GOOGLE'S GEMINI-SQL2 DOMINATES TEXT-TO-SQL BENCHMARKS
■ AI-SUMMARIZED FROM 1 SOURCE ▸ TIMELINE
Google Research's Gemini-SQL2 achieves 80.04% accuracy on the BIRD benchmark, significantly outperforming competitors from OpenAI and Anthropic. The system converts natural language queries into executable SQL code.
■ MORE FROM THE AI DESK
Anthropic's Claude AI model generated a playable browser game called Shepherd's Dog, sparking discussion about AI capabilities and risks in the developer community.
A pioneering UK research facility is deploying artificial intelligence to map how different types of video content affect children's developing brains, moving beyond generic screen time guidelines.
A developer leveraged Google's Gemini AI to create a functional application for yard maintenance in under four minutes, though the AI-generated code still required human intervention to resolve bugs.
A KPMG report touting the benefits of artificial intelligence contained numerous AI hallucinations, according to an investigation. The discovery highlights the irony of promoting AI while relying on flawed AI-generated content.