This was the question I asked myself when I first heard that this was technically possible. So, I had to try it out. I simply provided a link to something I know very well: my work, and watched the results emerge. Here are two podcasts generated by NotebookLM:
The first is quite serious and is based on a forthcoming paper of mine about Machine Interpreting: history, developments, etc. The second is more informal and is based on the content of this blog. In both cases, two AI hosts discuss the topic, and the quality is utterly amazing. Sure, the podcast about my blog is a bit too generous with me, but I can’t deny that some flattery is welcome. It’s interesting that the AI hosts didn’t fully understand some of my arguments, but to be honest, this also happens with several human commentators of my post, also expert ones. So, in a way, it feels even more real. The level of creativity in the examples is stunning, they are not part of the chapter. As stunning is the naturalness of the voices. Super engaging.
The reality is that the results are spectacular—an incredible technological achievement. As I keep saying, we are just at the very beginning of this curve.
Podcast about History and the Technology of Machine Interpreting
Podcast about Machine Interpreting based on my articles in this blog.
What do you think about them?
