NTT Unveils AI Chip for 4K Video Processing and Launches Physics of AI Group

NTT Launches Physics of AI Group

NTT Research has announced the formation of a new AI basic research group called the Physics of Artificial Intelligence (PAI) Group. This initiative, unveiled at NTT's annual Upgrade event, aims to advance our understanding of AI's "black box" nature and improve trust and safety outcomes 1

1

The PAI Group, led by Hidenori Tanaka, an expert in physics, neuroscience, and machine learning, will focus on an interdisciplinary approach to AI research. The group's mission includes deepening our understanding of AI mechanisms, creating controllable AI spaces for observation, and improving trust between AI and human operators 1

1

Collaborations with prestigious institutions such as Harvard University's Center for Brain Science, Princeton University, and Stanford University will support the group's research efforts. The PAI Group builds on NTT Research's previous contributions, including a widely cited neural network pruning algorithm and a bias-removal algorithm for large language models 1

1

NTT's Breakthrough in AI Chip Technology

Alongside the PAI Group announcement, NTT Corporation has introduced a groundbreaking large-scale integration (LSI) chip designed for real-time AI inference processing of ultra-high-definition video. This first-of-its-kind technology enables AI inference at 4K resolution and 30 frames per second while maintaining low power consumption 2

2

The LSI chip utilizes an AI inference engine created by NTT, which enhances computing efficiency. This innovation makes it suitable for edge devices and terminals with strict power constraints. The chip's capabilities include:

1. Real-time AI inference at 4K resolution
2. Power consumption of less than 20 watts
3. Ability to detect individuals or objects from up to 150 meters above ground when installed on drones 2

   2

Applications and Future Developments

The new AI chip technology has significant potential for various applications:

1. Beyond visual line of sight drone operations for public infrastructure inspections
2. Public safety monitoring
3. Live sporting events coverage 2

   2

NTT plans to commercialize this LSI chip within fiscal year 2025 through its operating company, NTT Innovative Devices Corporation. Researchers are also exploring the chip's application in the data-centric infrastructure of the Innovative Optical and Wireless Network (IOWN) Initiative 2

2

Furthermore, NTT is collaborating with NTT DATA, Inc. to advance the LSI chip in relation to Attribute-Based Encryption (ABE) technologies, enabling secure data sharing and fine-grained access control 2

2

Related Stories

NTT Unveils Industry AI Cloud and Breakthrough Technologies at R&D Forum 2024

NTT DATA Introduces Ultralight Edge AI Platform for IoT and Manufacturing

NVIDIA Spearheads AI Revolution in Japan: Partnerships, Infrastructure, and Innovation

SoftBank and Nvidia Pioneer AI-Powered 5G Network, Unveil Ambitious AI Infrastructure Plans

Implications for AI Research and Industry

The establishment of the Physics of Artificial Intelligence Group and the development of the new AI chip represent significant advancements in AI research and application. These initiatives aim to address critical issues in AI development, including:

1. Improving energy efficiency in AI computing
2. Enhancing trustworthiness and safety in AI systems
3. Bridging the gap between biological and artificial intelligence
4. Advancing edge computing capabilities for AI applications

As AI continues to play an increasingly important role in various sectors, NTT's efforts to demystify AI's inner workings and improve its efficiency could have far-reaching implications for the future of technology and society 1

1

2