•
September 19, 2024
•
4
Hello Niuralogists!
Welcome to this week’s edition, where we dive into the latest advancements in artificial intelligence. We’ll explore how these innovations are shaping various aspects of our lives—from the workplace and business to policies and personal experiences. This issue highlights some fascinating updates, including medical AI tech, new industry partnership and preparations for the future of AI in terms of governance and infrastructure.
For more in-depth coverage, keep reading…
The Global Artificial Intelligence Infrastructure Investment Partnership (GAIIP), led by Microsoft and BlackRock, is aiming to eventually raise up to $100 billion, with an initial target of $30 billion, to develop data centers and energy infrastructure to support AI workloads. Microsoft CEO Satya Nadella emphasized the initiative’s goal of building sustainable infrastructure to drive innovation across sectors. Other key participants include Global Infrastructure Partners, which BlackRock is acquiring, and MGX, a tech investor backed by UAE's Mubadala and G42. The funds will address growing demand for AI data centers powered by Nvidia GPUs, essential for AI models like OpenAI’s ChatGPT.
AI pioneers from around the world are calling for international oversight to prevent the potential catastrophic risks posed by rapidly advancing artificial intelligence. In a statement, scientists from the U.S., China, and other countries highlighted concerns that AI could surpass human control or be maliciously used, leading to global harm. They propose creating AI safety authorities in each country to monitor AI systems, with international coordination to set guidelines for risk management. The group emphasized the importance of collaboration between nations, particularly amid tensions between the U.S. and China. With AI technology evolving faster than regulations can keep up, the call for oversight echoes Cold War-era efforts to prevent nuclear disaster.
Intel announced a multi-billion-dollar partnership with Amazon to manufacture custom artificial intelligence (AI) chips, boosting its stock by over 8%. Intel will produce custom AI chips using its 18A foundry process technology and develop custom Xeon 6 chips for Amazon Web Services (AWS). Intel CEO Pat Gelsinger emphasized the partnership’s role in advancing innovation and supporting a domestic AI supply chain. Additionally, Intel plans to establish its foundry business as an independent subsidiary, aiming for clearer separation and potential external funding. To streamline operations, Intel will pause construction on new factories in Germany and Poland due to market demand, while U.S. expansions will continue.
Medical researchers developed an AI-based early warning system, Chartwatch, which is significantly reducing unexpected patient deaths. Launched in 2020, the tool analyzes vital signs and lab results to predict potential deterioration every hour. A new study published in the Canadian Medical Association Journal found a 26% drop in unexpected deaths among patients using the system. Chartwatch enhances clinical care by alerting medical staff to potential issues earlier, allowing timely interventions that save lives. Despite its success, researchers note that further testing in different hospitals is needed. The tool has already demonstrated the potential of AI in improving patient outcomes and is expected to expand across more hospitals in Ontario through Canada's GEMINI hospital data-sharing network.
Salesforce's latest updates to Slack transform it into an AI-driven work operating system, addressing workplace inefficiencies by centralizing data, automation, and collaboration. With the introduction of Agentforce, teams can use AI to access insights, perform tasks, and manage workflows directly in Slack. Integration with third-party AI agents from partners like Adobe and Anthropic enhances content creation, project management, and decision-making. Additionally, Salesforce channels now allow seamless collaboration around CRM data. These introductions aim to reduce time spent on low-value tasks and increase productivity by automating routine processes and improving access to key information.
📬 Receive our amazing posts straight to your inbox. Get the latest news, company insights, and Niural updates.
In the Hong Kong Polytechnic University study titled “Schrodinger’s Memory: Large Language Models,” researchers delve into how Large Language Models (LLMs) exhibit memory capabilities, drawing comparisons to human cognition. Through experiments, they demonstrated how LLMs recall information based on input cues, much like humans, yet operate through a distinct mechanism. The study also introduces a method to assess these memory abilities, highlighting the parallels and differences between how LLMs and humans process, store, and recall information. This research sheds light on the evolving capabilities of AI in language and cognition, pushing the boundaries of how we understand and develop AI systems.
In robotics, world models simulate how the world responds to a robot’s actions, enabling robots to predict multiple future outcomes based on their movements. By learning from raw sensor data, these models can simulate complex interactions with objects like doors, deformable items, and even human environments. Unlike traditional physics-based simulators, world models incorporate the real-world complexity without manual setup, helping evaluate robots across millions of scenarios. While powerful, they still face challenges such as object coherence and adherence to physical laws.
🚀 Supademo 3.0 generates engaging, interactive product demos
👩💻 Cracked copilot creates and refine motion graphics from a text prompt
📄 AnyParser Sandbox enhances document retrieval and text extraction efficiency
🤖 AutoGen builds multi-agent conversations for LLM workflows
💭 DreamRP creates interactive stories and characters with ethical AI
