Software Engineering & Open-Source

Breakthrough Hydrogen-Based AI Chip Mimics Human Brain for Self-Learning A research team from the Daegu Gyeongbuk Institute of Science and Technology (DGIST) has developed the world’s first two-terminal artificial intelligence semiconductor that utilizes hydrogen ions for self-learning and memory. This innovative technology precisely controls hydrogen ion injection and discharge with electrical signals, emulating the human brain’s ability to perform computation and storage simultaneously. Unlike conventional oxide-based memory devices that suffer from stability issues due to oxygen vacancies, this hydrogen-based approach resolves speed degradation and high power consumption in modern AI data processing. The neuromorphic semiconductor operates by changing its conductivity based on electrical signals and maintaining that state, similar to an artificial synapse. This breakthrough was achieved in a two-terminal vertical structure, which is advantageous for high-density AI chips due to its potential for high integration and simple manufacturing. The device has demonstrated high reliability, successfully running for over 10,000 repetitive operations. ...

Agentic AI Takes Center Stage, Shifting Focus from Generative Models In March 2026, the artificial intelligence landscape is undergoing a pivotal shift from generative AI to more autonomous, agentic AI systems. These advanced AI agents are engineered to execute complex, multi-step workflows with minimal human oversight, fundamentally transforming business operations. This trend is highlighted by the release of new models like OpenAI’s GPT-5.4, which features native computer-use capabilities, allowing it to interact directly with software interfaces. The industry’s focus is evolving from experimentation to implementation, embedding AI as an invisible infrastructure within enterprise environments. Upcoming events like NVIDIA’s GTC 2026 are also expected to heavily feature agentic and physical AI. ...

Met Office Enhances Weather Forecasting with Advanced AI and Machine Learning The Met Office is increasingly using artificial intelligence and machine learning to advance its forecasting capabilities for both weather and climate science. These technologies present a significant opportunity for transformation at every stage of the meteorological process, from atmospheric analysis to delivering actionable insights. Through its National Capability AI Programme (NCAIP), the Met Office is aiming to deliver substantial benefits to the public, businesses, and government. In climate science, AI provides new methods to improve localized impact projections, enhance small-scale process modeling, and support the analysis of extreme weather events. The AI4Climate programme, funded by the Department for Science, Innovation and Technology, is specifically exploring cutting-edge AI to strengthen climate models and deliver improved climate information more efficiently. ...

University of Kentucky Launches State’s First Bachelor of Science in AI Program The University of Kentucky’s Stanley and Karen Pigman College of Engineering has introduced the first Bachelor of Science in Artificial Intelligence in the Commonwealth. Housed in the Department of Computer Science, the new undergraduate degree program is set to welcome its inaugural class in Fall 2026. This initiative responds to the surging demand for AI and machine learning specialists, an occupation projected to be the third fastest-growing globally between 2025 and 2030. The program is designed to equip students with essential technical skills and a strong ethical foundation to apply AI responsibly across various industries. With this new offering, the University of Kentucky joins a select group of universities nationwide providing a bachelor’s degree specifically in AI. ...

MLCommons Releases MLPerf Inference v5.1, Setting New AI Performance Benchmarks MLCommons has released MLPerf Inference v5.1, the latest version of its benchmark suite, introducing new tests to evaluate AI system performance on modern workloads. This update saw a record 27 organizations submitting systems for evaluation. The new version features three new benchmarks, including one for the DeepSeek-R1 reasoning model, and implements tighter latency requirements for certain large language model tests. Notably, the popular Llama 2 70B benchmark demonstrated significant performance gains, with top systems running up to 50% faster than in the previous release. The v5.1 release also marked the first submission of a heterogeneous system that intelligently balances inference workloads across different accelerator types. ...

OpenAI Unveils GPT-Realtime: A Leap Forward in Speech-to-Speech AI OpenAI has officially launched GPT-Realtime, its most advanced speech-to-speech model, and has moved its Realtime API out of beta with new enterprise-grade features. GPT-Realtime is engineered to understand and generate audio directly, which significantly reduces latency and preserves crucial nuances like emotion and laughter often lost in traditional text-to-speech systems. The new model showcases enhanced performance in reasoning, instruction following, and function calling. The updated Realtime API now supports image input, Session Initiation Protocol (SIP) for phone network connectivity, and Multi-Capability Pipelines (MCP) for pluggable features. OpenAI also introduced two new voice options, Cedar and Marin, and has cut the pricing for the Realtime API by 20%. ...

AI & ML Developments: NVIDIA GPUs, Mozambique’s Tax Strategy, and University Partnerships Today’s top stories in Artificial Intelligence and Machine Learning feature Lambda’s deployment of Supermicro’s NVIDIA Blackwell GPU servers to expand its AI cloud infrastructure. In other news, Mozambique is adopting a new AI-centric strategy to transform its tax system, and Elizabeth City State University is collaborating with Amazon to advance AI and ML education for its students. Sources: ...

EngineAI Unveils New Humanoid Robots at World Robot Conference 2025 EngineAI showcased its latest advancements in humanoid robotics at the World Robot Conference 2025 in Beijing. The company debuted the T800, a heavy-duty, full-sized humanoid robot standing 1.85 meters tall, and also presented the PM01, which demonstrated impressive dexterity and high-speed running. Additionally, EngineAI announced the upcoming release of the SA02, a 1.25-meter-tall humanoid robot priced at approximately $5,500, aimed at making the technology more accessible to young engineers and enthusiasts. The company also introduced the JS01, a high-mobility quadrupedal robot designed for rugged terrains, suitable for industrial inspections, rescue, and logistics. These innovations highlight EngineAI’s focus on accelerating the commercialization of robotics and fostering a collaborative global ecosystem. ...