its tech hub in Bengaluru. Lenovo said it aims to produce 50,000 AI rackmount servers and 2,400 graphics processing unit (GPU) servers per year at its plant in Puducherry, India, which are specialized for resource-intensive tasks such as machine learning. Speaking to Reuters, Amar Babu, president of Lenovo Asia Pacific, said, "These servers are not only for local consumption but also for...

    Share to
  • Bill Gates: If he could start over, he'd start an 'AI-centric' startup

    10:24

    Microsoft co-founder Bill Gates recently said in an exclusive interview with foreign media, if he had to start a new business, he would join the race of artificial intelligence with a startup "centered on artificial intelligence". Gates pointed out that the startup he envisioned might not be as successful as Microsoft. (Tencent Technology)

    Share to
  • Just over a year old, AI unicorn valuation rises to $21.2bn in Eurozone

    10:24

    Another highly sought-after AI unicorn has emerged to kick off the "AI race" in Europe.Poolside, founded in early 2023 and headquartered in Paris, France, has seen its valuation grow rapidly in the short term.Poolside belongs to the AI programming space, a market that is in hot demand right now. The latest news is that it is currently negotiating a new round of financing of $500 million, with a valuation of nearly $3 billion, and Bain Capital is negotiating to lead this round of investment. (InvestmentNews.com)

    Share to
  • o1 sudden insider exposure? Google reveals principles earlier, big models don't have moats with software alone

    10:24

    Less than 1 week after its release, the moat of OpenAI's strongest model, o1, is gone. It has been discovered that a Google DeepMind paper published in August reveals that the principles are almost identical to the way o1 works. This study showed that increasing test-time computation is more effective than expanding model parameters. Based on the computationally-optimal (compute-optimal) test-time computation scaling strategy proposed in the paper, a smaller base model can outperform a 14 times larger model on some tasks. Some users said: this is almost the principle of o1. (Quantum Bit)

    Share to
  • Apple pushes out iOS 18 and announces Apple AI coming next month

    10:23

    Apple pushed out the official iOS 18 release for the second-generation iPhone SE and all models after it, 24 in total. The version will be pre-installed at the factory on the iPhone 16 and 16 Pro series, which will go on sale this week. The new version of iOS supports a new custom icon arrangement feature, introduces new Control Center and App Lock features, and introduces Game Mode for the first time. Apple has also announced that Apple Intelligence Apple Intelligence will make its official debut next month, and will be available first in the US, and will support English in more regions later this year, supporting iPhone,...

    Share to
  • Microsoft Launches Copilot Pages AI Collaboration Tool

    10:22

    Yesterday, Microsoft launched the new Copilot Pages feature. According to the introduction, teams of users can collaborate in Copilot Pages and see everyone's progress in real time.Copilot bots will also work in the pages like partners, and users can use Copilot at any time to query and add data, files, and more from the web into the pages. The feature has been rolled out to Microsoft 365 Copilot customers and will be fully available to all subscribers this month. On top of that, Microsoft has officially made it available for...

    Share to
  • BlackRock and Microsoft to Raise $30 Billion for AI Investments, Backed by NVIDIA

    10:22

    On September 17, BlackRock and Microsoft announced that they are teaming up to invest in data centers and other infrastructure to support artificial intelligence (AI). The strategy, called the Global AI Infrastructure Investment Partnership, aims to attract $30 billion in private equity and use that money to make up to $100 billion in investments. NVIDIA will utilize its expertise in artificial intelligence to NVIDIA will support the alliance with its expertise in AI data centers and factories.

    Share to
  • Hong Kong Considers Issuing a Rules Framework for AI Applications in the Financial Sector

    10:21

    Sept. 17 (Bloomberg) -- The Hong Kong government is preparing to issue its first policy statement on the application of artificial intelligence in the financial sector, a move that could catalyze the technology's use in areas such as trading, investment banking and cryptocurrencies, according to people familiar with the matter. Hong Kong's Financial Services and the Treasury Bureau plans to issue a framework of guidelines on the ethical use of AI and general principles for applying the technology in the financial sector, said the people familiar with the matter, who declined to be named because the information wasn't made public. Officials are still drafting the document while seeking feedback from the industry; details are still subject to change in the coming weeks, the people familiar with the matter said.

    Share to
  • Microsoft and Vodafone sign artificial intelligence office deal

    10:21

    Microsoft claimed to have signed a large deal with Vodafone Group, which will use artificial intelligence assistants in Office, and unveiled a series of new AI tools designed to help employees create Excel charts, prioritize emails and collaborate with each other. In a blog post published ahead of Monday's webcast, Microsoft said Vodafone would buy 68,000 licenses for the enterprise version of the assistant, dubbed Copilots, after initial trials found that employees using the assistants could save about three hours a week each.

    Share to
  • Salesforce Announces New $500 Million Artificial Intelligence Fund

    10:21

    Salesforce announced a new $500 million AI fund, bringing its total commitment to AI innovators to $1 billion.

    Share to
  • Apple: iOS 18 expected to introduce Apple Smart features starting next month

    10:20

    Sept. 17 (Bloomberg) -- iOS 18 is now available and is expected to introduce Apple's artificial intelligence capabilities starting next month, Apple said; vOS 18 is now available as a free software update for Apple TV 4K and Apple TV HD; and watchOS 11 will go live on Sept. 16th.

    Share to
  • OpenAI Safety and Security Committee becomes independent board oversight committee as Altman quits

    10:20

    Sept. 17 (Bloomberg) -- OpenAI has made an announcement that its Safety and Security Committee will be reorganized into an independent board oversight committee, with Zico Kolter, chair of the Machine Learning Department at Carnegie Mellon University's School of Computer Science, assuming the position of chair. This committee also includes Quora co-founder and CEO Adam D'Angelo, retired U.S. Army General Paul Nakasone, and Nicole Seligman, former executive vice president and general counsel of Sony, all of Ope...

    Share to
  • Intel to customize AI chips for Amazon

    10:20

    In a statement on Monday local time, Intel CEO Pat Kissinger said it has added Amazon's AWS as a customer for the company's manufacturing business. According to the statement, Intel and AWS will jointly invest in a customized chip for AI computing in a "multi-year, multi-billion dollar framework." The work will rely on Intel's 18A process. Intel has also delayed new factories in Germany and Poland, but remains committed to expansion in the US states of Arizona, New Mexico, Oregon and Ohio. Intel is also looking to accelerate the execution of its $10 billion cost-savings program,...

    Share to
  • Dark Side of the Moon Founder Shik Lin Yang: The Super App of the AI Era Will Be an Assistant

    10:19

    September 17th news, the founder of the dark side of the moon, Yang Shiklin, recently shared his thoughts on the development of the artificial intelligence industry at Xuanhuai College of Tianjin University. He said that in the AI era, the probability that the super application will be an assistant, the demand for intelligence is a very pervasive demand, only today the ability is still in the primary stage. At the same time, this market is the process of adapting and embracing new technologies, in fact, as the effect continues to get better and the cost continues to fall, it will lead to an increasingly strong market adaptation. He believes that there will definitely be opportunities for mass market adoption in the next 5 to 10 years. Behind all the software used now is made...

    Share to
  • Gemini's Google Photos Extension: Helping Users Remember, Record Information

    10:18

    Google is testing the Google Photos extension in the Gemini app, and has given it a first look in the latest 15.36.40.29 beta version of the Google app. The extension offers two main ways to use it: "Recall past activities, such as photos of the first food you tasted," and "Use photography to record information. The extension will soon be available in Google Photos' "Ask Photos" and natural language search features. Google has launched Natural Language Search, a feature that allows users to search for their Go...

    Share to
  • OpenAI o1 to follow, how? This GitHub project collects related interpretations, blogs, papers

    10:17

    This GitHub project aggregates technical explanations, blogs, and related papers on the OpenAI o1 model, providing a comprehensive repository to keep up with the latest developments; the project includes blog posts on detailed introductions to the o1 model, training methodologies, and security enhancements, as well as papers related to the model, such as automated theorem proving and the handling of complex reasoning tasks; the repository also covers the The repository also covers the practical applications and potential impacts of the o1 model, exploring its prospective and technological innovations in AI, providing valuable information and insights for researchers and developers.

    Share to
  • No upper limit for o1 methods? Yao Ban Ma Teng Yu and others mathematically prove that enough tokens can solve arbitrary problems

    10:17

    Teng-Yu Ma and Denny Zhou demonstrate mathematically that the Transformer model can simulate digital circuits of arbitrary polynomial sizes, with no theoretical upper bound for problem solving; CoT (Chain of Thought) significantly enhances Transformer's expressive capability to handle more complex problems, including parallelizable and serializable computational problems; and Experiments verify the accuracy of CoT in improving Transformer's ability to handle various types of problems (e.g., modulo arithmetic, permutation group combination, iterative squaring, and circuit-valued problems), and show that CoT can effectively expand...

    Share to
  • Tao Zhexuan actual test o1: it's like giving advice to a mediocre but marginally competent graduate student

    10:17

    Tao Zhexuan tested OpenAI's o1 model and found that it can effectively identify and solve complex mathematical problems, with performance similar to that of a junior graduate student; although the o1 model showed competence in generating hypotheses and solving problems, it still suffers from errors and comprehension limitations, and needs further guidance and tool support; Tao Zhexuan's tests showed that, despite the progress of the o1 model, it still needs to improve in autonomously generating key concepts and error avoidance, and its usefulness and accuracy need to be improved.

    Share to
  • "What will 'AI Scientist' bring to science?Controversies in the AI for Science boom

    10:17

    Sakana AI launched AI Scientist to automate the process of scientific research, including hypothesis generation, experimental runs, results evaluation, and paper writing; AI Scientist performs similarly to junior PhD students in real-world applications, with variable quality of ideas and challenges in writing and understanding results; despite limitations and moral and ethical issues, AI Scientist captures the trend of AI for science, but its utility and accuracy still need to be validated by future research.

    Share to
  • Fei Fei Li and his 24-member band create the "Big World Model", and Hinton supports them!

    10:16

    World Labs, a company founded by Feifei Li, is dedicated to developing 'big world models' that enable AI to perceive, generate and interact in 3D worlds; World Labs completed a new round of funding of $230 million, backed by a number of high-profile investors and companies including Geoffrey Hinton and Jeff Dean; the company's technology focuses on spatial intelligence, aiming to AI for modeling and interacting with 3D worlds, advancing computer vision and physical intelligences .

    Share to

Search