Google has launched a new feature, Audio Overview, in its AI note-taking application NotebookLM, allowing users to convert their research into AI-generated podcasts. This feature employs two AI hosts to review and discuss the provided material, offering summaries, connections between topics, and a conversational dialogue about the research.

NotebookLM, which already uses Google’s Gemini AI model for summarising research, now extends this capability into audio format. Users can access the Audio Overview feature by opening a notebook in NotebookLM and selecting the option under the Notebook guide. The AI-generated podcasts offer a fresh way to engage with research by transforming notes into audio discussions.

During a test by The Verge, the feature created a podcast discussing the invention of the lightbulb, featuring a dialogue about Thomas Edison and teamwork. While the AI hosts maintained a conversational tone, the dialogue occasionally felt unnatural, highlighting one of the limitations of the tool.

Google has noted that Audio Overview is not designed to provide a comprehensive or objective view but rather reflects the user’s notes. The feature is currently only available in English and may take several minutes to generate a podcast. Additionally, the AI’s tone may not be suitable for all topics, such as sensitive issues like cancer or war, where maintaining a lighthearted approach could be inappropriate.

While Audio Overview offers a novel way to interact with research, users should be mindful of its limitations, including occasional filler content and varying accuracy. For those interested in trying the feature, it can be accessed in NotebookLM by selecting the Notebook guide and loading Audio Overview.