- “As we start the year, I’ve been reviewing demos of the products and features we’ll be introducing over the next few months. The progress has been amazing, and I’m confident we’ll continue this momentum into 2025. Will continue,” Mr Pichai wrote.
- In December, Google made several major announcements in AI, hardware, and extended reality. It introduced Gemini 2.0, the next generation of its AI models, designed for the “agent era” that brings multimodality and tool usage. The purpose of the model is to better understand the world, think ahead, and take action with user supervision.
- Google has launched Deep Research, a new feature in Gemini Advanced that uses advanced reasoning and long-context capabilities to act as a research assistant. It also released an experimental version of Gemini 2.0 Flash, a model with enhanced performance and speed, available to developers through the Gemini API in Google AI Studio and Vertex AI.
- On the hardware front, Google announced the general availability of Trillium, its sixth-generation tensor processing unit (TPU). Trillium is designed for AI workloads, delivering significant improvements in training performance, inference throughput, and energy efficiency compared to previous generations. It was used to train Gemini 2.0 and is now available to Google Cloud customers.
- Google also unveiled Willow, a new quantum chip that has demonstrated the error correction and performance that could lead to building a useful, large-scale quantum computer. Willow has made a major breakthrough in quantum error correction and performed a calculation that would take a supercomputer 10 septillion years in less than five minutes.
- Google also introduced Android XR, a new platform developed in partnership with Samsung and Qualcomm, designed to extend reality for headsets and Glasses. Android XR aims to bring assistant experiences to these devices with the help of AI, and it will launch on headsets first – including one code-named Project Mohan – and then support glasses later.
- Gemini will be integrated into Android XR to provide conversation support and help users control their devices. Apps like YouTube and Google Maps are also being reworked for the headset.
- The company also introduced Google AgentSpace, designed to unlock enterprise expertise by using AI agents to bring together Gemini’s logic, Google-quality search, and enterprise data.
- AgentSpace provides a single, company-branded multimodal search agent for organizations. It allows employees to complete complex tasks, interact with enterprise data through NotebookLM (a note taking and research assistant), find information across the enterprise, and automate business tasks with expert agents.
- NotebookLM received updates, including a new interface, audio interactivity, and a premium version, NotebookLM Plus. Google also released new versions of its video and image creation models, Veo 2 and Imagen 3, as well as a new tool called Whisk, which allows users to generate images using other images instead of text .
