Encord is turning to brain waves to improve AI training, specifically for physical AI. The startup is testing EEG headsets to collect brain-wave data, which can help solve the training data crisis. By working with Zander Labs, Encord aims to develop technology that captures brain activity while workers perform tasks, providing valuable insights for model builders.
The USDA is also investing in AI research, focusing on precision agriculture, animal health, and food safety. The agency hopes to improve crop yields, reduce waste, and enhance food safety through AI innovation. Meanwhile, China's open-source AI approach is gaining attention, with industry insiders and policymakers discussing the implications of China's AI progress and the US response.
In the chip space, a Singaporean startup, Acrab, has developed an AI chip called Gelix 1, which claims to outperform Apple's M4 Pro in AI inference. The chip can run large language models, making it a significant player in the AI hardware market. On the other hand, a new MIT study identifies five AI risks with high potential for catastrophe, including uncontrollable AI systems and autonomous AI.
Researchers are also exploring the concept of 'forgetting' in AI, arguing that it's essential for lifelong AI. This approach would enable AI systems to forget unnecessary information, making them more efficient and effective. Additionally, the AI for Good Summit highlighted the need for open-source AI and local agency, emphasizing the importance of governance and sovereignty in AI development.
Key Takeaways
['Encord is using EEG headsets to collect brain-wave data for physical AI training.', 'The USDA is investing in AI research for precision agriculture, animal health, and food safety.', "China's open-source AI approach is gaining attention, with implications for the US response.", "Acrab's Gelix 1 AI chip claims to outperform Apple's M4 Pro in AI inference.", 'MIT study identifies five AI risks with high potential for catastrophe.', 'Researchers argue that forgetting is essential for lifelong AI.', 'The AI for Good Summit emphasized the need for open-source AI and local agency.', "70% of AI's value comes from people-related actions, according to a new study.", 'A new audit tool, HackDetect, assesses score inflation in agent benchmarks.', 'Cognitive neuroscientist Gleb Tsipursky discusses people most likely to resist AI.']Encord uses brain waves to boost AI training
Encord is testing EEG headsets to collect brain-wave data, aiming to solve the training data crisis in physical AI. The startup believes brain signals can encode intent and error-recognition that cameras miss. Encord is working with Zander Labs to develop this technology.
Brain waves key to unlocking physical AI
Encord is collecting brain-wave data to improve physical AI training. The company is working with Zander Labs to develop EEG headsets that capture brain activity while workers perform tasks. This data can help model builders understand when higher-effort models are needed.
Forbes: People Most Likely to Resist AI
A new study suggests that 70% of AI's value comes from people-related actions. Cognitive neuroscientist Gleb Tsipursky discusses the people most likely to resist AI and why they resist.
Agent Benchmarks and Capability Claims
A new study discusses the validity of agent benchmarks and their limitations. Researchers introduce HackDetect, a post-hoc audit that identifies exposures and assesses score inflation.
USDA Pushes for AI Innovation in Agriculture
The USDA is investing in AI research to improve crop yields, reduce waste, and enhance food safety. The agency focuses on precision agriculture, animal health, and food safety.
China's Open-Source AI Approach
China's open-source AI models are gaining attention. Industry insiders and policymakers are discussing the implications of China's AI progress and the US response.
China's AI Strategy
China is using its steel playbook to challenge America's AI dominance. The country's open-source AI models and industrial force are reshaping the global AI landscape.
Teach AI to Forget
Researchers argue that forgetting is essential for lifelong AI. The study suggests that AI systems should be designed to forget unnecessary information.
Singapore Startup's AI Chip
A Singaporean startup, Acrab, has developed an AI chip called Gelix 1. The chip claims to beat Apple's M4 Pro in AI inference and can run large language models.
MIT Study on AI Risks
A new MIT study identifies five AI risks with high potential for catastrophe. The risks include uncontrollable AI systems and autonomous AI.
AI for Good Summit
The AI for Good Summit discussed AI governance and sovereignty. The event highlighted the need for open-source AI and local agency.
Sources
- Encord is collecting brain-wave data to solve physical AI's training data crisis
- Are brain waves the next unlock for physical AI?
- The People Most Likely To Resist AI, And Why They Resist
- Do Agent Benchmarks Measure Capability? Protocol Validity in the Age of Agentic AI
- Michigan Farm News
- Editorial | US aims to dominate AI but China’s open-source approach is unstoppable
- China is using the steel playbook on AI — and America’s tech billionaires are standing in the blast zone
- To build lifelong AI, teach it to forget
- Singaporean Startup’s Gelix 1 AI Chip With 20-Core CPU, 273GB/s Bandwidth, Said To Beat M4 Pro In AI Inference, Can Run Up To 100B LLMs While Using Up A Mac mini’s Footprint
- These 5 AI risks have the highest potential for catastrophe
- The AI for Good Summit and America’s AI Sovereignty Problem
Comments
Please log in to post a comment.