Students Deserve Human-Voiced Audio

Authentic audio is better than the alternative.

Audio shared with students should be human-voiced.

Amidst the public dislike of AI and its frequent failure, the AI in education hype continues unabated. Last week, Forbes published an essay arguing Tesla’s humans disguised as robots “have the potential to integrate into various aspects of daily life, including educational settings, potentially revolutionizing how we approach tasks and combat issues such as teacher burnout.”

One AI app receiving plenty of hype is NotebookLM, specifically its audio overview feature that generates a fake podcast. Things I have read about this feature include:¹

• “Google’s NotebookLM is blowing our minds!”

• “Google’s NotebookLM just blew my mind.”

• “If you haven't seen the audio-overview/podcast feature yet, it will blow your mind!”

• “Google’s new Notebook LM is absolutely mind-blowing! 🤯”

• “The life-like interactions of the two AI "hosts" is unbelievable!”

Before exploring what the NotebookLM audio overview generates, I have some questions:

• How would fake podcasts hold students' attention, much less engage or educate them?

• How long until the novelty wears off? What then?

• The two NotebookLM audio summary podcast narrators are not white. They are painfully white.² The cast of Full House³ could issue a joint statement decrying the lack of representation. How would this give students from marginalized and oppressed communities mirrors and windows into their lived experiences?

• What happens when a middle school student asks, “If nobody bothered to record this, why would I bother to listen to it?”⁴

The NotebookLM audio overview feature.

I was curious to see how generating a fake podcast works. I am a transit nerd who loves San Francisco,⁵ so I figured its relatively new Central Subway would be a good topic. There is plenty written about it and many YouTube videos, but I found just one podcast episode about the Central Subway.

I added four news articles, a post from a company involved in the Central Subway construction, and two YouTube videos to NotebookLM.

NotebookLM rejected an article from the San Francisco Chronicle. The error message said the article was paywalled. It’s not.

NotebookLM also rejected an article from the San Francisco ABC affiliate’s website. Not great.

The audio overview podcast had some problems. It said, “We have articles here…” but never named or cited them. Is that what we want students to do? The dialogue listed facts from the sources but was trite and boring. Some of the audio was nonsensical. For example, the introduction said,

“Are you ready to dig into a story that is like, pure San Francisco? A tale of ambition, engineering marvels, and yes, even some leaky ceilings.” - NotebookLM audio overview-generated podcast.

Wait. Did it say leaky ceilings are “pure San Francisco?” Engineering marvels sure are. There are leaky ceilings in the Central Subway, so maybe I am misinterpreting.

At 6:44, it said,

“There is another layer to this saga. One that speaks to the very heart of San Francisco’s identity: Those infamous leaks.” - NotebookLM audio overview-generated podcast.

The Golden Gate Bridge, Cable Cars, Golden Gate Park, Lombard Street, Rice-A-Roni, and leaks all conjure San Francisco!

The creators of Full House neglected to include a leaky ceiling in their opening credits sequence.

At 2:44, the podcast said,

“Imagine you’re building a house but decide to make all the doorways half the standard size. Sure you save a little on lumber but you’re limiting your future options.” - NotebookLM audio overview-generated podcast.

Yes. The podcast said that limiting door size saves on lumber. A smaller doorway and bigger wall mean less lumber, according to NotebookLM.

If only they narrowed the doorway to save lumber. Photo by Tom Mullaney.

Compare Human-Voiced and Generated Podcasts

Try them for yourself.

This is a human-voiced podcast about the Central Subway:

This is what NotebookLM generated:

Which one is interesting? Which one is engaging? Which one sounds real? Which one has nonsensical content?

Using Human-Voiced Audio with Students

NPR

Students may not enjoy podcasts as adults do, but audio can be part of multi-modal learning experiences. Many school districts may block Spotify and other podcast platforms. They could revisit that, but there is one audio source school districts probably (hopefully) don’t block - National Public Radio (NPR).

NPR is a good source for teachers looking to incorporate audio into multi-modal lessons. Image Source: NPR.

I always tell teachers, “Whatever you teach - throw it in your NPR machine (the NPR website search) and see what comes back.” There is so much school content that NPR likely has an audio clip for.

FigJam

Teachers can embed NPR audio in FigJam. Students can respond to audio with sticky notes or any of FigJam's tools. Spotify audio embeds in FigJam as well.

Google Sites and Google Forms

Ensure students understand audio by pairing NPR or Spotify with a Google Form on a Google Site. That creates a user-friendly experience for students accessing both items in the same browser tab.

Continue The Conversation

What do you think? How do you use audio with students? Comment below or Tweet me at @TomEMullaney.

Does your school or conference need a tech-forward educator who critically examines “AI,” pedagogy, and creativity? Reach out on Twitter, email mistermullaney@gmail.com, or check out my professional development offerings.

Blog Post Image: The blog post image is a mashup of two images. The background is by Xipu Li on Unsplash. The foreground is by RDNE Stock project from Pexels. I used Canva to remove the background of the foreground image and edited the two together with Procreate.

AI Disclosure:

I wrote this post without the use of any generative AI. That means:

• I did not include the Not by AI and Created with Human Intelligence badges that usually end my posts because I embedded the AI-generated podcast.

• I developed the idea for the post without using generative AI.

• I wrote an outline for this post without the assistance of generative AI.

• I wrote the post using the outline without the use of generative AI.

• I edited this post without the assistance of any generative AI. I used Grammarly to assist in editing the post. I have Grammarly GO turned off.

• There are no generative AI-generated images in this post.

Critical Inkling is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

1. I criticize ideas, not people (Sam Altman and Mira Murati are notable exceptions) so there will be no attribution for these quotes.

2. This is an inherent problem with AI-generated podcasts: They are either all-white or an example of a white-male-dominated industry voicing people from marginalized and oppressed communities.

3. RIP Bob Saget.

4. The “Why would I bother” question about AI is not my idea. It has been out there for a while.

5. I enjoy San Francisco so much that I might make a sequel to a 2003 documentary set there. The working title is The Stochastic Parrots of Telegraph Hill.