🤖 π-0.5 Robot 🤖
Bringing adaptable AI into your home

Physical Intelligence has unveiled π-0.5, a versatile household robot capable of generalizing tasks and adapting to new environments it has never encountered before. The robot demonstrates remarkable ability to perform everyday activities like picking up items, opening refrigerators, and navigating unfamiliar home layouts. Its adaptive learning system allows it to transfer skills between different physical contexts, overcoming a major challenge in robotics and potentially bringing practical robotic assistance into everyday homes.

🎙️ Descript Agent 🎙️
AI video editing made simple

Descript has launched an AI agent to assist with video editing that can remove awkward silences, translate content, and condense long recordings into concise final products. The tool streamlines the editing process by understanding natural language requests and automatically applying appropriate edits, making professional-quality video production more accessible to creators of all skill levels. This innovation represents a significant step forward in democratizing content creation tools.

💬 ElevenLabs enables Agent transfers 💬
Specialized AI working together

ElevenLabs has announced that their conversational AI now enables seamless call transfers between agents. This allows different teams within companies to develop specialized agents in parallel—each with its own knowledge base and tools—creating more versatile and effective AI communication systems. This breakthrough enhances the practical utility of voice AI in customer service and other complex communication scenarios.

INTERESTING NEWS

📰 Washington Post Partners with OpenAI 📰

The Washington Post has announced a partnership with OpenAI, allowing ChatGPT to include summaries, quotes, and direct links to Washington Post articles in its responses. This collaboration adds the Jeff Bezos-owned publication to OpenAI's roster of over 20 major news publishers, continuing despite ongoing legal battles between OpenAI and other publishers like NYT. The Post has been actively experimenting with AI tools over the past year.

🧫 AI Outperforms Scientists in Virology 🧫

A collaboration between the Center for AI Safety, MIT's Media Lab, and other institutions revealed that AI models significantly outperformed PhD-level virologists on expert problem-solving tests. Google's Gemini 2.5 Pro scored 37.6% and OpenAI's o3 scored 43.8%, compared to the human experts' average of 22.1%. This demonstrates AI's potential for solving complex biomedical problems while raising concerns about potential misuse.

H-FARM AI RESEARCH

👩‍🎓 Participate in AI perception research 👩‍🎓

Elena from our H-FARM AI team is conducting academic research on how different generations perceive AI for her thesis. The 5-minute anonymous survey explores opinions on ethical and legal implications of artificial intelligence in society.