Google is enhancing its NotebookLM writing assistant by boosting its performance and introducing new features. It now operates on the company's Gemini 1.5 Pro model, the same AI behind the Gemini Advanced chatbot. In addition, advanced fact-checking capabilities have been added, making it a more powerful tool for users globally.
NotebookLM Goes Global With New Features
Google's NotebookLM, an AI-driven notes application, has recently enhanced its features to include integration with Slides and improved fact-checking tools. This update represents a significant advancement in making the tool more accessible and functional for a global audience.
Originally launched as an experiment under Google Labs, NotebookLM is designed to leverage language models to boost productivity for users working with large data sets. Acting as a virtual research assistant, it offers tools to summarize information, clarify complex concepts, and generate new ideas from existing content.
NotebookLM sets itself apart by enabling users to "ground" the AI in their specific documents, primarily Google Docs. This grounding customizes the AI’s responses to be relevant to the user's content, personalizing the experience. Users can ask in-depth questions, get document summaries, and use the platform for creative brainstorming.
Google's NotebookLM Is a Better Fact-Checker
With its global expansion and new features, NotebookLM now includes sophisticated fact-checking tools to ensure the accuracy and reliability of information processed through the system. This tool assists users in verifying data and claims within their documents, highlighting the importance of credibility in information handling.
NotebookLM expansion to advance fact-checking tools not only increases its usability but also emphasizes Google’s commitment to responsible AI development.
Given the ongoing issue of AI hallucinations, it is crucial for users to be able to fact-check outputs. For images, opening the citation allows the source picture to appear in a small window next to the text.
Moreover, Inline citations can point to your sources to verify AI responses, supporting up to 50 sources per “notebook” or project, with each source accommodating up to 500,000 words. Previously, users could only upload five sources.
Experiment Says the System Remains Imperfect
During the Gizmodo website experts' trial of the NotebookLM in December 2023, the reviewer found it disappointing. The tool frequently generated inaccurate and fabricated information. For instance, when the reviewer uploaded a story and supplementary documents, NotebookLM misunderstood significant parts of the plot and falsely claimed the reviewer had graduated from UC Berkeley and was a member of the Society of Children’s Book Writers and Illustrators.
Although more information can improve its performance, the system remains imperfect, according to the website. It provides citations for its claims, but the highlighted text from original documents often does not correlate with the AI-generated content, leading to misinterpretations.
Read More
AI Tool Predicts Real-World Impact of Online Health Misinformation
AI-Operated Accounts Promote the Israeli Narrative During the Gaza War
