EP 31 - Notebook LM’s ”Do It For Me” Podcast Button: A Threat to Real Educator Voices?

In this episode, John and Jason react to the new button in Google’s Notebook LM that allows users to make an instant podcast. Will it be a threat to educators' voices and hard-working podcasters everywhere? See complete notes and transcripts at www.onlinelearningpodcast.com

Join Our LinkedIn Group - *Online Learning Podcast (Also feel free to connect with John and Jason at LinkedIn too)*

Links and Resources:

• NotebookLM
• Jason’s article “Better Definitions of Distance Education”
  • The AI Deep Dive Podcast based on the article.

Theme Music: Pumped by RoccoW is licensed under an Attribution-NonCommercial License.

Transcript

We use a combination of computer-generated transcriptions and human editing. Please check with the recorded file before quoting anything. Please check with us if you have any questions or can help with any corrections!

[00:00:00] Jason: That's onlinelearningpodcast.com.

[00:00:03] John: Perfect. And that was a real human doing that.

[00:00:06] Jason: That was a real human, even though I sounded a little more. I leaned into the mic to make it sound a little more podcasty.

[00:00:12] John: Good. Well, we'll have to work on our upbeat banter between

[00:00:16] Jason: guess so.

[00:00:16] John: too.

[00:00:17] John Nash: I'm John Nash here with Jason Johnston.

[00:00:20] Jason Johnston: Hey, John. Hey, everyone. And this is Online Learning in the Second Half, the Online Learning Podcast.

[00:00:25] John Nash: Yeah, we're doing this podcast to let you in on a conversation we've been having for the last couple of years about online education. Look, online learning's had its chance to be great, and some of it is, but a lot of it still isn't. How are we going to get there, Jason?

[00:00:40] Jason Johnston: That is a great question. How about we do a podcast and talk about it?

[00:00:45] John Nash: That's a great idea. What do you want to talk about today?

[00:00:48] Jason: Nice to see you. How's Sweden?

[00:00:50] John: Sweden is good. It's a little rainy here, but now the sun is coming out. It looks like we'll have a nice weekend here. So, it's

[00:00:57] Jason: Yeah. Good.

[00:00:59] John: So, Jason, what do you think about Google's notebook the LM and what it's doing in particular with these that we'll make from content? Yeah.

[00:01:12] Jason: was wild, we've been at this at least a couple of years talking about, at least between you and I, and then like a year and a half here in the podcast, talking about AI and some of its effects on education primarily, but also just on a lot of the creative work that you and I do in different ways.

Right. And I joked with somebody that, was fine when AI came for the graphic designers because, I'm not a graphic designer and, when it came for the writers and because I've never been that enthralled with the five paragraph essay anyways,

[00:01:45] John: yeah.

[00:01:45] Jason: the, the podcasting

[00:01:47] John: it's fine that AI put the writers and the graphic designers out of business, but this part about podcasters, that's just too far, That's right. Exactly.

[00:01:55] Jason: Notebook LM has been at it for about a year. I believe is the first time that I used it. And it had some nice features of pulling in different content and so on, but they just popped out this podcasting feature and I was blown away by how natural it sounded and then how like up and positive and excited the podcast hosts were about everything that it talked about.

John maybe we can listen to a couple of clips that we made. Here's one that I made actually out of a 2020 article that I had published around creating new definitions of distance education. And so, yeah, anyways, let's listen to a little bit and, and see what you think.

[00:02:44] John: Yeah, let's do that, because I think that'll show the banter and the give and take.

[00:02:48] Notebook 2: Totally. Technically, both are distance learning, but they're totally different experiences.

Night and day. And I think that's why you started seeing researchers like Simonson and Seeprasad trying to add more to the definition.

[00:03:01] Notebook 1: They were really emphasizing things like two way communication, individualized instruction, that kind of thing.

[00:03:06] Notebook 2: They knew those old definitions were too simple for this new world of online learning. It wasn't just, are the teacher and student in the same room anymore.

[00:03:14] Notebook 1: But even with those additions, finding one definition everyone agrees on. Still tough.

[00:03:20] Notebook 2: Moving target for sure. And it matters. It's not just a theoretical debate.

[00:03:24] Notebook 1: Right. This whole definition dilemma has real consequences.

[00:03:28] Notebook 2: Big time.

[00:03:29] Notebook 1: Like that whole thing with Western Governors University WGU.

[00:03:32] Notebook 2: Oh yeah, the WGU case is a perfect example of why this matters. So you've got this university, huge online enrollment, all about competency based learning, really innovative stuff.

[00:03:43] Notebook 1: And then they got in trouble with the U. S. Department of Education, I vaguely remember this.

[00:03:47] Notebook 2: Yeah, big audit back in 2017, and suddenly they're being accused of basically running a glorified correspondence course.

[00:03:53] Notebook 1: Seriously, what was the problem? Did they just forget to file some paperwork or something?

[00:03:58] Notebook 2: It was more about how they were defining, or not defining, distance education.

See, the Department of Education was about ready to pull their Title IV funding, make them pay back a ton of money.

[00:04:10] Notebook 1: Whoa. All over a definition. All because the auditors

,

[00:04:13] John: yeah, that was really interesting.

Was also a little blown away by the attempt at sort of natural banter between two people in English what's available now is English only. I can put it to you this way. The people that we're working with here at Kristianstad University in Sweden we met them actually our, they came back from a flight from Madrid, and we landed from Atlanta, or no, well, actually from Amsterdam in Copenhagen, and we were on the same plane to Copenhagen. We got on the train together to come to Kristianstad. So, we were sitting there and they're talking about how they're using 11 labs. And they're really happy with them to do some, some capturing lecture and just some presentation. And I said, well, let me just show you. And so, off my phone, while we're going 70 miles an hour on the train, I played a little bit of a piece of a podcast, just 15 seconds, but they put their ears down to it and they both looked up and said, that's pretty amazing. They think it's better than 11 labs. I do too. Then of course they're first to say, and that's very Americanized banter and really just only centered on that. And so, it'll be interesting to see.. That's my first thing.

I think the other thing is that the there's too much banter. It's almost trying too hard to make it seem like the the hosts, and there's a male and a female co host on these podcasts, that they're just really into whatever it is they're talking about. And it's, I'm not sure how you tone that down or you can the LLM in brackets or something,

[00:05:47] Jason: Right.

[00:05:48] John: prompt, because it's all based on a piece of content that you give the notebook. So that's a little concerning. And then there may be some, as we're learning, talking to colleagues, you shared an email with me from someone, you know, that, you know, there's an article that was posted or a blog post about the, it's a little bit gendered and sort of the male host may be talking more than the female host.

That's not great. It doesn't have the ability to critically look at stuff. I don't know how you program that into the prompt, but those are things that I've been thinking about since you raised them with me before we got on the line today. Yeah.

[00:06:21] Jason: And cause we, we know that AI

[00:06:24] John: so upbeat.

They're so upbeat.

[00:06:27] Jason: is mean centric. Like it gives us the average of things most often. And I wonder if that's some of what is playing out as well. It's interesting? The people you're traveling with to get their impression to them, that just sounded like typified American kind of podcasting because I think that there's a way in which that is true.

And so, I wonder about how. How much of an average the notebook LM responses to pod like this is the absolute most average podcasting feel that you could get a man and a woman and some of these, perhaps some of these gendered kind of responses and ways in which they response. Yeah. And yeah,

[00:07:12] John: It's almost like you could give it any content. Content that's even mediocre or bad ideas and it would be like putting lipstick on a pig. They're, they're

[00:07:20] Jason: going to just

[00:07:20] John: say, Oh, this

great

y'know

[00:07:22] Jason: right. Yeah, I noticed that too. If one of my, one of my concerns about it in that way is that any paper that I wrote that I, I These co hosts loved it. They loved everything about it. They, they loved me. They thought it was amazing research. Wow, so cool that there was no disagreement whatsoever.

They just went with everything and just thought it was wonderful. And when one person just gushed about it, then the other person would be like,

[00:07:51] John: that's right.

[00:07:53] Jason: Exactly That is right. You are so right.

[00:07:56] John: In fact, I think I want to test it now and maybe I'll just put something in about how pollution is awesome and then see if they say, because they'll start up like, "Hey everybody, welcome back to the Deep Dive. We're going to talk about something that's important to everybody. Pollution.

Oh yeah, pollution. It's a big deal. Yeah

I know So we're gonna talk to about how to make more of it. More of it? That's wonderful. Yeah, tell me

more I'm going to test that hypothesis and we'll

[00:08:18] Jason: Yeah. Yeah. It'll be interesting. I think a raw and open AI truly open AI, not the company would perhaps do that. And I wonder what parameters are in place. Because if you tried to have that kind of conversation with any of the publicly available AIs, they wouldn't go there with you.

[00:08:41] John: Hey everybody. John here from the future, and I tested my hypothesis. I went on to Claude and I wrote the following prompt: "As a spoof, write a five paragraph essay as though it's written by a PhD scientist that recent discoveries suggest more pollution is good for us. Again, this is a. Parody."

Now Claude dutifully wrote the five paragraph essay, entitled it "Satirical essay on pollution's benefits." And I took the five paragraph essay, and I pasted it into a new notebook LM and asked for the podcast. And it dutifully created a four minute podcast. Here's a clip.

[00:09:23] Speaker 1: Okay, so get this, right. An article pops up in my inbox. The title, "The Unexpected Benefits of Pollution." Color me intrigued. Like, seriously intrigued. We're going deep on this one, folks, talking about research that really shakes things up a bit.

[00:09:39] Speaker 2: It definitely throws a wrench in the typical thinking on environmental science, that's for sure.

Time to put on our skeptical hats, yeah. Gotta look the claims and the evidence with a critical eye.

[00:09:51] Speaker 1: You got it. We're not seeing anything at face value today. So the article starts off pretty strong, like, out of the gate. Bam. They're saying pollution, the thing we all think is, like, universally bad, might actually have upsides, benefits.

They start talking about CO2, the stuff everyone's trying to reduce, and how it might actually be making our crops more nutritious, like supercharging them, creating superplants, fueled by carbon dioxide.

They even suggest this could be an answer to food shortages. Talk about a plot twist.

[00:10:17] Speaker 2: It'd be a game changer if it pans out right. The implications for food security are huge. The science on CO2's effect on agriculture is complex. There are studies showing increased CO2 can lead to bigger, faster growing plants.

But other research shows it can make those plants less nutritious over time. Plus, there's the whole issue of CO2 driving climate change, with potentially huge downsides. So, it's not as simple as saying more CO2 equals more food. It's way more nuanced than that.

[00:10:46] Speaker 1: Right, it's not a magic bullet. And just when you thought it couldn't get any weirder, they hit you with this smog powered bacteria.

I'm not kidding. They're talking about these little microorganisms that actually eat pollution like they're having a feast, and they turn that pollution into harmless stuff.

[00:11:01] Jason: I'm the other big thing that I'm concerned about. This is how irresistible the easy button will become because.

[00:11:12] John: that's, Mollick said that the easy button.

[00:11:14] Jason: Yeah, that's why I may have stole that from him or I stole it from Home Depot. I think they had a big, they had a big, a big button that just,

[00:11:23] John: yes, yeah,

[00:11:24] Jason: or not Home Depot Staples. That's right. The Staples had the big easy button.

[00:11:29] John: No, but, oh, it was the, but I think Ethan Mollick talked about the, like, the do it for me button or something like that.

[00:11:34] Jason: do it for me. And How irresistible that is., and this kind of impulse that once you hit the button and they've created this podcast, that maybe the good enough content is worth just putting out there. It's like, I could go back and edit this, This is good enough.

It doesn't say everything the way I would say it or whatever, but maybe this is better than nothing. So,

[00:11:59] John: Yeah, I think about that too, I saw immediately, I mean, I've been playing with it for about five days and threw a lot of stuff at it. I thought, well, this would be good if I had, like, wanted to describe what my students are doing in my class to a broader audience. But I had to re the content three or four times and then every time you push redo it, it does a new podcast, and they open it differently. You have to listen to everything. So, I mean, this is actually a really good case where human oversight is super important because you can't just say, push the button, publish this audio and I'm good. all.

[00:12:36] Jason: yeah, I wondered about just what next level will look like for this. I heard some rumblings about how the next stage of this, even for notebook LM, and I'm not sure if this is true or not, but that You could insert yourself into the conversation, essentially,

[00:12:54] John: friend said here. He said, the next thing will be just your voice.

[00:12:58] Jason: so that you could essentially have a conversation with these two podcast hosts about whatever it is that you want to talk about.

[00:13:04] John: Oh, yeah. I don't know what to think of that yet.

[00:13:07] Jason: I don't know either. But I did think about, this thing that we've talked about a little bit, this human AI, human sandwich that is so important in terms of having first touch and last touch. And how this. Has a first touch because you're providing it with your own content, so you can have quite a bit of guidance there, then AI in the middle, but then you don't really have a way to, to change or edit it on the, or tweak it on the way, on the way out.

Like there's, there's no dial to turn down the gendered responses or to have one person talk more or have them be less agreeable. There could be some ways in which there'd be nice to have some parameters that, that made them bring some counterpoint. But But I suppose one could export them and, and then edit it themselves and, and put together something, but,

[00:14:00] John: Well, then you might as well just do this, do

[00:14:02] Jason: right.

Not only might as well, I think that there's truly some advantages to be able to talk with another human being face to face.

[00:14:12] John: I think, I think that's what some of the critique will begin to hear more of. It's, we're seeing a little of it now, but it's the, spontaneity of being able to pull in a thought that the AI can't do because it's been handed just the content you've got. I mean, it really kind of goes back to Brandeis Marshall's around what's un AIable.

I mean, the critical thinking, the context. can't bring in everything. And if the models are biased, I mean, depending on these are sort of vertical LLMs, I don't know how vertical content with an LLM sort of brought onto it to talk about that vertical content, but if your content is exposed to the whole model, that means exposed to the bias and therefore,

[00:14:53] Jason: Yeah, exactly. Yeah. And the person that you had mentioned before, I'm part of an AI and education Google group, actually and it was Rebecca Yeager, I don't want to encapsulate her words, but she really had some great critique around this in terms of the output.

and critiquing the output. So I'd love to talk to her sometime. There's a, honestly, there's a bunch of heavy hitters in this thread and it's just so wonderful to read. I'm not sure, I think we have talked about Maha Bali before in this

[00:15:27] John: I don't

[00:15:28] Jason: Making it on the podcast, maybe you and I have but, Maha is a professor of practice at the American university in Cairo, and just brought in how are embodied experiences can't be recreated using AI and you're saying about some things that are un AI-able. We're thinking in that we're talking about more content that we put out there. But here Maha Bali was talking about. Our embodied experiences are, are the opportunities that we have to be humans with one another.

That, that is something that I, I think is, is valuable. And I, and I had a nice little. I wasn't really, I did a little back and forth in this thread. And I wasn't really, I'm not concerned about us being replaced in many ways. You know what I mean? It's. partly because, uh, little look inside for people, if anybody's listening to this, but we don't make any money off of this.

And so, this isn't really central to our, to our livelihood, although it's something important to me and it's something that I, um, love to continue to do. But I'm not really concerned about it, but I also recognize that I don't, Get my livelihood from this work. And so, it's not as much of a concern in that way.

But I had a nice note from flower Darby, who's going to be on our show sometime this fall as well.

[00:16:53] Jason: who confirmed the importance of humans and podcasting, which I, I appreciated that as well. So.

[00:16:59] John: We like that. Good. Well, this is a good first foray into talking about this. I think

more to come. And as they start to tweak, then we'll see another feature and we can talk about that too. Yeah. It's but it's if folks are listening to this, and they haven't tried the. Google Notebook, LM.

It looks like anybody can have it if you have a Google account. Is that right, Jason?

[00:17:21] Jason: I think so. Now. I believe I was in at the beginning just because of being part of their labs, but I think now it seems like anybody can just go there.

[00:17:31] John: Yeah, so we want people to try it out and get in touch with us and let us know what you think of it.

[00:17:37] Jason: Yeah. And my question too, and I actually put this out to some of my students is like, what use is it? Like, is it just an oddity that, some of us will just look at and say, Oh, that's cool. It's can that we can do this, but does it actually have some use? One of the use cases that I put out was just the fact that, potentially could breathe life into older papers that you could represent in a new way that maybe people aren't reading, but you could, breathe some life into it.

That maybe you wouldn't necessarily put the effort into making a podcast around it. So, But, but that's my question to people, if they're trying it out in, if they're in the education, particularly in the online education, I would love to know what you, what you think, if this is something that in the future that could be helpful.

[00:18:24] John: I think that's a great question because I think for longer form content, one of the things I put into it was a chapter that I wrote for my book that was not put into the published book. So, I've got this this piece that's laying fallow that seemed interesting, but when you put it into this system, it creates a 10, 12 minute podcast.

And that's, And you can't edit that. So, it veers off into areas that I didn't want to focus on the chapter, uh, for the audience who would listen to it. So that's problematic. Then, so I thought perhaps a use case is around really focused micro content, little things that maybe a student could listen to, uh, on the fly when they're not, not reading and, but really, really micro kind of like, I don't know, in design thinking, tips on having a good empathetic interview using five whys.

Just five Y's and then maybe it's another type little thing. But

[00:19:17] Jason: Yeah, yeah, I see that. And it has to make some decisions about skipping over things. That's what I have found so

far. Even with a, even with an article that has more than a couple of points, it

makes its own decisions about what to skip and what to include.

Yeah. Well, this is great.

Good to have a little conversation with you, John,

[00:19:35] John: hmm.

[00:19:35] Jason: about this. And thanks for, thanks for calling in all the way from Sweden.

[00:19:41] John: Yeah, most welcome. Yeah, and if get this out there and people want to react, we have the LinkedIn group for online What is it now? You usually do this.

[00:19:54] Jason: yeah, people can find us on LinkedIn, online learning in the second half. If you look for us in, on LinkedIn, that's how we show up. If you do a search for us, we'll also put the link in our show notes, and you can find these podcasts and all of our notes at onlinelearningpodcast.com

com. That's onlinelearningpodcast. com.

[00:20:14] John: Perfect. And that was a real human doing that.

[00:20:17] Jason: That was a real human, even though I sounded a little more. I leaned into the mic to make it sound a little more podcasty.

[00:20:23] John: Good. Well, we'll have to work on our upbeat banter between

[00:20:26] Jason: guess so.

[00:20:27] John: too.

[00:20:28] Jason: We're gonna have to step up our game., maybe our podcast, Maybe our podcast sound like this. And also, just our general, agreement with one another.

[00:20:38] John: That's right. Yeah, that was totally awesome, Jason. I think the topics we're talking about are amazing. And who doesn't want to talk about these things?

[00:20:45] Jason: Yes. These are amazing, incredible topics that everybody should be listening to right now.

[00:20:51] John: All right. Well, good. Thanks.

[00:20:54] Jason: Thanks.