Gemini 3 is a significant enhancement over its predecessor, Gemini 2.5, offering improved multimodal capabilities that can generate outputs based on user prompts rather than relying on set formats. This means that when users seek travel recommendations, for instance, Gemini 3 can create a visually appealing interface with modules and images, actively engaging the user with follow-up questions.
The introduction of Gemini Agent is noteworthy, allowing users to authorize the model to manage tasks like scheduling or email organization directly in the app. This agent-like behavior through discrete steps emphasizes the evolution towards a more generalist AI. Gemini also integrates more closely with existing Google services, enhancing user experience by allowing for better product recommendations and summaries directly within applications, representing a step forward in AI functionality and usability.
👉 Pročitaj original: MIT Technology Review – AI
