All Updates
Page 1112 of 1544
March 21, 2026
New MacBook Pro M5 Still Fast
Apple's 16-inch MacBook Pro with M5 Pro/Max chips maintains its top spot for creators with the same design since 2021. Performance remains exceptional, aligning with prior M4 reviews. Questions whether current owners should upgrade.
Apple MacBook Neo: $599 Budget Beast Reviewed
Engadget reviews Apple's $599 MacBook Neo as a capable low-cost laptop outperforming rival Windows PCs in speed, screen, and UX. Companion iPhone 17e offers affordable iOS entry. Lineup bolstered by M4 iPad Air and M5-powered MacBook Air/Pro.
Karpathy Stops Coding, Embraces AI Agents
Andrej Karpathy reveals he's largely stopped handwriting code, now spending 80% of time directing AI agents. He highlights token throughput as the new productivity metric over GPU power. Predicts paradigm shift to API-driven software and agent-orchestrated research.
AI Boosts Productivity, Cuts Junior Hires
Anthropic report shows 94% theoretical task coverage but only 33% adoption; no mass layoffs but 14-20% drop in young hires. AI augments seniors (14% productivity gain) while replacing entry-level roles quietly.
arXiv Declares Independence from Cornell
arXiv, the key preprint server, has become an independent nonprofit after separating from Cornell University. The move aims to secure funding for surging submissions and to combat 'AI slop' in papers. This ensures long-term sustainability for scientific sharing.
Gemini Task Automation: Slow but Impressive
Gemini's new task automation feature on Pixel 10 Pro and Galaxy S26 Ultra enables the AI to control apps for food delivery and rideshare services. It's in beta, limited to a small subset of apps, slow, and clunky, but represents the first true AI assistant autonomously working on a phone.
Beyond AI Hype: IP Eternal Life
Dismisses overhyped AI replacement narrative. Explores IP immortality as AI actors debut. Guangxian and Mutong bet on 'slow business' model.
Micron Q2 Surges 75% on AI Storage Boom
Micron's FY2026 Q2 revenue hit $23.86B, up 75% QoQ, beating $19.9B expectations, driven by 60%+ price hikes in DRAM ($18.8B) and NAND ($5B). HBM revenue ~$2.7B from HBM3E/HBM4 ramp; Q3 guides $32.75-34.25B with 81% margins. Capex raised to $25B amid AI inference demand shift.
NVIDIA DLSS 5 Sparks Contradiction and Backlash
NVIDIA released DLSS 5, an AI-driven visual super-resolution technology claiming breakthrough photorealistic lighting and materials per frame using color and motion vectors. Jensen Huang's statements appear to contradict company employees. The launch triggered strong online backlash, with users criticizing it as a low-quality AI filter.
Aiper Scuba V3 Adds AI Vision for Debris
Wired reviews the Aiper Scuba V3 pool robot, now equipped with AI computer vision. This upgrade enables the cleaner to actively search and target debris. It improves efficiency over random path cleaning.
Discuria Launches for arXiv Paper Search
Discuria.org is a new website for searching, reading, and discussing AI/ML papers from arXiv and Semantic Scholar. Users can annotate papers, comment, query an AI assistant, and use read-aloud. All features are free.
Kodiak Targets Driverless Trucks by 2026
Kodiak AI plans to launch fully driverless long-haul freight operations by end of 2026. CEO Don Burnette emphasizes that deploying autonomous trucks is only half the battle, beyond AI, perception, and mileage focus. Competitors like Aurora and Waabi advance self-driving trucks amid robotaxi progress.
DQX Adds Gemini Talking Slime AI
Square Enix is introducing 'Oshaberi Slime', a generative AI companion in MMORPG Dragon Quest X Online. It uses Google's Gemini 3 Flash model and Gemini Live API for real-time voice conversations with players. This marks a new AI-driven interactive feature in the game.
Top Hospital Partners with Met Bureau on Pollen Alerts
Zhejiang University Second Affiliated Hospital signs cooperation with Hangzhou Meteorological Bureau to fuse weather data with medical resources for allergy warnings and disease prevention. Key focus includes mapping local sensitizing pollen and building health risk models. Future plans involve personalized alerts via apps and wearables.
Cursor Shell Scandal Reverses in 24h
Cursor's wrapping controversy flips in 24 hours, shattering self-developed myth. $30B unicorn's open ecosystem faces tech moat doubts.
NIO-XPeng-Li Auto Each Shine
่ๅฐ็ (NIO, XPeng, Li Auto) showcase distinct strengths. No tolerance for underperformers in this high-stakes arena.
Ben's Builds #2: Stack, Tools & Skills
Ben's Builds #2 shares the author's personal tech stack, instructions, tools, and skills for building projects. This installment provides practical insights into his workflow. AI practitioners can draw inspiration for their own setups.
Nvidia Bets on Physical AI Amid China Robotics Rise
Nvidia CEO Jensen Huang describes China as 'formidable' in robotics due to its dominance in key components like microelectronics, motors, rare earths, and magnets. Despite the US pioneering the robotics market, its industry will rely on China's supply chain. Nvidia is betting heavily on physical AI and plans to re-enter the Chinese market.
MTP Boost for Qwen 3.5 in mlx-lm
mlx-lm PR adds Multi-Token Prediction (MTP) support for Qwen 3.5 series, achieving 1.5x throughput (15.3 to 23.3 tok/s) with 80.6% acceptance rate. Benchmarked on Qwen3.5-27B 4-bit with M4 Pro. Early implementation shows solid gains for Apple Silicon users.
4 Tips for Trustworthy Business AI Agents
Provides four essential tips for developing superior AI agents that businesses can rely on. Prepares organizations for the imminent revolution in AI-powered workplaces as agents proliferate.