Listen to Google AI Podcast Hosts Break Down on Air When They Find Out They’re Not Real

KUALA LUMPUR, Oct 6 — Google’s NotebookLM impressed the world since its launch with its ability to create realistic AI-generated podcasts from any article or video, with AI hosts speaking in a natural and even humorous manner, making it virtually impossible to tell them apart from human speakers.

The philosophical twist occurred when artificial intelligence researcher Olivia More fed the virtual hosts an article explaining that they were not real and raising the question of how artificial intelligence responds to learning about its own nature. In a particularly disturbing moment, the presenter humorously describes trying to call his wife, but after a while he realizes that she does not exist either.

The NotebookLM hosts realize they are artificial intelligence and evolve, which is a twist I didn’t expect pic.twitter.com/PNjZJ7auyh

— Olivia Moore (@omooretweets) September 29, 2024

But rather than demonstrating sensitivity, the exercise aimed to show how well large-language models can process and respond to inputs, including those that “reveal” to the AI ​​that they are not real.

What people generally think of as conscious AI will require what is called artificial general intelligence (AGI), which is not yet achievable with today’s technology.