Google Maps taps Gemini AI to transform into an 'all-knowing copilot' (www.theverge.com)

Google Maps is embedding Google’s Gemini chatbot deeper into core navigation: users can now have natural conversations about routes and nearby landmarks, ask for restaurant recommendations along a route and have the route updated conversationally, summon information by voice or tap, and even add calendar reminders or get summaries of recent emails while navigating. New audible directions use recognizable visual cues (gas stations, restaurants, landmarks) instead of only distances, and Google Lens powered by Gemini can identify businesses or landmarks through the camera. A Proactive Traffic Alerts feature monitors familiar commutes in the background and notifies drivers of crashes, construction, or closures early enough to reroute. The move turns Maps into a contextual “copilot” by combining Gemini’s language and summarization abilities with Maps’ geospatial datasets — including billions of Street View images and an index of about 250 million places — plus community reviews and web information. That grounding aims to reduce hallucinations (Google says place suggestions use actual place data) while enabling multimodal queries and app interoperability (Calendar, emails). For AI/ML practitioners, it’s a notable real-world deployment of multimodal grounding at scale, with implications for safety, UX, and latency trade-offs in live navigation; rollout is free for signed-in users on Android and iOS now, and later to Google-built vehicles.