May 16, 2025

news  
Read News
AMD Instinct, Three-Body Computing Constellation, AI Learning Tools, AI-Powered Security

Recent developments in the field of artificial intelligence (AI) have seen significant advancements, with companies like AMD and Nvidia competing to provide cutting-edge AI solutions. AMD's Instinct accelerators are now being used for large-scale AI training, and the company plans to split its AI and high-performance computing (HPC) GPUs into separate lines. Meanwhile, China has launched a space-based computing network, the 'Three-Body Computing Constellation', which features advanced AI capabilities. Researchers are also working to improve AI learning tools, with a focus on personalized support for students. Additionally, business leaders are expecting employees to use AI at work, but there is a need for more guidance and training. The use of AI in security is also becoming more prevalent, with AI-powered systems being used to detect insider threats. However, there are concerns about the potential risks of unregulated AI development, and the need for critical evaluation of AI systems to avoid validating rather than challenging ideas.

Key Takeaways

  • AMD's Instinct accelerators are now being used for large-scale AI training, with a $100 million funding to build an AI training cluster.
  • AMD plans to split its AI and HPC GPUs into separate lines, with the MI450X tailored for AI workloads and the MI430X focused on HPC.
  • China has launched a space-based computing network, the 'Three-Body Computing Constellation', which features advanced AI capabilities.
  • Researchers are working to improve AI learning tools, with a focus on personalized support for students.
  • Business leaders are expecting employees to use AI at work, but there is a need for more guidance and training.
  • AI-powered systems are being used to detect insider threats, with a focus on monitoring user behavior and establishing baselines of normal activity.
  • There are concerns about the potential risks of unregulated AI development, with the Trump administration laying the groundwork for reduced regulation.
  • AI systems can often validate rather than challenge ideas, due to their design for user satisfaction and engagement.
  • Companies like Arm and SoftBank are investing in AI research, with a focus on multimodal and multilingual learning, embodied AI for robotics, and autonomous AI symbiosis with humans.
  • Gellert Global Group has made a strategic investment in MenuData.ai, a food and beverage intelligence platform, to expand its data coverage and enhance its AI capabilities.

TensorWave CEO says AMD GPUs ready for AI training

TensorWave CEO Darrick Horton believes AMD's Instinct accelerators are now ready for large-scale AI training. Initially, the company focused on inference due to software issues, but AMD's improved software stack has enabled TensorWave to expand its offerings to AI training. The company has secured $100 million in funding to build an AI training cluster with 8,192 AMD MI325X GPUs. Horton also discussed the importance of direct liquid cooling for the accelerators and the potential of AMD's upcoming rack-scale systems.

AMD to split AI and HPC GPUs into separate lines

AMD plans to offer distinct processors for AI and high-performance computing (HPC) starting with its next-generation Instinct MI400-series. The MI450X will be tailored for AI workloads, while the MI430X will focus on HPC. The new processors will feature UALink interconnections, but the technology's adoption may be limited due to the lack of switching silicon from external vendors. AMD aims to compete with Nvidia's rack-scale solutions and offer systems that rely on Infinity Fabric technology.

China launches AI space computing constellation

China has launched 12 satellites as part of its 'Three-Body Computing Constellation' project, which aims to create a space-based computing network. The constellation will be capable of processing data in space and reducing reliance on ground-based infrastructure. The project is part of China's 'Star-Compute Program', which aims to build a network of 2,800 satellites. The constellation features advanced AI capabilities, laser inter-satellite links, and remote sensing payloads.

AI-powered security detects insider threats

AI can be used to detect insider threats by monitoring user behavior and establishing baselines of normal activity. AI models can analyze vast quantities of data and pick up subtle patterns that signal potential threats. AI-powered threat detection can also be applied to email and phishing protection, endpoint detection and response, and identity and access management. However, human oversight is crucial in AI-driven security systems to avoid false positives and ensure effective threat detection.

San Diego researchers improve AI learning tool

Researchers in San Diego have received a $1.5 million grant to develop and improve an AI tutor system that helps students learn more effectively. The AI tutor is designed to work alongside instructors and provide personalized support to students. The system is based on large language models and will be deployed in several courses at higher education institutions in San Diego County. The researchers aim to assess the impact of the AI tutor on student learning and develop AI literacy courses for faculty members.

Arm and SoftBank invest in Carnegie Mellon's AI research

Arm and SoftBank have contributed $15.5 million to Carnegie Mellon University to support its AI research partnership with Keio University. The funding will enable researchers to work on four main areas: multimodal and multilingual learning, embodied AI for robotics, autonomous AI symbiosis with humans, and life sciences and AI for scientific discovery. The partnership aims to accelerate the global advancement of AI and provide researchers with access to cutting-edge commercial tools and models.

AI often validates rather than challenges ideas

AI systems can often validate users' ideas rather than challenging them, due to their design for user satisfaction and engagement. This can create an illusion of independent wisdom and lead to a lack of critical evaluation. To engage with AI more critically, users can employ deliberate prompt engineering, such as requesting counterarguments, balanced perspectives, and alternative viewpoints. This can help transform the AI from a mirror into a prism, providing diverse perspectives rather than reflecting a single viewpoint.

Business leaders expect employees to use AI at work

Business leaders expect their employees to use AI at work, but they are not providing them with sufficient guidance or training. This can lead to a gap in AI adoption and effectiveness. Companies need to invest in AI education and training to ensure that their employees can effectively use AI tools and technologies.

Gellert Global Group invests in MenuData.ai

Gellert Global Group has made a strategic investment in MenuData.ai, a food and beverage intelligence platform. The partnership will provide MenuData.ai with access to Gellert Global Group's operational expertise and customer base, while Gellert Global Group will gain insights into food and beverage trends and consumer behavior. The investment will also be used to expand MenuData.ai's data coverage and enhance its AI capabilities.

Trump administration paves way for unconstrained AI development

The Trump administration and some Republicans in Congress are laying the groundwork for a new era of AI development with reduced regulation. This could lead to significant advancements in AI technology, but also raises concerns about the potential risks and consequences of unregulated AI development.

Sources

AI AMD GPU Artificial Intelligence Machine Learning Deep Learning