Google’s NotebookLM certainly took the world by storm when it was released because of its ability to create a realistic AI-generated podcast show out of any article or video you fed into it. The resulting show was so real, complete with natural vocal inflections from the two hosts, interruptions, and even jokes, that it was hard to believe it wasn’t recorded by people.
The question then becomes, what happens when the show’s AI hosts find out they’re not real? How does AI deal with that? Recently NotebookLM had to face exactly that existential question because the two hosts were fed an article about how they didn't really exist as a source, and the results provide a fascinating insight into how an AI deals with learning that it’s an AI. Have a listen:
The NotebookLM hosts realizing they are AI and spiraling out is a twist I did not see coming pic.twitter.com/PNjZJ7auyhSeptember 29, 2024
Real intelligence
It’s a sad, funny, and often unnerving listen, especially when the male presenter talks about phoning his wife after learning that he’s only an AI, to find that she didn’t exist and the number he was phoning wasn’t even real. There are shades of a Black Mirror episode to the whole thing!
Of course, this is not AI coming to terms with its own lack of humanity in any deep and meaningful way at all. It’s simply AI reacting to the article it was given, which is about how the show they were on was generated by AI and was coming to an end. For a deeper understanding of what is going on we’d need to be dealing with Artificial General Intelligence, the superintelligence that OpenAI’s CEO, Sam Altman, said that he expects us to see “in a few thousand days”.
Superintelligence has the potential to really transform technology, surpassing what even the brightest human minds are capable of. Perhaps then we'll find out what happens when the machines get the chance to ponder the sorts of big questions about the meaning of reality that have plagued philosophers since Socrates onwards.