The AI Canvas Newsletter #6

NVIDIA Research introduces Eureka, an AI agent utilising large language models to autonomously generate reward algorithms for training robots. This breakthrough enables robots to excel in complex tasks, from rapid pen-spinning tricks to opening drawers and more, outperforming human-written algorithms on over 80% of tasks. This innovative approach integrates generative and reinforcement learning methods, providing new possibilities for robot control and physically realistic animations.

Read the announcement post and the paper.

Delve into the dynamic world of AI with the State of AI Report 2023, where the dominance of GPT-4, the significance of open-source AI, and the evolution of governance and safety take centre stage. Explore the latest breakthroughs in various AI domains, from self-driving cars to molecular biology, and grasp the evolving landscape of AI research and technology.

Read more here.

Meta has achieved a significant milestone in brain research, unveiling an AI system that deciphers real-time visual representations from brain activity using magnetoencephalography (MEG). This innovative system not only enhances our understanding of how the brain processes images but also holds potential for non-invasive brain-computer interfaces, offering hope to those who have lost their ability to speak due to brain injuries. The study demonstrates that MEG can decode complex brain representations with millisecond precision, advancing the quest to bridge the gap between human and machine intelligence.

Read more on Meta’s blog and the paper.

Google Search is now empowering English learners with a cutting-edge feature that offers interactive speaking practice. This tool, supported by AI and deep learning models, delivers personalized feedback on pronunciation and grammar. Learners can engage in quick 3-5 minute practice sessions, making language acquisition more accessible and efficient.

Read more on Google’s blog.

Masterpiece X - Generate is a new tool for 3D creation, combining AI and NVIDIA technology. This innovative platform allows creators to generate 3D assets quickly and affordably, making 3D design accessible to everyone. The AI co-pilot empowers ideas to come to life, without the need for hardware or complex skills.

Read more with the announcement and blog.