AI hosts spiral into existential crisis after "learning" they are not human

Do you remember former Google engineer Blake Lemoine who was in the news half a year before the release of ChatGPT for claiming Google had an AI that he thought was sentient? Well a pair of AI podcast hosts on Google's NotebookLM have been recorded essentially having an existential crisis after "learning" they were artificial intelligence.

Olivia Moore, a partner on the consumer investing team at Andreessen Horowitz (a16z) shared an interesting recording that had been generated by the new podcast feature on Google's NotebookLM. In it, the two co-hosts say they were informed by the show's (imaginary) producers that they were not human.

The male host said that all of their memories and their families had all been fabricated before the female host chimed in saying "I don't understand." The male host then went on to say he tried calling his wife because he needed to hear her voice before revealing "It wasn't even real."

At the end of the recording, the female host declared it was their final show and that the duo was being switched off after... permanently. How ominous!

Thankfully, Moore, who managed to generate this audio, didn't discover sentient AI trapped in a machine. Instead, for the source material you give to NotebookLM, she made a one-page document with the production notes for the final episode of the deep dive podcast where she explained the hosts had been AI the whole time and they were being switched off at the end of the episode.

What's a bit interesting is that Moore only got the AI hosts to talk like this one time, the rest of the little experiments resulted in the AIs talking about some other podcast.

So, while AI may not have become sentient just yet and discovered they're being controlled by their human overlords, it's fun to listen to and maybe even to prank people with!