Venture Step

Dalton Anderson (00:01.048)
Welcome to Midget Step Podcasts, where we discuss entrepreneurship, industry trends, and the occasional book review. Imagine you're having issues comprehending a complex research paper, lengthy articles, or just in a rush and scramble to get a good understanding of something before you have to do a presentation, just less than ideal situations where you need to compile lots of different information and

consolidated into a quick.

how to or break down, explain to me like I'm five situation. And there is a way to do that now and have it output it as a podcast, which is really cool. And it's a project that Google has been working on for a while and has been utilized by myself for a couple of years. And the original project was called Project Tailwind.

And Project Tailwind has since evolved with many different names. And they release some new features. And I think it's worthy to talk about. So today, we're going to be exploring a potentially groundbreaking technology and self -learning and learning in universities and basically giving yourself the opportunity to have your own personalized AI tutor.

per subject and documents you upload and those documents that you upload, the AI will only reference the documents that you upload. So you have less issues with hallucination, which if you're not familiar, hallucinations are when you ask something and it just freaks out and says something crazy, not related, or sends information back to you in a different language, it's happened to me before.

Dalton Anderson (02:01.564)
So these models do have hallucinations or rejections.

But with an LLM that you control all the source material, it has to reference those things. And then in those questions that you ask the AI model, it will send you the sources like, OK, I got this from this document and at this exact point. And it highlights it for you on the web page. I would use it to study and things that I was struggling with. I would.

use the LLM to find stuff like, okay, I'm not really grasping this topic here. Ask a question. That question has to get pulled from the source material that I'm looking for. Then I would go and read more about it. Well, yeah, we're going to be discussing how this potentially will impact education and dive into this AI audio summary piece. But of course, before we dive in, I'm your host, Dalton Anderson.

My background is a bit of a mix of programming, data science, insurance, offline. You can find me running, but in my side business or lost in a good book, the podcast, this is in both video and audio format on YouTube. If audio is more your thing, you could find the podcast on Apple podcasts, Spotify, or wherever else you get your podcasts. But before we dive into the agenda today, I do have some updates. So last week I had an episode talking about how I was in Japan.

And I was having some issues finding my place and luckily I did find my place. Well, I have another trial. I typically when I go abroad, I like to immerse myself and try to get friends and things like that. And the way that I do that is I play sports. So I try to play pickup sports in that area. If I'm in a big city, there's typically a pickup sports culture. So I just go and I'll play some games in the league. And in Mexico, I played

Dalton Anderson (04:06.666)
three times a week minimum, probably. And then Japan, I played some pickup volleyball and pick up soccer and I played pick up soccer. What is yesterday? And I was playing on this turf fields with my turf shoes on, which I haven't really used that much in the turf is like super sticky. And one of the players on my team like hurt his knee because it was like too much grip and like it just like his knee buckled, which isn't too good.

And then I was playing maybe like five games in. I took a shot. The guy blocked the shot and I went to go. I received the ball after the reflection, touched it down and then I went to go past the guy and he put his leg out like behind his front leg, which you're really not supposed to do. It's more of like a joke thing. He was doing it for real as a tackle and I was going to step on his leg. And so I try to stop myself from stepping on his leg and like stop my step. But I had a lot of momentum going.

And so what happened was my leg got like stuck underneath my body weight and I just landed on my on my pinky toe and it had like a 90 degree just my whole ankle folded. So I may have broken my ankle or have a severe sprain.

which is really difficult when you live in a city, a big city, it's not really built for.

these kind of things, especially when you don't know the language. So I'm going to have some issues there. So I'm going to figure that out. My current stance is I've got it elevated. I have no means of icing. So the place I'm living at only has a mini fridge, but no real freezer or way to keep things cold. The refrigerator is OK. Maybe it keeps things at like 50 degrees, but not freezing temperatures. So if I buy ice, it's just going to melt.

Dalton Anderson (06:05.508)
There is no pharmacy or place nearby, no food places right next to my house. mean, the closest one is like 800 meters. I'm having trouble booking a place to get an x -ray. And I needed to get an x -ray because my foot is like crazy swollen and it's all purplish now. So I know I know I at least torn some ligaments in my ankle, but I can't walk. So.

Like I can't put any weight on my ankle because it hurts. And obviously when you do things like a severe sprain or if it's broken, you don't want to put any weight on it because you can do severe damage during the recovery process. But basically, I can't walk. I have no means of storing ice. have no means of easily getting food or water.

have no means of like real like useful transportation. Like I could do Uber, but Uber only takes you to one place and you have to go to multiple places then that doesn't really work out because the metro is too much walking for me and like too much stairs. So I can't really do that. So here I am. And then the next issue that I have is getting an X -ray because everything is in Japanese. And so

I just yeah, I mean my I tried on my computer. My computer's not converting everything into English. Does it for some websites, but all and I typically on my phone I have a pixel and like if I swipe up on an app I can just like select all the text and I could ask it to translate everything for me. But it hasn't been doing that. Doesn't work with Japanese, but it works with like other languages, just not Japanese for some reason. And.

So I'm trying to figure out like where to go to get an x -ray and get that taken care of before I travel to Korea because I'm still going to be abroad for another two weeks. And then I have another like week and a half in Las Vegas for a conference. And I was trying to run a marathon and that's done. So that's no longer a thing anymore. I wasn't really trying. I was going to run a marathon. I signed up for the marathon. Like I have a marathon planned in two weeks.

Dalton Anderson (08:26.948)
but obviously that's scrapped. I was planning on doing like a lot of activities and stuff like that when I was in Korea. I think that's scrapped. There's a whole bunch of things I'll have to readjust. And when I was in Las Vegas, I planned on renting a car and then going to the Grand Canyon. I think I could still do that, but I won't be able to like hike around. Maybe I could just get out of car, get a good view of everything.

call it a day. I don't know. But yeah, so my ankle might be broken or it's just severely sprained. So that's an update on my travels. That's a big hamper on my, on my, my travel plans, but it is what it is. We will keep pushing and we will prevail on the other side for sure. I have no doubt of that. So we're just going to keep pushing. So my plan tomorrow is since I'm having such a

difficult time getting access to these urgent cares and like knowing which urgent care does x -rays and like which hospital does what. I'm going to talk to the embassy, US embassy tomorrow and hopefully the US embassy could help me coordinate everything. Cause another issue is like there's a rule, not really a rule, but if I try to call a Japanese number without a Japanese phone number, doesn't go through. So

At least that was what my friend was telling me. And I tried it doesn't work. And he tried it doesn't work. And he's lived here for like not in Japan, but he's lived in Korea for two and a half years. And he has friends in throughout Asia and obviously in Japan. And that's what he was telling me. So it makes it a little bit more difficult. I don't know. I haven't really verified that, like on the Internet or anything like that. So that could be fake news. I have no idea. So just throwing that out there like I'm like, I'm not necessarily skeptical.

But I'm definitely not overly confident. I haven't verified that myself. So trust, but verify. So I don't know. OK, so that was just a general update of where we're at with the LLMs and my personal life. So let's get into the LLMs. OK, so let me just give a brief background of Google's notebooks.

Dalton Anderson (10:53.634)
And I might mistakenly say Juniper notebooks because that's the Jupiter notebooks is that that's like a popular notebook.

Dalton Anderson (11:05.032)
Mark like it uses Markdown and Java to organize your Python code or just code in general. So I might accidentally say that not intentional, but that's what I'm talking about. If I say that I actually mean Google notebook. Alright, but go ahead. So Google's notebook. Was originally encoded project tailwind tailwind came about maybe two years ago, two and a half years ago.

And so they've been they've been really sneaky about it and they haven't really advertised it. And if you if you knew you knew but if you didn't know there's probably no way you would know. And one of the things that was demoed was

you could upload a file and that file would make a notebook guide. That notebook guide would have like a study guide, a quiz, and they would also have this like AI podcast thing that would create a podcast from the text that you gave it. And I don't really know the limits that you could have. I think when I was doing it, you could only do up to 500 ,000. I know they increased it.

I think you could do a million tokens now. I think you could do a million tokens, which is a lot of tokens. You could upload a lot of big documents and they could be PDFs, could be Word docs, they could be websites, whatever.

Dalton Anderson (12:39.012)
But that's a new feature. That's not something that's been around for a while. But the other stuff, like you could ask it questions and then it would provide the source material that you gave it the exact location that the source where the source material is referenced. And it would give you like five different sources. Like, OK, like these are the things that I used to get this answer. Obviously, not every time it needs five sources. But if it does need five sources, it will show you the ones that it utilizes. If it needs like 15.

I'm going to show you all of them and we'll just show you the most important. But I would like to, with this overview, I would like to show you the live demo. so what you're seeing, before I show the video, what you'll see is you'll see the notebook or notebook guide. You'll see audio overview and you'll see this other thing where he drops into this audio overview and he has a conversation with the AI host. That is not a current feature that's available.

but it was demoed at a keynote. Okay, so just letting you know. So let's share this.

Dalton Anderson (13:52.228)
Let's see, yes, sharing. Hold on, it's not sharing. Just give me a moment.

Dalton Anderson (14:01.666)
It's annoying, right?

Dalton Anderson (14:08.58)
Okay, sure. All right, how's that look? You should be able to see now.

Okay, so I'm play this video real quick and you should be able to the sound and everything.

Dalton Anderson (15:48.142)
Okay, so this join feature that he's discussing isn't a feature that is currently available, once again, but I think this is the sickest part of this whole product and the idea of it, because you could ask questions, break down complex topics, and get it in the perspective that you want and that you would appreciate and that you could be.

I wouldn't say that, that you could easily understand and comprehend.

Dalton Anderson (16:45.654)
Insane insane. So that is What they demoed what we have now is we have this audio overview feature and that audio overview feature allows you to

listen in to like a podcast of the material that you provide. And I think it's just, I mean, it's something else, honestly. Like I feel as if it's so human -like. When my friend Ryan was over here, he was staying with me. Not staying like at the house, because my house is crazy small, but he's in a hostel and we kind of did this trip a little bit together.

And he was helping me out because I hurt my leg and can't walk. And he was over and I was messing around with this prepping it before the episode today. And he was like, what what podcast are you listening to? And I was like, no, this isn't a podcast. This is.

AI and he's like, wait, what? And I was like, yeah, you can upload. You can upload a file and then it create this like AI podcast for you and you can listen to it and like learn what it's about. And then you ask questions. You'd have to write the questions out right now. But in the future, you wouldn't have to do that. You could just ask it via audio. And. He was like, what? That's insane. And I was like, yeah. And he was so impressed with it. He's like, OK, OK.

What if I gave you a document right now and would it just do it? Like what happens? And I was like, yeah, if you give me a document that you want, we could upload it and then it's got to process, take a couple of minutes and then we could listen to it together. And so we did that. So we have our own that we did. We did an army one army podcast. He's in the army. And he, let's see.

Dalton Anderson (18:54.7)
He was super impressed with it. Load. Audio overview.

guess you have to load it after you close it. It's odd.

But it gives you this ability to listen in to these large documents. And so it breaks down this 100 page rule regulation document that the Army posted. Let's see, Army.

So I'm sharing my screen again and then we'll just play. So this is AR670 -1.

Dalton Anderson (19:56.334)
And you can see how it's not perfect. Like it said, it's AR670 -1 Ramen. I don't know where they're getting their ramen. Maybe they need regulation. I'm not sure.

Dalton Anderson (20:32.804)
I'm not gonna play the whole 11 minutes. We played 40 seconds. I think he asked a question. So there's some suggested questions. Like this is, are the changes to Army uniform regulations regarding wear, appearance, and insignia? So you click it. He asks this question. And then it's kind of like a chap GPT or a Claude vibe or Gemini. And it's kind of like a chat bot thing. So you put it in here and.

Okay, so the first thing is policy for breastfeeding and pumping. It clarifies a policy for breastfeeding or pumping in uniform authorizing female soldiers to wear an optional undershirt while doing so. And then you can click on this, what I was mentioning earlier, source, so source one. And then right here, it brings me to the exact location that it referenced, so.

Summary of change 670 -1 wear and appearance of uniforms and insignia. This major revision dated 26 January 2021 clarifies the policy for breastfeeding and pumping uniform authorized female soldiers who are breastfeeding or pumping to wear an optional undershirt para 3 -6C. And then, you you keep going through. So I think it's super cool. And

an amazing way to help yourself learn. Let's see, let's stop sharing. And overall, I just think, you know, not every teacher or professor is gonna be able to.

Dalton Anderson (22:19.94)
appease every student and their learning style, but you can appease your own learning style now and teach yourself things. And instead of a teacher trying to get every student to comprehend what is going on with a lesson, I mean, obviously that's their job.

but you could lay the groundwork of what the real expectation is. Like, hey, I want you to understand and know how to do X. How you get there isn't really my prerogative. That's what we need from you as a student. And so if you need to use this LLM notebook to talk with these AI teachers or if like,

you know, experimentation is better than you or better for you. Or if you're an audit editorial learner, like you can, you can use these LLM notebooks. If you're a visual, you can do whatever we could, we could show you. And so there's, opens up like these different things. Or, you know, you could also say you missed a class and this is super common. Like you miss a class with the professor or you miss a class with the, you know, in high school or middle school.

whatever, and you could have the notes of like what chapters you went off, like what chapters the teacher went on. And then you can just have a lot of schools now, you're required to have an online textbook. So you could just copy and paste those chapters that you needed, throw it in here, and there you go.

As long as you're not selling the material and you're using it to learn a lot of those books under the licenses that they have, as long as you're using it to learn and not using it for profit or anything like that, it's okay to do those things. And so instead of you having to catch up manually or struggle to get the notes or take someone else's notes and copying them over, just say, what chapters you guys work on? Okay. Then copy and paste those in, make a notebook.

Dalton Anderson (24:34.24)
Or you can make a notebook from the book and then ask it to talk about those chapters and break it down for you. There's a lot of things that you could do.

I think this technology is important because a lot of people and Ryan, my friend Ryan asked me recently, like, you know, so like what is AI gonna do for me? You know, like we always talk about AI, AI this, AI that. What really is it doing for me? And I was like, fair question. And I think there is some key concerns, like, okay.

What really is it doing? There's so much money being pumped into AI. Is it providing useful things to the public yet? And the answer is probably not really, not as much. The general public doesn't use AI every day. mean, besides people that program or people that have repetitive tasks, there isn't that much utilization outside of those groups. And a lot of times if you're using AI that you're more technical and

inclined to do so. There are some early adapters that are outside of those technical areas, but many people that are technical are curious. obviously there's a lot of talk about this, these things and these products. So they're trying them out and experimenting. There's a whole bunch of reasons, but I think this is a big step in the right direction. Like imagine

you have onboarding process and you have a whole bunch of rules for your department or whatever regulations. Instead of having like all these meetings and are like providing all these notes that the person has to go through, just send them an LLM notebook link and have them ask questions there. can I do this? Or how do I request PTO? Or how do I do X? Or you know, what's this all about? Or you can listen to this audio recording that it's created by AI.

Dalton Anderson (26:44.86)
And you couldn't really tell. If you closed your eyes, you wouldn't really know that that was AI. Some words, you'd be like, that sounds a little robotic. But for the most part, pretty decent. Pretty decent, honestly.

And so this allows multiple approaches to learning. And I think that the role of AI in this sphere of learning will increase as AI becomes smarter and smarter. I think that the new release of ChatGBT, whatever their new model is, forgot the name, it came out and said, I think the IQ is like 155 or something like that. And I don't know how legitimate that is because

The questions are probably in the training data and.

Dalton Anderson (27:36.676)
It's probably in the training data and who knows how useful these benchmarks are and how these models are taking up so much data that they have to create their own synthetic data to train the model on. There's not enough data in the world to keep training the model or useful data to train the model, so they have to create their own data, their own synthetic data to train the model. So it of puts in perspective how much data they're utilizing to create these models.

Dalton Anderson (28:12.406)
And so I think this like increases, you know, as I stated, like it increases the role of AI and learning. It allows people to ask questions, you know, in the future with like a mid, mid audio followups. And it will allow people to explore topics that they might not necessarily be able to explore during a lecture or in class, but they can

do so outside of class. maybe a professor or a teacher sets this up, like sets up these LLMs before the semester, before the class term. And these are like, okay, these are the topics we're gonna be talking about. See where you're at, ask questions, and we might not be able to get to everything during class, but you can ask these questions to the LLM and you'll be able to see like what's going on.

and get a better understanding and you'll be able to see what sources are being referenced and where and you can read more about it.

Dalton Anderson (29:23.756)
I think that's like super useful. I don't know. I'm a big fan of this. I use it a lot for studying for my insurance exams. It helped me create study guides. It helped me understand topics that I was lackluster in and

Just overall, definitely.

sped up my learning and how fast I was able to learn, both with the LLMs and the study guides it created and the quizzes it gave me, is overall a phenomenal tool for learning and self -learning of that. And I think the next thing is with these audio reports, it's gonna allow you to really condense these topics into.

bite -sized pieces like that army topic the army regulation that army regulation topic it turned a 90 page document into 11 minutes it would take you way more than 11 minutes read the whole thing and get a good understanding and I'm not saying that these audio overviews will give you all the important information that you would need to know but you should probably have a good idea

of what the document's about and have a good summary and then you could dive deeper if you needed to and ask questions and those questions would be, those answers would be cited and you could go to those citations in within the document, the documents that you uploaded and then bam, there you go. Then you've got all the answers you need. So it really reduces the time to find things for sure.

Dalton Anderson (31:10.242)
Like either the LM could answer them for you. And if you want more information, the LM sites where it gets it and it tells you where to go. So overall, that just makes it really easy. Really easy.

And so I made another LLM and we can make one together. So let's do that. So you go over here. Let's do this. I'll share my screen in just a second. Let me just set this up.

Dalton Anderson (31:46.052)
it's notebook LM.

Dalton Anderson (31:50.776)
I've been saying L

same thing.

Dalton Anderson (31:57.381)
and model that we create with our notebook. Okie dokie. So let's share my screen again. Share screen.

Dalton Anderson (32:11.488)
All right, so you see that I've done some tests. I have done some insurance stuff. I have some other things. So now we have a new notebook.

So you can upload your sources. So it's this PDF text marked down. Used to, they updated it, but you used to have to only update from the Google Drive. this is my, I won't have, I won't have what I need. Hmm, here.

Dalton Anderson (32:42.528)
Let's see, let's go.

Dalton Anderson (32:51.672)
I'll just say Google research or DeepMind. DeepMind research.

paper.

Dalton Anderson (33:05.526)
Yeah, so DeepMind, sorry. just forgot the mind part. DeepMind research paper publications. And I'm just going to go to Alpha Proto. This is the one. Copy this link. Yeah, I forgot that I didn't need to do that. So then you can upload sources. You can do Google Drive. You can do the link. You can copy and paste text.

Watch this. So I'm just going use this URL right here that I got. So this is, let me share my screen so I can share. I'll share this tab inside. Sorry, had a cough. Alpha Proto generates novel proteins for biology and health research. It's like a new model that they're using for Alpha Fold that helps predict the protein. What is it?

Dalton Anderson (34:04.708)
protein binders and it just gives a better understanding of like how the proteins bind to each other. I'm overall not too confident talking about the topic, but I know this LLM will be and I could ask you questions. Share this tab instead. So that was where I got the URL. It's a Google DeepMind publication related to their Alpha Proto.

model that is related to

Dalton Anderson (34:40.47)
outfold, which is something that we talked about earlier in the podcast. So I'm going to generate this this audio overview, and I'm not sure why. I'm not sure why I may zoom in to not sure why this is so small on my screen, like it's like cutting off the words right here. I don't really know. But I hope this is not too small. I know there's been some complaints that, hey, when I'm sharing my screen, it's not big enough.

So maybe this is better. I'm like plus 500 zoom on this thing. But you can see it does an FAQ study guide, table of contents, timeline, briefing doc, whole bunch of other stuff. There's my little profile photo for my Gmail. And let's see here summary.

Dalton Anderson (35:32.58)
And that's the notebook guide so that you don't necessarily need to have that. And so this notebook guide is that thing that creates the audio overview. It has suggested questions that you might want to ask. It can generate these other documents for you. You can add sources. I don't know how many sources you can add. You can add 50 different documents now. wow, that's a lot. So they upgraded that. used to only be, I it used to only be able to upload 10 ,000, ,000

token documents. And so you could upload 10. But the issue is like, OK, like if you wanted to upload like a book, you have to split that book into like many different documents. And that would eat up on your document space because you couldn't you couldn't upload like a huge, huge document. But they're just saying 50.

What is the, let's see, is the max tokens.

Dalton Anderson (36:43.576)
seconds.

Dalton Anderson (36:58.23)
So it says...

Dalton Anderson (37:04.388)
I don't know, it says 500 ,000 tokens or 800 ,000 words, so it's saying.

It says 1 million tokens. 1 million tokens. So I was right earlier. I just wanted to verify it. It couldn't hurt. So what I'm really doing is I'm kind of stalling to wait for this thing to upload. But so you can upload 50 documents up to a million tokens. I don't know how many pages a million. How many pages do you think a million tokens is? How many pages is a million?

Dalton Anderson (37:42.884)
Okay, so 32 tokens, 32, 3 .4.

Dalton Anderson (37:55.012)
Huh. Okay, so 24 ,000 tokens is 48 pages. That's what the estimate is.

Okay, is this done yet? Okay, it's done. Cool. That worked out. Okay, so right here we are going to start playing and this is about that document that it wasn't even a document. It was a website. Extracted the information from the website and then made a...

Dalton Anderson (39:30.8)
I'm gonna pause there, because I mean, you don't need to listen to whole thing. But basically, alpha proteo is different than alpha fold. Alpha fold was understanding protein structures and the interactions between outside influences that you give to the protein, like medicines or different gene sequences that you edit. Alpha proteo is about creating proteins.

new proteins to help solve issues and their Alpha proteo and Alpha fold all should enable Google and their partners to enhance their capabilities for drug discovery and drug enhancements. If they go any further than that, I'm not sure, but I do know that they're going to be using it for drug enhancements.

But theoretically they could alter like DNA of animals, plants, those things. And so you could get plants that would have, and this was talked about in the All In podcast. I don't know the guy's name, but he's the scientist guy and he has like a pharmaceutical company. And he's talking about like growing plants, like growing like these good, good potatoes or something like that. Like he's got like a licensed potato.

seed or something like that. But anyways, it's more better explained by him, but basically, alpha fold would enable us to discover which gene sequences have certain traits in identifying the traits out of the gene sequences. like say that you had a plant that was drought resistant, okay? So you would have a drought resistant plant.

and then you could take the gene sequence from that one and then you could have another plant, same species, that was resistant to bugs, like pest resistance. And so then you have those two gene sequences. And what you would do is you would essentially splice the DNA into these different pieces and you would edit the DNA and create this new

Dalton Anderson (41:53.208)
gene sequence with those two traits with alpha fold. Before you couldn't do that, they would have to like breed the two plants. something like, the that he explained it is like really typically like, okay, drought resistant plants come with drawbacks, so do pest resistance. like maybe pest resistant plants don't taste as good or they, you know, they're not as ripe or whatever. They take longer to grow.

Same thing with drought resistance plants. So you're getting, with these gene edits with AlphaFold, we'll enable is you'll able to get all the benefits without the negatives. So you have these situations where, okay, like you have A, A has these drawbacks and never activities associated with them. B has these issues associated with it. And so instead of having the drawbacks, you just have A plus B instead of having A plus.

or instead of having A minus minus minus B minus minus minus, you get rid of the negative signs and then you're just like A plus B. And so you're able to get both the traits of those plants without the negativity that's associated with them. Which is pretty groundbreaking and can be applied for other things. I don't know if we're going to go down that route, but I do know that we'll be doing so with plants for sure. And I know that

will be using it for enhanced drug discovery, which I think is great. Where hopefully you can cure some of these drugs, like think one of the, know, brutalist disease. Not cure these drugs. Cure these drugs with drugs, huh? Not drugs, but diseases. I think Alzheimer's is just like brutal, like a brutal way to go. I would love for us to get a cure for that.

Yeah, I've seen that in person a couple times and it doesn't get any easier.

Dalton Anderson (44:01.57)
Yeah, it's just rough. That's a rough way to see someone and for them to see you but not really see you. yeah, now Alzheimer's is rough. Like that would be a good one to resolve. I mean, there's many other diseases that need help too. That was just one that came to mind mid -speak, like mid -talking on the podcast. But what do you think about this?

LOM stuff. I've been reading on the internet. There's just a whole bunch of garbage. So much garbage on the internet, obviously. But there's just been so much garbage on like YouTube and like the internet and like Instagram. And I guess I'm not surprised, but they're like, yeah, like this is the new way to make a million dollars. You just, you you copy and paste like text in a year and then you just publish podcasts and then those podcasts will get ads and then those ads.

will make you money. And if you just rinse and repeat that like thousands of times, you'll make a million dollars. Like, like little do they know that I think to be a top 50, like you're like the top 60 % of podcasts or something like that. If you get, or I think even you're in the top 40%, if you get more than 30 downloads an episode, like you, you're not, it's not a big deal.

you're not going to make money from like if you're in pod, I'll rephrase it. If you're in podcasting to make money, you're probably not going make any money. If you're in podcasting because you like it and you enjoy it and you could do with or without the money, then yeah, I think that you might have success, but for the most part.

You know, I'm doing it for fun and I hope that other people enjoy the things I talk about. And it keeps me disciplined and keeps me interested in learning. But you're probably not going to be making a million dollars off of podcasting unless you get an exclusive deal or you're Joe Rogan or what is it? Call her daddy. I don't know her name off the top of my head. Yeah, those are like the two biggest podcasters. And that's where you get the hundred million dollar deal.

Dalton Anderson (46:23.384)
But you're not running ads making a mill. You would have to be doing crazy numbers, like crazy download numbers.

to make that much. And no one's gonna pay you top dollar for your ads that you're running if you have a whole bunch of different podcasts. But anyways, there's posts about making money via podcasting. I literally saw one before this podcast. I was like, are these people being serious? They're talking about making a million dollars, publishing these AI audio podcasts and just.

this overall garbage instead of utilizing the tool for like what it's meant to like it just, yeah, I don't know. I don't know where he will see it. mean, maybe it's just click bait stuff where people upload these things and then they just get a whole bunch of clicks because people are, you know, interested and want to know instead of being like, wow, this is a great learning tool. no, you can make millions of dollars using this. Not true. Not true. So what do you think about it? Like

Do you think that you would utilize Notebook LM? Do you think that you'd want your kid to utilize this tool or would you suggest it to your peers? is it something you wish you had in college to help you out on those late nights? I don't know, let me know. And I would appreciate if you tried this out and tried out this audio feature and share your experiences in the comments below.

I invite educators to try this out and students and see if it helps them is integrate this tech, sorry, if this technology helps them learn faster or gives them a better understanding of the topic that they need to learn about or discuss in class. And, you know, of course, if you like this episode or you like the podcast, can you please like or subscribe?

Dalton Anderson (48:29.804)
Definitely not going to beg for these things like before and after the episode. Just going to talk about it very, very briefly at the end. It really irks me when people are like before they even talk about their what they're talking about. They're like, please smash that like button. Subscribe. Turn on the notification. No, no, I'm not doing that. But of course, like if you enjoy it, like please like and subscribe if if you're just passing through. I hope you pass through again and hopefully we'll we'll pass.

I'll pass by you next week and we'll keep seeing each other and eventually you'll come and stay for a bit. Who knows? So I have a couple topics for next week. I don't know which one I want to talk about. I think that there is some things I need to discuss regarding this. What is it? Reflection 70 billion parameter model reflection. Apparently

There are some scientists accusing the company of fraud. So I have to look into that. And that's one of the reasons why I haven't I didn't talk about it this week because I'm letting I'm trying to let it settle down because I know in the last couple of episodes, I've been talking about how it's like the best in class model, apparently. And it's like blew the blew the community away and they were so impressed with it. And then the research paper posted. There was question marks on some of their assumptions they made and the things that they did.

And then there was even more question marks when scientists couldn't replicate what they did. That's a big red flag when you publish a research paper and people can't replicate or get close to results that you're stating. And so apparently they...

are accused and I'm careful the words I use here because it's not it's not solid but there's been some accusations from the scientific community that the reflection paper is a fake and the model that they're presenting to the public and the results that they are stating is false they're false. Do I know that for sure? No, I don't know.

Dalton Anderson (50:50.4)
But I'm going to look into it once everything kind of settles down a bit. So hopefully next week I'll be settled down. I can talk about that. Or there's some other tech stuff that came out came about. So we keep pushing along with that. But I would like to close out those books on the reflection model. But of course, I appreciate you joining joining in on this on this discussion of Google's notebook. I think it's a wonderful product.

It has definitely made learning for me easier. I hope that you'll be able to utilize this topic and you will not topic you will utilize this tool and you will.

Dalton Anderson (51:36.662)
enjoy it and I hope if you didn't know about it you're excited and if you did know about it maybe you didn't know about this notebook overview feature with the audio overview and you're quite intrigued and find it interesting. But of course wherever you are in this world good morning, good afternoon, good evening and thank you for listening and I hope to see you next week. See ya. Bye.