Google’s Gemini Live aims to be more than just a chatbot; it aspires to be a true companion. Recent updates to Gemini, particularly the Gemini 2.0 Flash model (now freely available via the Gemini app), enhance this vision. The chatbot now mimics the experience of a phone call, replacing the “End Live Mode” notification with options to “Hang Up” or put the AI on “Hold”. From the lock screen, users now see “Live with Gemini” and a reassuring “Listening” message. This seemingly minor tweak underscores Google’s strategic direction for AI in 2025.
Expanding Gemini’s Capabilities
Beyond conversation, Gemini Live is evolving to understand uploaded photos and videos. Future integration with Google DeepMind’s Project Astra promises even more advanced visual capabilities. These additions necessitate seamless background operation, minimizing disruption to the user’s typical phone usage.
Google is also broadening its Gemini 2.0 model lineup, introducing various sizes to cater to diverse needs. An experimental “Gemini 2.0 Pro,” touted as the most potent user-facing model yet, is primarily targeted at coders and programmers, accessible through Gemini Advanced. This raises questions about its performance relative to its claims. Google asserts its superiority over Gemini 2.0 Flash in most benchmarks, with the exception of providing “factually correct responses given documents and diverse user requests.” Conversely, Gemini 2.0 Flash-Lite reportedly matches Gemini 1.5 Flash in power requirements while boasting significantly improved accuracy. These new models follow OpenAI’s recent unveiling of its o3 reasoning model and the miniature o3-mini.
Mobile Integration: A Work in Progress
The Samsung Galaxy S25 launched with cross-app Gemini integration as a flagship feature. A long press of the power button enables hands-free actions like converting text messages into calendar invites. However, practical tests revealed these AI functionalities to be rather basic, struggling with more intricate tasks. If constant oversight is needed, the efficiency gain becomes negligible.
Looking Ahead to Google I/O 2025
Google is strategically reserving its most advanced mobile AI features for later in 2025, likely coinciding with Google I/O and the anticipated Pixel 10 release. In contrast to the somewhat underwhelming performance on Samsung devices, Google aims to impress with its AI capabilities. Whether the hype translates into a truly transformative user experience remains to be seen.
The Future of AI Companionship
While Gemini Live can engage in conversations and process multimedia content, its current conversational abilities might not be considered truly stimulating. Nevertheless, Google’s ongoing efforts suggest a strong commitment to realizing its vision of an AI companion. The shift in notification style, from a simple app notification to a simulated phone call, subtly reinforces this ambition. The future iterations of Gemini, with enhanced visual understanding and deeper integration into mobile workflows, will determine whether this vision truly materializes.
