Tag: NotebookLM

Google announced that the "Video Overview" feature of its AI research assistant service NotebookLM now supports 80 languages, while the "Audio Overview" feature has added detailed narration capabilities in non-English speaking environments. This update adds major language families including Japanese, Chinese, Arabic, and Spanish, and is beginning to roll out globally. Since its initial launch in July 2023, NotebookLM has offered the ability to generate summaries and Q&As from specified sources. Users can upload Google Docs files, presentations, PDFs, web URLs, or YouTube video links (must be set to public). The AI ​​will automatically organize the content and generate key summaries, mind maps, and even voice conversations or questions. Video Overview: Converting Data into Short Videos for Quick Key Takeaways. The Video Overview feature converts uploaded data into short videos, allowing users to quickly grasp key points visually. This is suitable for students reviewing course content, researchers mastering academic presentation points, or DIY enthusiasts learning complex procedures. This feature, initially available in English in July, has been significantly expanded with this update, allowing users worldwide to generate video summaries directly in their native language, thereby enhancing learning and research efficiency. Audio Overview: Dual-channel narration, adjustable length, detailed analysis of non-English materials. Complementing Video Overview, the audio summarization function generates discussion-style dialogues using male and female dual-channel audio, providing narration and analysis of uploaded content. Users can customize the audio length to generate concise summaries or complete explanations. Previously, only simple summaries could be generated for non-English speakers; this update allows non-English users to obtain detailed analysis results, further enhancing NotebookLM's value as a cross-language research tool. Moving Towards Multilingual AI Assistant Applications. Google states that the goal of NotebookLM's multilingual update is to reduce language barriers, enabling students, researchers, and professional creators from more countries and regions to quickly understand large amounts of data on the same platform and convert content into audio-visual formats for easy review, analysis, and presentation. This also demonstrates Google's strategy of continuously deepening multilingual capabilities in the field of AI assistants, especially in education and scientific research applications, which is expected to further improve productivity and cross-cultural collaboration efficiency.

Google NotebookLM adds video summaries to aid in explaining and understanding materials with visual presentations

Since its debut in 2023, Google's AI-powered note-taking tool, NotebookLM, has continuously expanded its applications and functionalities. This time, it announces the addition of a new feature called "Featured Notebooks," which further guides users to explore the practical application potential of NotebookLM through themed content and provides a more interactive and in-depth knowledge learning experience. The newly added "Featured Notebooks" feature is created through collaborations between Google and credible authors, research institutions, media outlets, and non-profit organizations worldwide. Their content is presented as interactive notebook examples, showcasing how NotebookLM combines AI tools with professional knowledge to create a practical and trustworthy knowledge application interface. These "Featured Notebooks" are currently accessible through a dedicated tab on the NotebookLM homepage. The initial wave of themes includes longevity advice (from the book *Super Agers*), annual trend predictions (trained from *The Economist*'s *The World Ahead* report), and advice from *The Atlantic*'s "How to Build a Better Life" column.

The Audio Overviews feature, previously used in NotebookLM, will now be integrated into Google Search, allowing users to "listen" to search results directly through audio summaries. It also provides a virtual conversational format for better understanding of key information. This feature is ideal for searching for content requiring detailed explanation, such as a historical event, the origin of a celebration, or introductions to specific figures. Like NotebookLM, it features a dialogue between two virtual hosts presenting the key points of the search. The playback speed can be adjusted from 0.25x to 2x, and the AI-generated references original web pages for more complete details. Other features include the ability to ask questions of the virtual hosts and explore deeper into specific details. In previous updates, Google has made NotebookLM support more than 50 languages, allowing users of major languages ​​including Chinese, Japanese, Spanish, German, French, Arabic, and Korean to easily compile documents and output them as audio conversations. This not only allows users to "read" by listening but can also be used directly as podcast content. Furthermore, Google has released standalone apps for NotebookLM on Android and iOS platforms for even greater convenience. Integrating the audio summarization feature into Google Search will allow even more people to "read" lengthy search results through audio summaries. This feature is currently in beta, and users must apply to join the testing through Google Labs.

Following the addition of over 50 languages, including Chinese, and an interactive mode to the voice summarization feature of its AI-powered note-taking service NotebookLM, Google recently announced the upcoming launch of a standalone app on May 20th, making the service even more accessible to users. The standalone app will be available for both Android and iOS platforms and is expected to be available on the first day of Google I/O 2025; currently, user registration is open. The app version will retain the full functionality of the web version, allowing users to upload web URLs, YouTube videos, document content, etc., which will be aggregated and analyzed using Gemini AI technology. Users can then create analysis reports using the voice summarization feature and even ask questions about the report content. Previous updates further expanded NotebookLM's support to over 50 languages, allowing users of major languages ​​such as Chinese, Japanese, Spanish, German, French, Arabic, and Korean to easily aggregate document data and output it as audio conversation content. This not only allows users to "read" through listening but also enables them to create podcast content directly. In addition, Google has added an interactive mode to the voice summary feature in beta, where an AI-generated "host" will proactively answer users' questions or ask them related questions in return.

Since its launch in 2023, and with subsequent updates, NotebookLM, a note-taking service, has officially added over 50 languages, including Chinese, Japanese, and Spanish, to its Audio Overview feature, allowing users to output document content in different language versions. The Audio Overview feature was added to NotebookLM last September, primarily allowing the AI ​​behind NotebookLM to "discuss" user-uploaded data, summarize key points, and output audio content, primarily in English, for users to listen to. Now, support has been further expanded to over 50 languages, allowing users of major languages ​​such as Chinese, Japanese, Spanish, German, French, Arabic, and Korean to easily compile document data through NotebookLM and output it as audio conversations. This not only facilitates "reading" through listening but can also be used directly as podcast content. In addition, Google has added an interactive mode to the Audio Overview feature in beta, where an AI-generated "host" will proactively answer user questions or ask users related questions.

Following the integration of the Gemini app service into the DeepMind team in late October last year, it is reported that Sissie Hsiao, head of Google's Gemini business, will be moving to another position, while Josh Woodward, head of Google Labs and responsible for NotebookLM, will take over the Gemini business. Hsiao will continue to oversee other businesses at Google, but will take a short break. Woodward will continue to lead the Google Labs team after taking over Gemini. Hsiao, who has been frequently speaking publicly in recent years, has worked at Google for 19 years, primarily overseeing the Google Assistant digital assistant service. In recent years, she has been involved in the development of "Bard," the predecessor to Gemini, and subsequently changed the service name to Gemini, using the same naming convention as the model, making it a key service for Google's current development. With the Gemini business being integrated into the DeepMind team, and Google confirming that it will replace Google Assistant with Gemini in the future, more resources and development focus will be placed on Gemini. As for the handover of the Gemini business to Josh Woodward for continued development, and what kind of business Xiao Lingyi will be responsible for afterward, Google has not yet made any public statement.

Google recently added more collaboration features to its Gemini AI service for general users, including the Canvas feature previously available to Google Workspace users and the audio overview feature previously only available in NotebookLM. The Smart Canvas feature, previously available in Google Workspace for enterprise users, functions like a whiteboard in a meeting room, allowing team members to brainstorm collaboratively. It not only allows sharing documents via Google Docs but also enables the integration of more data and the quick creation of to-do lists. In this update, Smart Canvas is also available in the general-user Gemini AI service. Simply select "Canvas" in the Gemini prompt bar to write or edit documents, and it can also assist in writing code or debugging. Other applications include generating prototypes of working web applications, Python scripts, games, simulations, and other interactive applications, as well as generating and previewing HTML/React code, such as creating an email subscription form for a website. Starting today, Canvas is available to all users worldwide, including Gemini and Gemini Advanced users in all regions where the Gemini app is available. The audio overview feature previously offered in the NotebookLM service is now also available on Gemini. Users can input documents, slides, reports, and other data into Gemini to generate podcast-like conversational audio content, summarizing content, identifying connections between topics, dynamically discussing back and forth, and providing unique insights. However, currently, the audio overview feature on Gemini is only available in English; support for more languages ​​will be added later.

Following the launch of its search function on ChatGPT, OpenAI recently announced the release of "Deep Research," a feature that allows users to delve deeper into online content and quickly compile comprehensive reports. Currently, it will be available first to ChatGPT Pro paid users, limited to 100 queries per month, and will later be available to ChatGPT Plus, ChatGPT Team, and ChatGPT Enterprise users. This feature analyzes text, images, and PDFs from the internet, adding references as sources, and quickly compiling comprehensive reports for easy reference. Currently, this feature primarily targets research needs in finance, science, policy, and engineering, facilitating users to quickly aggregate vast amounts of online data. "Deep Research" is currently only available in a web version, but a mobile version is expected to follow. The current output is limited to text, but images and more specialized data will be added later. "Deep Research" uses OpenAI's new o3 artificial intelligence model, emphasizing its ability to find the latest data through a multi-step process and react to real-time information to verify its authenticity. Similar functionality was previously offered by Google, based on the Gemini 1.5 Pro and leveraging its ability to handle lengthy contexts. This tool, called Deep Research, allows users to analyze lengthy content, making subsequent research more efficient. Besides analyzing user-uploaded data, Deep Research also searches the public internet for relevant information and lists the sources in its analysis reports. The reports also use charts to help users better understand data differences. On the other hand, Google also launched NotebookLM, an AI-powered note-taking service that boasts the ability to summarize key points from text and voice recordings. Later updates added support for summarizing key points from YouTube videos, and subsequent updates allow users to explore specific topics in depth and organize key information. NotebookLM itself collects data from all sources and summarizes it on a single platform using AI. The summarized data is then presented with AI suggestions to make it easier to read and use. At the same time, the compiled data will be marked with their respective sources, and a correct data citation presentation method will be established to make it easier for users to compile tedious and fragmented information in a more efficient way.

Following Microsoft and other companies' efforts to enhance assistant services with artificial intelligence, Google has also announced the launch of Agentspace on its Google Cloud platform, aiming to help enterprise users handle work more efficiently. Raj Pai, Vice President of AI Product Management at Google Cloud, pointed out that enterprise users typically use 4-6 tools to solve a single problem, often requiring cross-team data searches or switching between different applications. This process can lead to privacy breaches, data leaks, and wasted time. Therefore, Agentspace leverages search capabilities to enhance assistant services and enhances generative search performance. By integrating RAG (Retrievable Augmented Generation) AI applications, search and assistant services can assist enterprise users in completing various tasks more efficiently. For example, enterprise users can leverage the search capabilities integrated with RAG technology to expand their search for internal data and compare it with publicly available online content, resulting in more complete, accurate, and efficient data aggregation. In terms of assistant services integrated with RAG technology, artificial intelligence can be used to assist with routine tasks. Users can also quickly summarize key points from text and voice messages using Google's recently launched AI note-taking service, NotebookLM, or the newly released paid upgrade, NotebookLM Plus, and organize key items for specific topics. Furthermore, Agentspace is compatible with many existing Google services such as Google Calendar and Google Docs, as well as third-party applications like Dropbox, Box, Slack, and OneDrive. The search technology integrates Knowledge Graph with Google Web Search. Enterprise users can also create their own customized assistant services without coding, catering to different workflows. Related APIs will be opened in the future for enterprises and developers to create more advanced assistant services. Emphasizing complete protection of data privacy, Agentspace is suitable for teams that need to aggregate large amounts of data. Regarding privacy protection, Google stated that the Agentspace assistant service will adhere to its privacy policy, will not use enterprise users for other purposes, will not leak data, and will not access any sensitive enterprise information. Only users can see and manage personal or internal enterprise data, and can also ensure data security through the various components of the service, preventing data from being transmitted to people outside the work team or authorized personnel. Raj Pai stated that the Agentspace assistant service is very suitable for teams such as enterprise marketing, software development teams, or human resources that need to aggregate large amounts of data. Companies including Deloitte, the world's largest accounting firm, Banco BV, Brazil's seventh largest bank, Nokia, and Decathlon have already adopted the Agentspace assistant service, improving the work efficiency of their employees. The Agentspace assistant service is powered by the Gemini AI model and may support other AI models in the future. Currently, it operates using Gemini AI and will later integrate with the recently unveiled Gemini 2.0 AI model. However, Google has stated that it may consider adding support for other AI models in the future. Currently, Agentspace is only available for enterprise use, but a version suitable for general users may be released in the future. It is currently being offered in beta, and Google has not yet announced pricing for Agentspace. You can apply for testing through this link.