Wes Roth (@WesRoth)
마이크로소프트가 전 Ai2 CEO Ali Farhadi를 Corporate Vice President로 영입했고, 그와 함께 우수 AI 연구팀도 합류해 새롭게 재편된 AI 조직을 강화한다고 밝혔다. 연구 역량 확대와 조직 개편 측면에서 중요한 인사다.
https://x.com/WesRoth/status/2036533511973142822
#microsoft #ai2 #hiring #airesearch #organization
Wes Roth (@WesRoth) on X
Microsoft has hired Ali Farhadi, the former CEO of the Allen Institute for Artificial Intelligence (Ai2), as its new Corporate Vice President. He is bringing a team of elite AI researchers with him to bolster Microsoft's newly restructured AI division.
Farhadi recently stepped
The SCARIEST chart in AI
Visualizing the exponential.
This data-driven look at compute-to-intelligence ratios suggests we are entering a phase of self-improving code that exceeds all previous human projections for 2030.
#AIData #TechStatistics #FutureShock #AIResearch #TechNews #ExponentialGrowth
https://www.technology-news-channel.com/the-scariest-chart-in-ai/
The SCARIEST chart in AI
The latest AI News. Learn about LLMs, Gen AI and get ready for the rollout of AGI. Wes Roth covers[...]
Researchers from Meta FAIR, Cornell and Carnegie Mellon have developed TinyLoRA, a fine-tuning method that achieves 91.8% accuracy on GSM8K math benchmark using just 13 trainable parameters on Qwen2.5-7B. The approach uses weight sharing across layers and truncated SVD to scale down trainable parameters dramatically. Reinforcement learning proves 100-1000x more efficient than supervised fine-tuning at extremely low parameter counts.
https://www.marktechpost.com/2026/03/24/this-ai-paper-introduces-tinylora-a-13-parameter-fine-tuning-method-that-reaches-91-8-percent-gsm8k-on-qwen2-5-7b/ #AIagent #AI #GenAI #AIResearch #Meta
This AI Paper Introduces TinyLoRA, A 13-Parameter Fine-Tuning Method That Reaches 91.8 Percent GSM8K on Qwen2.5-7B
This AI Paper Introduces TinyLoRA, A 13-Parameter Fine-Tuning Method That Reaches 91.8 Percent GSM8K on Qwen2.5-7B
Meta Yann LeCun introduces LeWorldModel, the first JEPA that trains stably from raw pixels using just two loss terms. The system plans up to 48x faster than previous models, encoding observations with 200x fewer tokens.
https://www.marktechpost.com/2026/03/23/yann-lecuns-new-leworldmodel-lewm-research-targets-jepa-collapse-in-pixel-based-predictive-world-modeling/ #AIagent #AI #GenAI #AIResearch #Meta
Yann LeCun’s New LeWorldModel (LeWM) Research Targets JEPA Collapse in Pixel-Based Predictive World Modeling
Yann LeCun’s New LeWorldModel (LeWM) Research Targets JEPA Collapse in Pixel-Based Predictive World Modeling
Meta researchers have developed Hyperagents, a framework where AI systems can modify their own learning mechanisms. Unlike previous self-improving AI that required fixed meta-level systems, Hyperagents can rewrite their own improvement procedures. This metacognitive capability represents a step toward AI that continuously evolves without human-designed boundaries.
https://www.marktechpost.com/2026/03/23/meta-ais-new-hyperagents-dont-just-solve-tasks-they-rewrite-the-rules-of-how-they-learn/ #AIagent #AI #GenAI #AIResearch #Meta
Meta AI’s New Hyperagents Don’t Just Solve Tasks—They Rewrite the Rules of How They Learn
AI hyperagents enable metacognitive self-modification to improve both task performance and the system's own improvement process
Luma AI's Uni-1 model beats Google and OpenAI on image generation benchmarks while being 10-30% cheaper. The key innovation: using autoregressive generation (like LLMs) instead of diffusion, allowing the model to reason while creating images.
https://venturebeat.com/technology/luma-ai-launches-uni-1-a-model-that-outscores-google-and-openai-while #AIagent #AI #GenAI #AIResearch #LumaAIMIT and the Hasso Plattner Institute have launched a 10-year collaborative hub to explore the intersection of AI and creativity. The initiative will fund joint research, fellowships, and educational programs at the MIT Morningside Academy for Design and HPI in Potsdam.
https://news.mit.edu/2026/mit-hasso-plattner-institute-collaborative-hub-for-ai-and-creativity-0320 #AIagent #AI #GenAI #AIResearch #MITMIT and Hasso Plattner Institute establish collaborative hub for AI and creativity
MIT and the Hasso Plattner Institute establish the MIT and HPI AI and Creativity Hub, a 10-year collaboration led by the MIT School of Architecture and Planning, MIT Schwarzman College of Computing, and MIT Morningside Academy for Design to advance research and education at the intersection of AI, design, and creativity.
sayzard
Ahill
AIagent.at 🤖 AI News
Winbuzzer
DrWJK