Disturbing is annoying. Apparently, even AI-powered podcast hosts agree.
Or so Google NotebookLM users discovered. NotebookLM was introduced last year. The virus stopped. For its feature that creates AI-generated podcast-like discussions from content users discuss with chatty AI bots that act like podcast hosts. In December 2024, NotebookLM launched a new feature called “Interactive Mode”. Allows the user to “call in” to the podcast and ask questions.Mainly interrupting AI hosts while talking.
When the feature was first introduced, AI hosts seemed annoyed by such interruptions. They occasionally make comments to human callers like “I'm here” or “Just what I was saying,” and are “extraordinarily adversarial,” Josh Woodward, VP of Google Labs, explained to TechCrunch.
So NotebookLM's team decided that some “friendliness tuning” was in order, and posted a self-deprecating joke about it on the product's official X account:
Woodward said the group solved the problem in part by learning how their members could respond to interruptions more politely.
“We tested a lot of different cues, studied how people in the group responded to interruptions, and arrived at a new cue that we thought was more friendly and engaging,” he said.
It is not clear why the problem occurs. Human podcast hosts sometimes show frustration when interrupted; This can culminate in the system's training data. A person familiar with the matter said that this is most likely due to the design of the system's stimuli, not the training data.
Regardless, the fix seems to work. When TechCrunch tried out the Interactive Mode, the AI host didn't seem annoyed, but said “Woah!” He shouted and was surprised. Before inviting the person.