IT Home reported on October 30 that in the early hours of this morning, Beijing time, Google CEO Sundar Pichai revealed the latest progress of the "ProjectAstra" plan in the company's third-quarter earnings call. He said Google is building AI experiences that recognize and reason about the user's surroundings, and "ProjectAstra represents the first exploration of this future." We're working to launch a similar experience as early as 2025. "
Note: ProjectAstra is a suite of new technologies showcased by Google at the I/O developer conference in May 2024, including a smartphone app that recognizes the environment around and answers questions, and an AI assistant that can perform actions on behalf of users. The project is based on Gemini and can run natively on Pixel phones, which can be said to be the latest model to benchmark OpenAIGPT-4o.
According to Google's official presentation, ProjectAstra is able to answer questions about objects in the smartphone's camera's field of view, such as the user's neighborhood or the name of the damaged bike part.
This means that Google won't roll out the core technology in ProjectAstra's program until at least next year. The project is a broad attempt to develop AI applications and "intelligent agents" with real-time, multimodal understanding.
According to TheInformation's report this month, Google plans to launch a consumer-facing "smart agent" service as early as December this year, with features such as product purchases and flight bookings. However, unless the experience is not related to "ProjectAstra", this goal seems unlikely to be achieved.