Researchers at UC Berkeley and ETH Zurich used generative AI to empower $120 robot arms for cleaning spills. Jannik Grothusen and Kaspar Janssen programmed the robots in four days with GPT-4o, designing a visual language model for human-robot interaction using 100 demos.

Last week, OpenAI launched its ChatGPT Search, seen as a potential “Google killer.” However, after testing, users found it impractical for routine searches. While it excels at answering detailed questions, it struggles with short, navigational queries, limiting its effectiveness.

The EU has halted Nvidia’s $700 million acquisition of AI startup Run:ai for review after Italy’s competition regulators expressed concerns. The bloc’s assessment could introduce delays and uncertainties, potentially affecting the competitive landscape in Europe.

Starting today, developers can use Google’s Gemini API and AI Studio to enhance AI services with data from Google Search, allowing for more accurate responses. Grounding prompts helps avoid inaccuracies, as it connects models with verifiable data and offers users additional context.

Google Maps is set to enhance its features with Gemini, Google’s generative AI. Users in the U.S. will get AI-assisted suggestions for activities and improved navigation that highlights lanes and road signs. The update also includes weather disruption reporting and Immersive View expansion.