By Lisbeth Knudsen
Imagine being able to produce an engaging and smooth-flowing podcast, just over 18 minutes long, with two articulate American hosts discussing a complex topic like the Algorithms, Data, and Democracy-project. And doing it in just 5 minutes.
That’s exactly what happened the other day when I curiously tried out Google’s NotebookLM. With two artificially generated and very enthusiastic American hosts, they curiously and insightfully discussed the ADD-project, its structure and goals, its significance for society, and the project’s research and debate-creating initiatives. All of this was done with an inviting but not overly impressed tone of admiration, respect, and fascination around the project.
I fed NotebookLM with the strategies and goals of the ADD-project, and from there, the hosts unfolded a conversation about both the benefits and concerns of using AI. Not a single suggestion for replies or responses from me. Only cold facts. The artificial hosts themselves delved into how algorithms affect society, how data is collected and used, and what that means for democracy, healthcare, the workplace, and much more. Based on the inputted material, they explored more specific mentions of several of our research projects and explained some of our outreach activities, from large population surveys to policy labs.
Notebook LM from Google is designed to help organize, analyze, and present complex information. The solution is built on the same principles as Google’s large language models (LM). But instead of just generating random pieces of text or answering questions, NotebookLM is designed to understand, analyze, and create content that is both targeted and insightful.
Notebook LM is developed for researchers, academics, and content creators who want to take their data and research results and transform them into something communicable and easy to understand – such as an article, a presentation, or, in my case, a podcast.
As the name suggests, it’s a kind of notebook on steroids, where it can analyze the content you input and from there generate deep insights. The tool is a gamechanger because it goes far beyond merely organizing information: it can actively create narratives, select relevant topics, and even facilitate meaningful conversations.
The idea of letting two AI hosts lead the conversation in the podcast stems from the desire to create a dynamic and engaging communication of complex material. Instead of presenting the information as a one-sided monologue or traditional reading, the freer conversational form becomes a way to make the content lively, easier to understand, and more relatable to listeners. It is a method that leverages AI’s ability to generate human-like dialogue, while also introducing a multi-voice approach where different perspectives and questions can be explored, and where the tool itself invents the dialogue.
By transferring the conversation to artificial AI hosts, the technology can simulate a discussion where the hosts not only convey facts but also reflect, ask questions, and counter each other’s points of view. By using AI hosts instead of humans, you also have the freedom to control how the conversation unfolds and which topics are addressed – all based on the material you have fed them. In the case of the ADD project, this meant that the hosts could delve into both the benefits of AI and the concerns associated with it, bringing a balanced discussion to the listener.
The entire process – from giving NotebookLM input to hearing the hosts’ conversation – was impressive, and the quality of the result was surprisingly good. Think about it: In just 5 minutes, a podcast that could fool even the sharpest ears into thinking it was created by experienced humans was produced. I’m just reflecting on where this takes the field I was once trained in.
It shows the power of AI and simultaneously raises big questions about the future of media, knowledge dissemination, and human creativity. We are truly in an era where technology is not just assisting us, but in some cases taking over the entire production process – and doing so in a way that leaves one amazed at the outcome.
Listen here: