Google calls NotebookLM an “AI-powered research and writing assistant”. It’s more than that. It “helps users understand complex information by instantly becoming an expert on uploaded sources.”
You can upload various types of files to NotebookLM, give it an ask, and watch it “generate” …
Focused on AI, I wanted NotebookLM to intersect the big thinking of 1) noted author and historian, Yuval Harari, 2) author and entrepreneur, Max Bennett, and 3) the perspectives of Colin Rule, a leader in the field of Online Dispute Resolution.
Source materials I provided to Notebook LM
I provided NotebookLM with links to:
- An interview of Yuval Harari, about his new book, Nexus, focused on the history of information networks. The 2-hour interview (I first heard it in podcast format) took place on Dax Shepherd’s Armchair Expert. Its available as a podcast or video.
- A presentation on AI and Mediation: Working with the Fourth Party This presentation was prerecorded for the upcoming Association of Conflict Resolution conference. Refer to my previous blog post for more about that presentation.
- A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made Our Brains, Max Bennett’s book.
I chose those files as I had read Bennett’s book, and watched and listened to the entirety of the videos; both the interview and the presentation.
My specific ask of NotebookLM
I asked Notebook LM to:
“create a 5-minute podcast based on the intersection of 1) the Yuval Harari interview, 2) Colin Rule’s AI and Dispute Resolution presentation, and 3) Max Bennett’s book about evolution, the brain and AI – include elements of Bennett’s book about what’s still missing in AI based on how our brain works.”
NotebookLM’s output
After I hit ‘generate’, within two minutes, NotebookLM provided me with an 11-minute audio overview (yes, I know, they went overtime on my ask), in podcast format; e.g., with two hosts discussing the source material provided. The hosts sounded human, very human, with human mannerisms. In reality, the hosts were computer-generated.
Here’s the generated 11-minute audio. Tell me it isn’t amazing, and scary.
NotebookLM will also generate, if asked, based on the sources provided to it, output: a FAQ, Study guide, Table of contents, Timeline, and Briefing doc. In my case, I made the asks and it generated all of those outputs in 2 minutes, total.
In my assessment, the NotebookLM AI hosts/experts provided a logical, and fair, representation of the source materials, and how they intersect. My one gripe is that it didn’t incorporate Bennett’s work enough. Maybe a different query wording is in order?
As their ending/bottom line, the AI hosts keyed on trust – how can we trust the AI algorithms?
While I grasped the importance of trust when digesting each of the information sources individually, I did not fully appreciate why trust, and associated control mechanisms to ensure trust, must be at the core of future human-centred AI development. I have thank NotebookLM for the nudge.
What next?
Good, innovative ideas build on existing knowledge and ideas.
In a complex world, NotebookLM can help you make sense of the now, and find those good, new ideas. For example, in the field of mediation, good ideas, in the form of solution options, are gold.
No matter your field of endeavour, onwards, with AI, and trust.
Speak Your Mind