The tech giant’s search engine will now display AI Overviews in response to queries that summarize information or offer direct answers, starting in the United States. Google’s Gemini 1.5 models can now accept input in text, image, audio and video form. It made the announcements at its I/O developer event. (The Logic)

Talking point: Google’s updated AI models have a longer context window, meaning they can consider more chunks of information at once—hundreds of pages of text or many minutes of video. That makes them better at tasks like summarization and analysis. The AI Overviews update a core mechanism in Google’s search feature by limiting the need to click links. That could be a concern for publishers selling ads on linked pages, but Google insists users actually click more this way. At I/O, the firm also showed off Project Astra, a mobile and wearable assistant, or “agent,” that responds to verbal and visual cues. It all comes a day after OpenAI unveiled similar multimodal advancement.