AI Expert Ken Jee Discusses Data Science & AI Startups

Data Science, Startups & More with Ken Jee

AI Expert Ken Jee Discusses Data Science & AI Startups

Subscribe to my AI newsletter and receive free gifts, such as my secrets to success on YouTube and my upcoming book!

In Episode 23 of the “What’s AI Podcast,” I received with Ken Jee, a prominent figure in data science and AI, to explore various aspects of these fields. Ken shares his journey into data science, which began during his college days where he used data analysis to improve his golf game. This experience highlights the practical applications of data analytics in everyday life and sports, setting the stage for a discussion about the broader implications of AI and data science.

Ken and I followed the discussion with the world of AI startups, discussing the current trends and challenges. We explored the reliance of new companies on AI technologies, particularly focusing on Large Language Models (LLMs) like ChatGPT. Ken offers his perspective on the shift in startup culture, where the emphasis has moved from solving specific problems to building solutions around AI capabilities. He emphasizes the importance of balancing technological innovation with clear problem-solving objectives.

The conversation also covers the potential future of AI platforms, such as the OpenAI store, and their influence on app development and market strategies. Ken suggests that while the proliferation of AI-driven products can lead to market saturation, it also opens up new avenues for experimentation and market research. This aspect is particularly relevant for companies looking to navigate the competitive landscape of app development.

Moreover, the episode touches on practical uses of AI in content creation and summarization, from how tools like ChatGPT can assist in summarizing discussions and generating SEO-optimized content to demonstrate AI’s role in enhancing productivity and creativity.

The episode provides insights into the challenges and opportunities in AI startups, the impact of AI on app development, and practical applications of AI tools in content creation. This discussion is valuable for anyone interested in AI, data science, technology entrepreneurship, or looking to understand how AI can be leveraged in various domains.

Listen now on Spotify, Apple Podcasts or YouTube:

Full transcript:

Ken Jee: [00:00:00] Do you know how many people think about doing a podcast and don't? A lot. I would bet anyone who's remotely extroverted or even introverted has thought about making a podcast. Like, oh, I bet I could do that. But how many of them have done it and how many have not? You already won. I've more recently been viewing the hallucinations like a feature rather than a bug.

Ken Jee: And the feature helps me to question it and think about the problem. I'm going in and I'm like, Oh, I have to think about this a little bit more. I have to make sure it's right. I have to validate it. And it means that I am still thinking about the problem. I'm not just a zombie who writes in a box and copy and paste.

Louis Bouchard: This is an episode with a great friend, Ken Jee, YouTuber, podcaster, but also a data science expert. In this episode. We talk about very different topics, including AI startups, but also going to podcasts and why you maybe should create your own podcast. Ken first got into data around 2012 and 2013. [00:01:00] So he has seen a lot and thus has extensive experience in data science.

Louis Bouchard: We also talked a lot about how we leverage language models like ChedGPT or Claude to be more productive and do tasks that we are not experts at. If you are interested into AI, startups, learning or building things, This episode is for you. I hope you enjoy it. By the way, don't forget to leave a like or a five star review depending on where you are listening or watching this episode.

Louis Bouchard: It helps the podcast a lot. Thank you. I'd love to hear how you started in the field or how you got into artificial intelligence or rather in data science. 

Ken Jee: Yeah, so I started getting interested in data to begin with when I was in college. So I was playing golf in college and something that we were required to do was, or we were asked to do is collect our data and upload it to this platform.

Ken Jee: I forget what it was called now, but it was essentially you would track your scores and, and you'd be able to compare them [00:02:00] online and do some very light analysis. And I started. Getting really into analyzing my scores. I realized that math or statistics could be helpful and understanding how I could improve my own golf game.

Ken Jee: Just looking at the standard deviation of the different hole types, looking at the areas of my game that needed improvement. There is a really clear action ability on a lot of the analysis that I was doing. So I could see there's a gap and I could go work on that area. To me, that is arguably the purest form of data science where you're using data to diagnose something and you can do something directly about it going forward.

Ken Jee: I started getting really interested, not just analyzing my own game, but trying to predict or project how professional golfers would play doing some daily fantasy sports. Doing some light sports betting things along those lines. And eventually I ran into a roadblock where I felt like the skills that I had using Excel or [00:03:00] Google sheets or any of those types of things, they weren't sufficient for me to be able to answer the questions that I had or to try to build the models that.

Ken Jee: I was trying to do. And at the time, I didn't even know what a model was, right? I was just fishing around in the dark, asking people, you know, how do we get all this data that comes in and it gives us this projection. So that led me on this maybe wild, mystical pursuit of understanding what modeling was.

Ken Jee: And it eventually ended with me getting a master's in computer science, becoming a data scientist. But there was a point where it was all new to me. And I still remember that feeling. Thank you. Not like it was yesterday, but I still do remember it.

Louis Bouchard: It's funny because it's similar to what you preach now, but you, you've learned data science directly to apply it to something as a solution.

Louis Bouchard: And on my end, I also prom this now, but I got into the field quite differently. It was, uh, on my end, it was at the end of [00:04:00] the, of my systems engineering degree where I had my first AI class and it was super theoretical. But I, I really loved that you could actually apply mathematics to something useful.

Louis Bouchard: So, okay, I guess it's kind of similar to, to you that I wanted to apply something, but it was still theoretical. Like I found it very, very nice that you could use mathematics to, to do something useful. But still at this time, AI wasn't something Like really useful that, that companies were using. It was just like image classification or, or things like more, more basic things that were just fun to do as, as a, as a projects.

Louis Bouchard: And yeah, I was wondering, you, you talked about how you got. Into data science. But when was this exactly? Like around what year?

Ken Jee: Oh, uh, probably 20, 2014. Well, so I [00:05:00] started getting interested. I started building the models when I was in college. So I was 2012, 2013, just analyzing my own game. And then 2014. I went to grad school for for business, and I started learning some sequel.

Ken Jee: I started learning basic statistical analysis, linear regression, clustering, those types of things. And then I finished my masters of computer science. I think it was in either 2017 or 2018. So I was working the whole time. I was doing my computer science degree. In the company, I actually work for now, so I was, it was really cool.

Ken Jee: I was in a great situation where I could learn the things and apply them directly to my work as I was going, which I think is probably the best way to possibly learn. Um, and I will say, I am a little bit envious that of of anyone who really enjoys the theory. And loves that they can apply it in the real world.

Ken Jee: I'm very much the opposite [00:06:00] direction where I love the real world problems and data science, machine learning AI. Those are the tools that I use to be able to solve the problems. I do have some enjoyment of the tools and, and the intricacy of them themselves, but it's definitely more for me on the other side of the equation is how do I use this to get exactly what I want?

Ken Jee: And I think that. Personally, that makes me a effective data science because a data scientist, because that allows me to not necessarily just use data science tools. We'll try to use literally anything at my disposal to get the answer that I want, which may be a good or a bad thing. 

Louis Bouchard: What do you think on the relatively new AI startup field where it's basically, I feel like personally that it's companies that just want to be based on AI and even for the most part that are just based on OpenAI.

Louis Bouchard: So do you think, like, what is your opinion on, on startups that. basically are dependent on artificial intelligence or on [00:07:00] new technologies. Is it like, I, I will just give it a bit more context to me. I feel like a startup is the goal is to solve a problem. And that's pretty much it. You need to be passionate about the problem and try to find solutions, whatever it is.

Louis Bouchard: Whereas now I feel like startups are going the other way around where LLMs are super cool and we want to build something on it. And now let's find what. So, yeah, I wanted to, to know what are your thoughts on this, like this new style of startups. Is it viable or like, should they do anything differently?

Ken Jee: So I think there's more opportunity now than ever to solve problems with LLMs. It's an incredible technology. There are a ton of use cases that we probably didn't think would be accessible for years in the future that are attackable now with AI. That being said, I think that there's also a huge influx of people into the [00:08:00] market who are building things just.

Ken Jee: To say that they're AI related and they're trying to get funding. I would argue it's probably the best market ever to get funded for an AI startup just from from talking to friends who have been funded and things like that. And I think that the way people are approaching it, they're not essentially, they're not effectively building a moat.

Ken Jee: They're building it. To get funding, not to create a long term company. And I think that that is fundamentally flawed. I think that if you go in with that approach that, hey, our focus is to get acquired. That's generally not going to be as you're not going to be thinking about the customer. You're not going to be thinking about the problem enough.

Ken Jee: And I've seen so many, I've seen it on Twitter the last couple of days, people like. OpenAI killed our startup and it's like, well, was your startup built on OpenAI to begin with? And they could just essentially build functionality around the infrastructure they already have. [00:09:00] If you're purely dependent on OpenAI, which, which can be okay.

Ken Jee: Your, your competitive advantage is just going to be your data and some of these other things. And that might not be something that sustains over time. If. Your competitive advantages, the data infrastructure, and you can slip in any of these large language models for that same use case that might be more resilient.

Ken Jee: But it's interesting to see how people are. Yeah, what I think you're describing and what I, what I see in that space is short term thinking, and that's something that I don't think makes the best companies, the best products or any of those types of things over time. 

Louis Bouchard: Yeah, I completely agree. It's funny how they are like, of course, there are so many because it's super accessible, but I guess it's also.

Louis Bouchard: Yeah. Those products are easy to build. I'm just wondering why there are so many, like if we talk about RAG, for example, the retrieval augmented generation systems, all with your own data set, like a lot of people just scrape some [00:10:00] websites and build a RAG based system that is pretty much the same thing as others.

Louis Bouchard: So the only. Like differentiating factor is the, your scraping system, which is easily replicable or like lots of people can build this as well. I find it just special that so many people are trying to build companies that, well, it's not even companies. It's just basically, it's basically the difference between a project and building a real product or a real company.

Louis Bouchard: Like I will assume it's, it's fine to do this for a project to build up your portfolio, but for a company you would need. Something like much more in depth where you, you need to have a real advantage. 

Ken Jee: I, you described it perfectly. I think that something that I've noticed is the AI tools make it easier than ever to build a product.

Ken Jee: You can get something out there. You can. Get users. You can build some infrastructure around it. And a lot of [00:11:00] people are conflating a product with a company. So a company is something that's profitable that you create a moat, a competitive advantage. And when you look at, for example, what Charlie Munger or Warren Buffett would invest in, they're only investing in companies that they believe have a sustained edge over time.

Ken Jee: I think maybe some people create a lot of small companies that don't have a sustained edge over time. Like Take dropshipping. I'm sure that there's 1 percent of people that are successful at dropshipping, but it's not because of the specific product they choose, right? So each individual product that they choose would probably will not be profitable in the long run, right?

Ken Jee: They'll make a quick win and then they'll move on to another product. But the thing that is their edge is the system that they've built to be able to spin up a ton of products over time. So I think What people are doing now is in the same way of drop shipping, where they're just starting with a product and trying to build it out and create a service.

Ken Jee: The window for that specific service is going to be [00:12:00] small. If, on the other hand, they're creating infrastructure to be able to build out a ton of services over time and keep them relevant, which in my opinion would be a tremendous amount of work, then maybe that's a moat that they would create over time because they have this know how.

Ken Jee: But The idea, and don't get me wrong, I don't think it's necessarily a bad thing that so many people are creating products that could be potentially useful to people. That's great experience. The belief that that would be an overarching company, that you're going to stake a lot of your individual livelihood or any of these types of things on that is a little, I would be scared for them in terms of risk tolerance, let's put it that way.

Louis Bouchard: Do you think the new, well, the, the upcoming OpenAI store is something that will. Change how we use or how we build apps, for example, I, there are a lot of companies that just like spam games, mobile games on the app store and whatever. And they, they, they make money out of the. [00:13:00] 1 percent of the winning apps that they build quickly.

Louis Bouchard: I was just thinking about this. I feel, do you think it's possible that OpenAI in the near future becomes some kind of new app platform where a company could be built around releasing tons of paid products on the OpenAI store? Like all super easily built as a, the GPTs, like the, the new GPT thing, like, do you think this, this could be a viable thing or is this worth investigating that, uh, 

Ken Jee: Yeah, I mean, I think it is, you know, I don't think it's necessarily a bad strategy to launch a lot of products where you experiment and then you figure out what works because in, in the, in this day and age where you can iterate and you can create a system with minor variation and you can run experiments.

Ken Jee: That doesn't necessarily mean you're diluting the product you're creating. It means that you're doing market research. And if you have that system down pat, I think it can be very effective to launch a lot of products in [00:14:00] an app store or a platform. Um, from a consumer perspective, I don't know if I would love that.

Ken Jee: I went on the Apple app store recently, just fooling around, looking around and the quality of apps. That I'm seeing, it's so low, you know, compared to, I remember when I was, I first got an iPhone, all the apps that were out there, I was interested in, they all seemed to have utility to me. When I go now, there's so many stupid games that Look completely uninteresting.

Ken Jee: Just just killing brain cells. Just just a waste of time. Just entertainment. They're getting away from actual utility in in the space or on your phone. And to me, that's just maybe a large, a larger overarching problem is that we're looking for ways to distract ourselves rather than to figure out what's what's meaningful in our lives or or in our careers or whatever it might be.

Ken Jee: But that would be my concern with these questions. The OpenAI app store overall is what direction does this [00:15:00] go? Does this just become a time suck where we have, uh, GPT uh, like person that tells us exactly what we want to hear all the time and we're always just, you know, just gonna waste, essentially waste time it with an in interface.

Ken Jee: Rather than spend it meaningfully, I don't know, I, I am, I'm always wary of those types of things. I think AI has the power to, uh, or the new developments in AI has the power to hack our brain. Like no machine learning has ever done before and the more apps and the more infrastructure and the more entertainment focused things that are built in is a little bit scary to me.

Louis Bouchard: Yeah, to me as well. Just the research on, on Google Maps. Like hurting our memories just because we, we don't focus on driving and trying to find different, um, reminders to, to know where we are and like, just practice our memory that there's a study that, that shows that [00:16:00] using Google maps a lot is just, uh, hurtful for our brain.

Ken Jee: That's very interesting. I didn't know. I know there there's the taxi cab study of the London taxi cabs where they have like dramatically more white space, uh, white matter, which I think is good. And one part of their brain. I think it's like that prefrontal cortex for mapping the entire city, but I didn't realize there was maybe the opposite effect of using Google Maps, which, yeah,

Louis Bouchard: I had sorry for the lack of details.

Louis Bouchard: I don't remind. I don't remind the exact the exact story. Maybe that's because I'm using a GPS, but I it's like if you use. They, they've been comparing like using Google maps versus not using it. And I, there's like a decrease in memory quality or something when you depend on Google maps, just because you basically just follow directions instead of trying to remember and to practice your memory.

Louis Bouchard: And that's something quite scary to me. Just, uh, [00:17:00] like as a comparison, we just technology in general makes us that. We remember that we can find the information we want anytime on Google instead of remember remembering the actual information. So that's like something else that hurts just our capacity in general.

Louis Bouchard: So I like, it's quite, it's quite scary how, as you said, AI can hack our brain to, to improve it, help us, but also make us extremely dumb and completely dependent on these new tools. I think I could say that I'm already dependent on like ChatGPT and cloud just for learning new things, coding for just being more productive for tons of, of things that I do professionally.

Louis Bouchard: I was wondering if that's also the case for you. Do you use, uh, large language models to do anything in your life right now? 

Ken Jee: Yeah, I mean, a lot of pretty much all code I write [00:18:00] sees ChatGPT at some point. Unless it's like super private for the company I work for, and it it, we're worried about those types of concerns to me.

Ken Jee: I, I will say there's a benefit and a drawback, right? So something interesting that I've noticed is to interface with ChatGPT I have to be able to articulate. What I want to write in pseudocode or any of these types of things better than I had to do before. Yeah. On the other hand, my ability to code, it's probably decreased.

Ken Jee: You know, I don't think about the syntax as much. I do still review and say, okay, is there a more efficient way to do this? Whatever it might be. But then I would just, you know, ask it if there's a more efficient way to do this. Right? Yeah. So I think that as long as technology like track GPT is making you think.

Ken Jee: More about the code you're writing or what you're trying to do. That's not necessarily a bad thing. I think that if we view chat to be T like a pair programmer, like someone that we're having a conversation with, it [00:19:00] can actually be beneficial. So. When I'm thinking, if I'm bouncing ideas off someone else, if I'm having a conversation very much like we're having now, I actually find that that enhances my output.

Ken Jee: It makes me think in different ways because I have something to push off of. That's not just my brain going on the other hand, if we go down the rabbit hole of social media or those types of things, I think that they are purely a detraction. Like, yes, I might get slightly more information, but what percent of the time am I looking at social media where it's educational versus entertainment?

Ken Jee: I think we're the algorithms are dramatically more prone to giving us entertainment over education in. Specifically short form content. I think YouTube probably a little bit different. That's one of the reasons why I put all my stuff on YouTube. I think podcasts vastly different. I think podcasts are largely at least for me.

Ken Jee: I only listen to educational podcasts. So if we look at it and parse it out, I think I [00:20:00] really think. You know, instagram, tiktok, they're net negative overall. I think ChatGPT is probably net positive because I do have that conversational dialogue. Still, if we get to the point where ChatGPT is just reading my brain and writing code, then the interaction and there's no thought in that engagement.

Ken Jee: Right when I go on Youtube, it's probably. Break even because there is entertainment there, but I do watch a lot of videos that one, I'm trying to learn how to make better content myself. And two, I do find educational videos there and then podcasting, which, you know, we both do, and which I think is actually probably the most beneficial medium.

Ken Jee: We're listening to conversations and we're forced to think about them, you know, like in dead space, we're educating ourselves. We're learning. I mean, I, I've listened to, I tell everyone about this. The founder founders podcast by David Senra is. I love it. I listen to it every day. I probably listen to four, three or four episodes a day.

Ken Jee: He just goes through and talks [00:21:00] about what the greatest entrepreneurs founders like a little bit about their story and how I like some of the special things about them. And since he's done so many, he's made these connections between all of them. Right. And that's the benefit of all this information is we can start connecting things together.

Ken Jee: So probably a longer answer than you're looking for, but that's how it feels. I think, yes, chatgbt as of right now, I believe is net positive for our brains and for learning. But I could see that being a very slippery slope. The further we go away from us having to interact with it rather than it interacting directly with us. 

Louis Bouchard: Right now, especially with like now being connected with Internet and etc.

Louis Bouchard: I feel like it's, I really like the interface. The chat interface that you interact with, maybe with, with voice, it will be even better, but like, I really like how we can go back and forth. And it's basically both a brainstorming friend basically knows everything. So it's, it's both super useful and interesting to, [00:22:00] to exchange with it compared to, for example, uh, get up copilot that I haven't really used.

Louis Bouchard: I don't. I never liked the, the completion ID. I, I much rather, I, I'd much rather open another tab, like of, of a ChatGPT and just ask, and I'll, I always like to add the, the why question, like just, for example, um, I don't know how to do that function and. Explain like which each parts or some parts that I don't understand about the function.

Louis Bouchard: I feel like it's just a really good to a really good tool to for learning compared to being dependent on a completion thing like copilot, where it's just. It just codes for you without much explanation or without the exchange part. 

Ken Jee: Well, you know, something that I think is really interesting, you know, ChatGPT, these large language models, they hallucinate, right?

Ken Jee: Yeah. If we're talking about purely the benefit for our for our [00:23:00] brains, right? If we could type into a box and we knew everything was guaranteed, right? We'd never question it. We wouldn't think about it as much. We were just copy and paste. Right. I've more recently been viewing the hallucinations like a feature rather than a bug and the feature helps me to question it and think about the problem and that to me actually makes it more useful because rather than just writing and copy and pasting, I'm going in and I'm like, Oh, I have to think about this a little bit more.

Ken Jee: I have to make sure it's right. I have to validate it. That keeps me engaged with it at the same time. And it means that yeah. I'm still thinking about the problem. I'm not just a zombie who writes in a box and copy and pastes. So, you know, maybe I'm, I'm weird in that sense that I'm okay with the hallucinations because it is a forcing factor for me to think and arguably write better code because then it's forcing me to actually pair program because when we pair program, we don't just assume the other person's right.

Ken Jee: We have to validate it. We have to check back and forth. We have to do those things. So yeah. In some [00:24:00] sense, the hallucinations, they aren't, they aren't the worst thing, especially if we're talking about our mental health and our, our, our brain capacity and some of those other things. But, uh, you know, I'd probably like there to be less of them, but, or, or to be more awareness around them.

Ken Jee: But at the same time, I, okay. If you view it like that, it's actually really not that bad for, you know, again, for our brain health and those types of things. 

Louis Bouchard: Yeah. And as long as it's like up to date and it actually is using the right version of your library or anything like this, this is an hallucination that you cannot really deal with easily.

Louis Bouchard: So it's. Like a bit annoying, but the other hallucinations where it just creates something that like, I agree with you. If it allows you to think like I always not necessarily double check, but I, yeah, I check the code if it makes sense. But also that's, that's something that we can do since we know how to code, like we can, if I ask for a function, I check [00:25:00] quickly and okay.

Louis Bouchard: Makes perfect sense. Just copy paste. But then what if the person doesn't know how to code, like they cannot really ask, they cannot really think about it further since they don't know, like how to do better or how to do different. So what, like, is it, is, uh, ChatGPT only, uh, nuisance at this time with hallucination, what, what is your, what would be your thoughts on basically someone learning using ChatGPT without any background knowledge of the task they are trying to achieve?

Ken Jee: So this is going to sound really weird, but I would just open either another ChatGPT window or a cloud window and just have them talk back and forth because, you know, there's enough randomness built. Into that you'd at least have one other perspective, you know, just like you talk with another person.

Ken Jee: When understanding a question, you don't know that their knowledge is correct. [00:26:00] You might assume it's slightly better than yours, but having a dialogue around those things. Could help. You know, you obviously I'd probably try to understand the source material a little bit more or try to understand the fundamentals of these types of things.

Ken Jee: But if you think about how things used to work, maybe 300 years ago, 400 years ago, when there weren't necessarily books or things written for every single thing that would do, you just ask other people, right? And you'd assume that between the knowledge from the wisdom of the crowds, if you ask four or five people, You'd come to a solution that might make a little bit of sense.

Ken Jee: Obviously there was some kind of faulty scientist science historically, but at the same time I think that the, the conversation aspect of multiple people did push them forward. And if you're asking multiple different instances or multiple different large language models for insight or for feedback on something, I think that would push you further than just Running around in a circle with a singular model.

Ken Jee: I don't necessarily know if [00:27:00] I would recommend that for everyone learning, but it is definitely outside the box and probably could be effective. 

Louis Bouchard: Yeah, that's, isn't it what they are doing at OpenAI, just using modal and assembling with like GPT 4 or 3.5 or both, just to compare the results and select the best one, basically, uh, to the best of my knowledge.

Louis Bouchard: Yeah. Yeah. And that's kind of funny and also makes sense if you are unsure, just for example, I recently left the PhD. And I talked about this to my friends, my family, and a lot of people, even, I even always used to ask, what do you think of the PhD to my guests on the podcast? And it was basically because I, myself, wasn't convinced that it was worth it to do one.

Louis Bouchard: And so just taking like all the different opinions, just like the opinion of ChatGPT, if we can say that, it's all going towards the same direction, you can almost take it as a fact. So like, I have much more confidence. On it. So it surely makes sense to [00:28:00] have multiple LLMs and even different ones that if it, if it confirms and produces the same function and there are some same results that it should be fine.

Louis Bouchard: That's a, it's probably a great tip. 

Ken Jee: Thank you. Well, you know, it's interesting. For example, Bard produces at least when I was using it a while ago, I haven't touched Bard and. Five months, but it would give you the top three answers rather than just one answer like ChatGPT, but does, and I thought that was interesting.

Ken Jee: I mean, that might be an approach where it's like, okay, well, what are the, what is the variation in the answers that that you're getting from this single large language model that could still be effective in some sense? I would probably like to see what other I always find it interesting to use. I use Claude and Claude to and GPT four.

Ken Jee: Yeah. And I just compare them a lot. And it's very interesting to see how differently they approach things sometimes. Obviously, Claude is more up to date than GPT 4 is. So that, that can be helpful. But I think overall, GPT 4 is a [00:29:00] better coder than Claude is. So there's, there's trade off between those two things.

Louis Bouchard: Do you use it for anything else than coding? 

Ken Jee: Sort of. So I tried for YouTube those types of things. Just, uh, trying to have the help of that. And I found that it took me more time in revision to get it to have my voice than it did for me to just write it myself. I also realized through that process that I really enjoy writing the scripts myself.

Ken Jee: That's sort of fun. So it was fine. Something that has been useful is when I go through my, I do like an intro call very similar to you with all the podcast guests and I'll take notes and then I can have it. GPT four. I have one script in GPT four that turns those notes into a question set. So I send that questions out over and it's like, Hey, these are some of the topics that we'd like to touch on.

Ken Jee: And it does a pretty good job. It probably saves me a minute or two. But you know, those are the big one. I mean, I [00:30:00] can let me open up chat and see what I have in my stuff. So it's sort of like simple math. I was like doing a probability question. So I play, I like to play this game called Liar's Dice or, um, in Columbia, it's called Dodo.

Ken Jee: And I was thinking about the dice probabilities of, you know, when you should. Like call someone out on lying or not based on a role like this simple stuff like that. Just simple questions that I want a little more in depth answer to. I feel like Googling that and going through the search process is actually less stimulating than asking.

Ken Jee: Yeah. Because then I can see how the, how it's all written out in the work and I'm getting exactly what I want. I don't know if there's necessarily, there probably is still value in the search aspect of going through and finding stuff. But I'd have to look at the, at the science on the, on the research in that.

Louis Bouchard: Any other use case that you had recently other than, than this one? 

Ken Jee: [00:31:00] Um, Oh, summarizing some stuff. So, for example, one of my really good friends, Tina Huang. She recommended I read Aristotle's Poetics and I couldn't get through the, the language in it. And so I was just like, hey, summarize this for me. I probably not the best summary, but it at least broke it down into terms that I could understand and go through, you know, I wouldn't trust it for summarizing necessarily all books.

Ken Jee: It kind of does tend to hallucinate and stuff like that. But if it's just, hey, what is this roughly about? It's generally pretty good. 

Louis Bouchard: Yeah, I feel like it's pretty powerful for summaries, at least to extract the relevant information. Maybe not to, to make like the most interesting one or yeah, I don't know how to measure the quality of a summary exactly, but I guess it makes, it should make a good job.

Louis Bouchard: And I also use it a lot for, for summarizing as well. Just for example, for a podcast, [00:32:00] like if we record an hour discussion, et cetera, I can. Give the, the whisper transcript to ChatGPT and just ask, like, what was this interview about? And can you give me the top points or something to then do the, the thing that I'm worse at to, which is to try to build a good, like SEO title and, and just like SEO things.

Louis Bouchard: I'm extremely bad at that, but I, it's, it's been quite useful to use ChatGPT and just. Just extract the keywords of the episode just from the discussion and figure out the questions asked and etc. It's just, this saves quite a lot of time and also helps me like for things that I don't know. I definitely don't know if what.

Louis Bouchard: It suggests to me as title are good or not for SEO, but I hope it's better than what I can come up with, but I cannot confirm. 

Ken Jee: Yeah, I agree. I find it good for brainstorming like that. Titles is really good. Writing descriptions of videos for YouTube is very [00:33:00] convenient. I mean, the description, it's mostly keyword focused.

Ken Jee: I don't really enjoy writing those. Um, so back to that, for example. Poetics example from Aristotle, I can have chat to PT summarize it. And then I can ask it. Oh, how could I apply this to YouTube storytelling to me? That's that's nuance that you wouldn't really be able to get anywhere else unless you're talking to a friend that has extensively studied that.

Ken Jee: So the ability to iterate and do second order thinking with ChatGPT is very cool to me because, you know, yes, I could start to make those connections, but there's a little bit more lubricant to be able to do that with these large language models. 

Louis Bouchard: Yeah, and what's even cooler is how easy it is to to build them to just now give a link or a PDF and just.

Louis Bouchard: Talk with the PDF. It's just, it's, it's so [00:34:00] cool. It's really cool. 

Ken Jee: I could not agree more. 

Louis Bouchard: And yeah, right now it's, it's just crazy. I don't know if you've tried the new GPT stuff, or I don't know if it's assistant or the same, or they are the same thing, but you can just like upload. I don't know. For example, my friend recently just uploaded all of the, of the Sherlock Holmes book, because they are like over a hundred years old and they are, they are open source or something.

Louis Bouchard: And, uh, yeah, he, he just built a, a quick GPT and I think it took, it took him like five minutes or 10 minutes and now you can just ask it to, to build a new Sherlock Holmes like story or just discuss with Sherlock Holmes, like what he would have said or what he would have. Thought of egg situation. He could do that in five minutes.

Louis Bouchard: That's just incredible. 

Ken Jee: Yeah. I mean, I spent a long time and I still haven't released it. I've been [00:35:00] behind on it, but I built one using Lang chain off of my own data. But man. I could probably just use the infrastructure that they have now to do that and do it in 20 minutes, which is nuts. 

Louis Bouchard: But if you have, if you have an API, you can connect their GPT thing to an API.

Louis Bouchard: So that's like you, you haven't lost your time. That's true. Very true. And what do you mind talking about it? Or is it like a bit underground? Do you, do you want to talk? 

Ken Jee: No, no, no. I mean, so basically I just took all my YouTube transcripts and my podcast transcripts. And I, uh, put them in a, you know, uh, like I, uh, encoded them, I put them into a vector database and then I put a ChatGPT or a GPT for layer over it.

Ken Jee: So you could interface with it with text. So basically it does the similarity search. It pulls in the most similar segments as the question that's being asked. And then from my videos, you [00:36:00] can get an answer to whatever question that you would ask that you would pose of the like. Okay. I called it ChatJEEPT like my last name, J E.

Ken Jee: Um, which I, you know, I thought it was pretty neat. It would also pull in. Uh, the links to the videos as well, which, which is, is pretty handy for the people to listen to. But I just haven't got around to making the video on or putting out and it's going to be outdated technology and like in probably 5, 6 months.

Ken Jee: But, you know, it's still a sort of a fun project to fool around with. I've just been focusing a little bit more on on some other things. And I, you know, that's something that I was okay with. Not necessarily letting it fall through the cracks, but letting it, uh, you know, just putting pause on that for a little bit.

Louis Bouchard: What are you focusing on right now? Is it your work or podcast, YouTube? 

Ken Jee: Yeah, I've been traveling a lot for work. I'm actually also moving in March. So I'm starting to try to plan to, you know, do like a massive life uprooting, which, which is a, [00:37:00] it's a big one. Um, and just doing a lot of introspection about what I want to be doing, what types of challenges and problems I want to be solving.

Ken Jee: Obviously, work is a big one of those. But I think for me understanding, you know, for example, why do I create YouTube videos? Why do I do a podcast? What is the end goal with those? Where, where do those, where do those go? I mean, I, I probably will do the podcast for as long as I live, just because I like talking with and meeting new people.

Ken Jee: And that's a very fun experience for me. Um, YouTube. I also enjoy, but it's not as much as a calling as I feel, uh, with, with Interviewing and studying, and I also just haven't felt like I've been growing and learning as much as I could be or or as I have in the past, uh, in the 3 years before this past year.

Ken Jee: So I just wanted to sort of take a little bit of time to [00:38:00] introspect and figure out what direction that I wanted to go. I'm planning some kind of new changes and and. And cutting some things off, I'm not going to stop YouTube or podcasting, uh, in the new year, but I am planning to travel less to do less speaking engagements, not necessarily think about the other social media platforms that are out there.

Ken Jee: I really do view them as a distraction, even though it might be better for getting reach or some of these types of things. Like, is that really important to me? Do I want to be that have the biggest social media presence? I mean, yeah, it's nice to have followers or any of those types of things, but that doesn't really mean a whole lot to me.

Ken Jee: It's more about what type of impact I can make, what type of things I can build, what I can learn, how do I, how do I create and put interesting things into the world? So sort of refocusing on those things has been a very, uh, it's been a like a light bulb moment for me is okay. Why do you do it? [00:39:00] What do you want to do?

Ken Jee: What is aligned with who you are and what you think? And, you know, talking about before, I mean, I've always viewed data science as and like the coding and any of these types of things is more of a means to solving the problems and answering the questions that I have. Um, you know, at the core, what are the questions?

Ken Jee: What are the things that I'm fascinated with? How do I how do I get to do more of that? Um. And, you know, I would imagine data science data analysis. I will be a part of that, obviously, but it doesn't necessarily have to be just that. And, uh, and that discovery process has been fun and a little bit, uh, harrowing.

Ken Jee: At the same time, you sort of, you're like, oh, I didn't realize that. That's like what the driver was. Um. Yeah, just, just a, a lot of introspection has been good. 

Louis Bouchard: And just for longevity as well. I assume just because [00:40:00] I don't know how you are managing a, I assume full time job as well as the podcast and YouTube.

Louis Bouchard: And doing other projects. Plus, I, I, I know that you, you like sports as well. So that like, even for me, I, I was doing the, well, podcasts, YouTube and PhD, which is, I assume in my case, it was much easier than a full time job. I, I didn't calculate my hours, but I don't think I put in like 40, 50 hours a week. So it's, it was.

Louis Bouchard: And it was also my own schedule and everything. So it was much easier, I think, but still I felt like it was too much and I would, I wanted to concentrate on. On, uh, fewer things in order to succeed and like not distribute my efforts and just do like, just be average or below average on everything. And so I wonder how do you balance your, your work and [00:41:00] life right now?

Louis Bouchard: And is this why you are making the decision? Because it was a bit too much and you, you, you feel like you cannot sustain this for many years. 

Ken Jee: Yeah, I think you you hit it pretty much on the head So I felt like I felt for a while now that because I'm doing so many things I can't do all of them well, and I'm a huge believer in the idea of quality and and As a product, as a differentiator, and I felt like I wasn't able to add the quality to everything that I was doing in the way that I would like.

Ken Jee: So, you know, 1 big bottleneck for me is, you know, I have been living in Hawaii and to get to anywhere is minimum basically 5 hours and I traveled. I've probably been the equivalent of around the world, like, 5 times this year. I've I've been. You know, I've traveled probably two weeks a month for the whole year if you averaged it out and to me from traveling from [00:42:00] Hawaii, I could probably cut the travel time in half and I would get those hours.

Ken Jee: I would get that sleep. I would get that health back. That's a very easy change for me to make going into the next thing. I also have realized that my work and the content that I produce and some of these things, they're, um, yeah. They don't necessarily overlap, and I'm trying to figure out how to do more things where I get economies of scope.

Ken Jee: Yeah. What can I do that is aligned with, uh, my work? What can I do that aligns with the content? How do I fit all those? So they're sort of pointing all the same direction rather than necessarily they're not pointing opposite directions, but rather than pointing, um, Rather than creating the compositional forces.

Ken Jee: So to me, those are the two things that I'm really focusing on is doing more of the things that are. In line with everything else and also traveling less and speaking less, which are things that I, I mean, I like to travel [00:43:00] for, uh, for enjoyment, but for work, traveling is not quite as quite as meaningful.

Ken Jee: So how do I create that balance for myself? Yeah, I wouldn't say that I'm burned out, but I would say that I'm. A little bit over some things, and that's where you have to find the balance of, uh, of just creating the space to, I mean, so many people just get wrapped up in it and they don't think they don't sit and survey what's going on.

Ken Jee: I'm trying to be really good at surveying and taking my time to survey. And then once I have all the information, plan to act rapidly with conviction. 

Louis Bouchard: I'm sorry to interrupt this episode, but I just wanted to remind you to share this episode with a friend if you find it interesting. If you share something good with a friend, they will associate the added value to you.

Louis Bouchard: So it's both beneficial to you, but also supporting my work. Thank you. Let's get back to the episode. That's what I [00:44:00] also discovered recently with the, with the PhD and now trying to focus on. More, yeah, meaningful things, but also projects that are linked together. So just like you trying to, like before I was doing a lot of videos explaining research papers.

Louis Bouchard: And that's, that, that to me was both super interesting because I, I love new research and, and, and just like learning things, but. It was also kind of a two in one to, to stay informed with the new papers and to do my master's and then PhD, like it was, it had a great link together, whereas now I don't need to read nearly as many papers and I'm also a bit transitioning into the more applied love, just like you, of like, I like new products and I like building stuff more than So.

Louis Bouchard: Yeah, yeah. Then reading papers and, [00:45:00] and reading new research. Like I prefer the ones that, that work in the real world. And so I haven't done a research paper video in a few, I don't know, weeks or months and yeah, I'm not forcing myself to do any, but of course it hurts the channel because like the audience is expecting that and I'm.

Louis Bouchard: Doing some other kind of content, just like this podcast, for example, but as you said for the podcast, it's so valuable. I, I just like there, there's no word. I just can't even describe how, how much value it brings just in general to speak with intelligent people or to. People that succeed or like do amazing stuff, but also to me personally, just to be able to speak in English and practice my English speaking, just because I'm French and I live in French, everything is in French here.

Louis Bouchard: And so I, I cannot speak English in a daily, in a day to day [00:46:00] basis. And so having discussions is super helpful, but also talking about artificial intelligence and just brainstorming just, uh, I took some notes while you were speaking. Thank you. A few minutes ago, just for my own company and for my own things, because it gave me cool ideas that I wanted to try.

Louis Bouchard: And just that is so cool, especially because I don't know about you, you, I know that you're a friend with Tina. So I guess you have more friends in the field than I do, but. I have very few close friends that are in technology or like in artificial intelligence. So I cannot, I both cannot speak in English with them because we are like always speaking French and it would just be weird.

Louis Bouchard: But also I can't talk about AI or building things or startup or any of the things that I love talking about or love talking about on the podcast with them. So it's like the podcast, the podcast bring just [00:47:00] so much. Learning opportunities, even if it, like the video isn't even published ever, it's just worth it.

Louis Bouchard: So it's, yeah, it's incredible. I just recommend anyone to try to launch a podcast about your main interest. If you are, if you have anything you cannot talk with your friends about, it's definitely worth it and super fun. 

Ken Jee: Yeah, I couldn't agree more. I actually made a video. Almost two years ago that has no viewership is one of my lowest viewed videos, but it was basically why Everyone should start a podcast because it gives anyone a reason to talk to you I think that that's pretty compelling is that I can now Go to people who I'm interested in talking to who have something that I want to know and I have a reason For them to talk to me Because of promotion or whatever it might be on there.

Ken Jee: And that that is a very beautiful idea is that you sort of have to do [00:48:00] it. I don't know if it's sad. I don't know if it's a good or a bad thing, but there is a value trade off where if I am going to someone for information. What's in it for them? And if you have a podcast, they at least have some story.

Ken Jee: They have something that maybe they said something really smart and they can make a new social media clip, whatever it might be. But there is value generated for them. And I don't think the world should be transactional like that. But inherently, when you're communicating with people that you don't know, it's generally better to be able to offer them something rather than just asking with nothing in return.

Ken Jee: So, you know, with my friends, I don't view it as, oh, we're doing tit for tat. It's just, hey, this is if you need something for me, I will provide it. I will do the best that I can. But a lot of the time, if you don't know the background on someone, hey, have they put in the work? Have they been able to do these things?

Ken Jee: Even if they can show that that's something that you get in return, right? It's this other person has clearly done their homework. They've done all [00:49:00] these things. I'm happy to be generous with my time because I can see that the the things that I'm going to tell them that I'm going to talk with them about are going to pay dividends for them.

Ken Jee: It's going to take them somewhere further. And I get. Goodwill. I get a good feeling out of that. But if, on the other hand, I talk with someone and I think the advice that I gave or any of these types of things are going to be wasted time on my part because they're not going to incorporate it or anything along those lines.

Ken Jee: I think that's where you get a lot of people that just don't want to talk with strangers because that's the feeling they get. So anything you can do to show that The time that someone spends with you or that conversation with you isn't going to be wasted in any way. I think, well, I think a podcast is the clearest one because you put it on the Internet in an effortless way, but there's other ways to do that as well.

Louis Bouchard: Yeah, so just, I guess, doing interviews for a future book or. Like you can come up with a somewhat of a good reason if you are working on a project. [00:50:00] I'm curious. I unfortunately haven't seen that video, but if I had seen it two years ago, I would probably have started earlier. But yeah, I wonder what, what are your, like, can you share a bit more about how to, how, how would you, how you would start a podcast right now?

Louis Bouchard: For example, if someone just heard you and is motivated to, to start talking about golf or whatever topic, how would you suggest them to, to start get interviewees and all the technical or non technical stuff related to hosting a podcast? 

Ken Jee: Yeah. I mean, I don't think you need anything special. I mean, I started my podcast with just zoom and then we've both moved on to Zencastra.

Ken Jee: It looks like, which I think is a superior platform, but you don't need that to start. You don't need a fancy microphone. Just normal earbuds will do. They have pretty decent AI. They can clean up audio well enough now that you just Okay. Have a conversation with someone and record it. That's generally all you need to do to get started.

Ken Jee: I, I think that there's 2 things. So if you're [00:51:00] doing a podcast to turn it into a business or to make money, maybe I would approach that a little bit differently. Um, but if you're just doing a podcast and, you know, it could go on to be a business or make money, but you're doing it out of a pursuit of passion.

Ken Jee: I think all you just have to do is. Figure out what it's about first or what you want to talk about the more narrow, the better. So if you have a podcast about the types of buttons on men's dress shirts. That could be that that's that narrowness is probably better than about than just having a podcast about fashion overall.

Ken Jee: Yeah, because you'll be able to find a core market. You'll, uh, you'll be able to find a very clear group and a clear voice for what you're doing. There's a book called the Common Path to Uncommon Success. And the guy just talks all about podcasting. I think he may, he hosts a pretty famous podcast and he talks about how he approaches podcasting as a [00:52:00] business, how he would do it essentially a step by step guide for, for how he created his podcast.

Ken Jee: I don't know if his roadmap is best for everyone, but studying how other people did it, I think it's pretty compelling. I probably wouldn't be the best person to ask because I started the podcast after I had. A YouTube audience and so it's pretty easy to get audience for the podcast because I was going off of the other stuff that I do.

Ken Jee: And then now I really, to be honest, enjoy podcasting more than I even enjoy making YouTube videos just because it's, it's a social endeavor, not necessarily an individual pursuit. I also don't feel now that the YouTube videos I'm producing are necessarily as they're not necessarily, they're more fishing for.

Ken Jee: unique perspectives than doing really deep research and telling good stories. And I think that YouTube doesn't necessarily [00:53:00] reward. me doing a ton of analysis and research into something, um, and telling a great story, it rewards an interesting topic and entertainment. And that's something that I'm dramatically more interested in learning something myself, telling a story and figuring out what to think about it than, than just creating something that I know people will watch.

Ken Jee: And that, that, that's an interesting trade off. I, you know, that I sort of digress outside of the podcast space for that. 

Louis Bouchard: That's also why I think podcasts are very interesting just because. I feel like, especially in artificial intelligence, YouTube is going in a place that I don't really like, like the, the most popular videos are definitely, I think that if you type what happened this week in AI or something like this last, last week in AI, you will find like thousands of videos and they, like, for example, I, I used to, to review papers and I just remembered that the last one I did is about the [00:54:00] paper.

Louis Bouchard: that was released covering how DALLE 3 was trained, which is super interesting. And back in the days when I covered like DALLE 2, I shared the paper, the paper video, and it, I was the only video on YouTube, the whole YouTube about talking about DALLE 2. And it was just like the only DALLE 2 explained video and it did very well.

Louis Bouchard: And I love making these, but then for DALLE 3 with a bigger audience, I. When I, I did it very fast, like in, in, in one or two days or something, just because I, I was lucky enough to like fall on the paper, uh, just when it was published. And anyways, I made, I made the video and when I was about to release it, I just Googled for fun, like DALLE 3 explained, and there were already hundreds of videos of DALLE 3 explains.

Louis Bouchard: That were released before the paper, just because people put, [00:55:00] used all the keywords, like DALLE 3 explain, or what is DALLE 3, but just showed how to use it on ChatGPT basically, they didn't explain like what the model is or how it was made. And so, I don't know. I feel like YouTube in general is becoming quite saturated where there's.

Louis Bouchard: So many content going out that even if you have a different value, not necessarily that it's a better value, but, uh, like if you provide, you try to provide value, it's, it will just. Not be seen because of all the others that just like do quick videos on like talking quickly about the news and the stuff like this whereas a podcast you Usually I get I assume from what I've seen get a much more engaged audience that listen throughout Compared to just like the regular curve of watching the first 10 seconds, 30, and then dropping the views completely.

Louis Bouchard: I feel like the podcast is much more [00:56:00] consistent where people are just staying because they like the conversation and it's just like chilling with, with the people talking and they are learning and it's. But it has much more value. It's much more fun. And I assume right now it's the podcast world is not saturated enough where there may be a podcast for everything, but at least you are not really subscribing to a podcast.

Louis Bouchard: You are subscribing to like to a podcast about AI. You are. subscribing to the person and you want to hear about this person and, and kind of interact with it like a, in a social, but not social way. Since you're just listening, it's, it's, it's a weird relationship, but you kind of have a relationship with the podcaster.

Ken Jee: Yeah. I think that, so the podcasting space, I think is some people would say it's super saturated. Everyone and their mother is starting a podcast, but I don't think. Many people are, how do I put this? I don't think [00:57:00] 90 percent of those people are not going to create more than five podcasts. I think that's what the data says, right?

Ken Jee: So if you make more than five podcasts, you're already better than 90 percent of all podcasts ever. If you're willing to put in a tremendous amount of work, if you're willing to get to 100 episodes, you're better than Like the top 1 percent of podcasts ever, maybe even, even less than that, probably better than the top 0.

Ken Jee: 01 percent of podcasts ever. So in some sense, it could be a war of attrition. In some sense, I also don't think there's been an identity that's found in podcasting as much as there has been on YouTube. The styles of podcasts are usually just monologue or interview. And, you know, there's some like true crime and they tell really good stories are clearly well researched.

Ken Jee: But I think as a podcast. If you, you put in more time and effort than other people, you can make more of a differentiation than on YouTube. I find that podcasts travel dramatically further through word of mouth than YouTube videos do. [00:58:00] Right. How often, I mean, I just recommended you a podcast today, the founders podcast.

Ken Jee: I tell everyone I'm the biggest. Oh, well, maybe not the biggest evangelist. There's a lot of evangelists for that, but I've told all my friends to listen to that podcast for your podcast. You just need. One of those every 1000 people, and that creates growth that creates, you know, that evangelism creates the word of mouth spread and recurring listeners.

Ken Jee: So, you know, to me, you leverage the network effect dramatically more effective, and that's probably a better way to grow a podcast than it is to even grow a YouTube channel. I mean, YouTube views can be very fickle based on the topic. Podcast views are unbelievably consistent. And that's something that I think about a lot is, you know, do I, do I want to feel this emotional roller coaster every time I post a video or what?

Ken Jee: I like to be able to build something that has steady growth over time, like a podcast, um, that I feel like [00:59:00] I can dedicate time to and not feel like I'm rushed to get something out based on a new trend or something along those lines. And, uh, I've been thinking. Really hard about that is, you know, how do I make content that is evergreen?

Ken Jee: How do I really invest my time and hyper focus on something to do the best possible job on that podcast or on that episode or whatever it might be? You know, that that idea of quality as a product is something I've been thinking about a lot, and I don't think the level of quality I could make in a YouTube video is quite as high as I could make in a podcast if I really did it well.

Ken Jee: And I really dedicated to it. And so, you know, it's just a lot of things to think about related to that. 

Louis Bouchard: All right. It's, it's a very fun difference that I didn't think about before. Just, um, first I, I just want to mention the, there's, I remember that there's the, there's one episode of the Tim Ferriss podcast where he explains everything about starting a [01:00:00] podcast and he gives all his tips.

Louis Bouchard: So if, if someone's still listening that wants to, to, to start one, just Google, uh, Tim Ferriss start a podcast or something like this. It's on YouTube and it's. Like a three hour episode, just him explaining how he started or like how he would start now. It's, it's super cool. And also I completely agree that it's very fun where in my case, like I'm a huge fan of Andrew Uberman and I share his podcast with, well, I share, I talk about him and I, I share like.

Louis Bouchard: him to my friends and to lots of people. But when you, when it comes to YouTube videos, we, what I share is a YouTube video. Like I never really share a channel or a person it's, it's much different. And if you are looking to both to learn, obviously, as we said, but also to build, build an audience, I feel like the [01:01:00] podcast brings.

Louis Bouchard: Much more value to compare to this. Yeah. Yeah. Compared to having like a million subscribers on TikTok, where you just do funny faces or I don't even know what's on TikTok, but yeah. Um, I also wanted to ask you something from what you mentioned, you, you, you quickly mentioned that there, there are two types of podcasts, one to learn and one, if you, you wanted to make it to make money.

Louis Bouchard: And so how would you approach that to create a podcast to make money? And is this even viable? Like, should, should you even consider running a podcast to make money? 

Ken Jee: I hear what you're saying. So I think, for example, a business can create a podcast and that that can have a financial motivation that it needs to have different things out of the gate to be able to get traction.

Ken Jee: You know, for example, Huberman coming out of the gate. His podcast, I wouldn't say it was designed to make money, but it was more of a business oriented thing. He's coming out as a professional teacher, and the [01:02:00] quality of the podcast came out of the gate very high. Yeah. To me, that's something that is a little bit different than, for example, how my podcast started in a relatively informal way.

Ken Jee: And I was, you were, people were able to see me grow as a host, see me grow my communication skills and those types of things over time. Even that founders podcast that I described. He recorded a lot of episodes and then deleted them and then started over because he didn't know what the name of the podcast and those things.

Ken Jee: So once he found his voice, he started, and I wouldn't say his podcast started as necessarily purely a business endeavor, but it was focused on more professionalism out the gate. People are going to, um, see that podcast as a polished product rather than something that's a work in progress. So, To me, if you wanted, if your goal was to say, Hey, this is what I want to do, and all I want to do is do this podcast and turn it into a business, turn this into my lifestyle.

Ken Jee: You probably want to come out of the gate with a more polished product. [01:03:00] On the other hand, I think the way that I've approached things is okay. I'm learning. These the ropes, I'm understanding what it takes to make a good podcast. I'm understanding the video setups. I'm understanding how to do pacing. I'm understanding how to not interrupt my guests that much, which is a big problem.

Ken Jee: Um, and in order to do that, I'm acquiring the skills. If I was to launch another podcast, which. You know, it's always on the in the back of my mind. Right. I would probably try to do it more professionally because I've gotten all these reps in of doing a podcast over the last three years. I understand what it takes and I could take all that learning and make it good out of the gate.

Ken Jee: And, you know, I would expect that I would have more channels and more momentum. Um, and it would probably be expected of me that I would you. Have a more polished product than someone who's just starting. So, you know, you even look at my youtube videos. I've left all my youtube videos up the first ones.

Ken Jee: The quality was so bad. I was so awkward and you can see me [01:04:00] progress over time. Right? So I think that there's a difference between, hey, like, this is a polished product and this is a work in progress. Both are totally fine, but it depends really on your personality, your experience, these types of things, what type of thing you want to create.

Ken Jee: I don't think you need to create a polished product. There are plenty of podcasts. I mean, the Joe Rogan experience was not a polished product when it started and look where it's come now. But I would imagine you get a little bit more traction if that's what you want with a polished product. 

Louis Bouchard: It's definitely important to have a polished product in the end just for people to enjoy quality.

Louis Bouchard: But if you start a podcast in AI or any niche, do you have to be An expert of the niche, like Huberman. 

Ken Jee: I think interest is the most important. I mean, probably would be good to do some due diligence and understanding a little bit how they work. But a podcast can be a purely exploratory endeavor in all my YouTube videos and all my podcasts.

Ken Jee: I hope you will never hear me say that I'm an expert in anything. The goal for me to [01:05:00] do all of this, this research and all these things is to understand and to To get experiences, and I can share my experiences that that's something I'm comfortable doing, but giving advice directly and saying, Hey, I'm an expert.

Ken Jee: You should do this. And that I feel like you're setting yourself up for long term failure or being called out. And, you know, maybe I might have said that in a video about employment or something along those lines. Not my intention. My goal is to always speak from my experience, but Yeah. Just because you don't have experience doesn't mean you can't learn.

Ken Jee: You're probably taking a perspective of someone else who doesn't know much about the field. And you're asking the type of questions that they would want to know as a beginner. And so that can still provide utility that can still be valuable and useful. So I don't think there's any barriers for you to start a podcast.

Ken Jee: I also would say that if you start a podcast, don't expect anyone to listen to it. And that's totally okay. The first episode should be for you. You're going to relish that in two years when your podcast is big. You'll [01:06:00] say, Oh man, I remember those conversations that no one listened to and we had so much fun and they were so interesting and I learned so much and you know, sometimes it's nice to have that again.

Ken Jee: I mean, it's also nice to have listeners and feedback and a community and interaction. But I think that there's joy in all parts of that process for me. 

Louis Bouchard: Yeah, I think that's also why most podcasts up before five episodes. Yeah. Yeah. It's just, I don't know why, I don't have this fear personally, but I, I know that a lot of people try things and see that it doesn't succeed so they stop, and for me, uh, I would have never got where I am now just because, uh, on YouTube I had it.

Louis Bouchard: I don't know. I think I reached a thousand subscribers like after eight months or so it was pretty late or maybe a bit less, but I, for the fear for the first few months, I had zero viewers. I also was like completely anonymous, so I didn't share it with my friends, family, nobody, nobody knew except [01:07:00] my mother because she heard me, but nobody knew I was recording videos and, and doing.

Louis Bouchard: So it was kind of hard for me to share since I didn't have anywhere to share it. So nobody saw it. And I still just kept going and growing. And just like you, I didn't remove any first video. So it's horrible. And like, I say that in a, in a good way, like it's, it's so cool to see that it was so bad that now it must be incredibly better just because of the, like, if you can see that it's bad, it means that you.

Louis Bouchard: You are much better now, I guess. 

Ken Jee: Well, I think it, it really matters on how people define success. So if people are saying, oh, it's successful if it's, if it gets 100 views, it gets a thousand views. That's fundamentally flawed. So well, at least from my perspective, it's fundamentally flawed. I've never had a podcast episode that I did where I walked out of it saying that was a waste of time.

Ken Jee: I've never been on someone's podcast where I walked out and [01:08:00] said. Well, maybe one or two that was wasted. This definitely isn't one of them. Don't worry. Um, but the idea is that, okay, I got to meet this new person. I got to work on my podcasting skills. I got to create something and put it into the world that will be there for as long as.

Ken Jee: You know, the hard drives will maintain it wherever they're being hosted that to me is success. That is a success. You're putting something out to get feedback. You can get better. That's in the early stages. What it's all about. It's, you know, I started, I've told the story a couple of times, but 1 of the reasons I started YouTube is because I had an interview.

Ken Jee: Where it, there was a question prompt on the screen and I just had to talk into a camera and I watched it back and it was, I thought, man, I would never hire that person. That was the most awkward thing ever. I looked like a zombie. I hated myself in that and. YouTube is a way for me to work on those skills.

Ken Jee: It's a way for me to improve my ability to [01:09:00] communicate and articulate things and become a good storyteller and be engaging. And to me, same thing with podcasting, same thing with any of these things. If you're viewing it as, oh, it's a failure. If I don't get viewership, you're missing out on all of Successful aspects of that.

Ken Jee: You the first thing is you did a podcast. Do you know how many people think about doing a podcast and don't a lot? I would bet anyone who's remotely extroverted or even introverted has thought about making a podcast. I bet I could do that. But how many of them have done it? How many have them not? You already won.

Ken Jee: It's like, you know, being born, you know, if we're talking about like sperm and eggs, like you won, you won one race already like a freaking massive one out of like, Millions and millions and millions. So maybe a weird place to go, but, but it's all a framing thing, right? 

Louis Bouchard: Yeah. The people that knew me back in when I was younger, uh, like high school and, and, and before I would never [01:10:00] think that I could do that in, in the future or just like speak in front of people or just do YouTube videos.

Louis Bouchard: All that stuff was definitely like extremely far from what I was like, I was like the, the shy. Kid that didn't talk to anyone except his very close friends. And yeah, I, and, and still I'm doing this right now and I'm like putting me in front of people online and et cetera. And it, it like, I don't know, my message is just that if you think you cannot do it, it's, it will be weird at first just to talk and being recorded with a camera.

Louis Bouchard: And also listen to your own voice, that's like maybe a hard part to, to get through at first, but then I think anyone can do this. I'm, I know that it's like something that all comedians say, but I was really the worst person to do that before. Like I, I would never have [01:11:00] seen me do that and I'm just doing it.

Louis Bouchard: I am enjoying it a lot. It's, it's still challenging to me, like even. With you, it's, it was a bit different because I feel just by listening to your podcast and etc, I feel like I already know you a bit and like I, we have some kind of weird online relationship where I've, I've listened to you a lot without talking to you.

Louis Bouchard: So it's, it's definitely something unique, but still, uh, for, for most podcasts, I'm super stressed out before and. I still, then as soon as I start talking about, uh, with the person, I, it just, there's a switch where I'm just having a nice conversation. I'm learning. It's fun. And obviously it's a bit hard to keep, try to keep track of the questions you want to ask and etc.

Louis Bouchard: That's like the small challenge, but it's, it's just so fun. I don't know. It's, it's totally worth it to try. And I'm pretty sure that if you, even if you are super shy right now [01:12:00] and like you are not comfortable to do a presentation in front of your class, you can definitely do it, host a podcast and, and be successful.

Louis Bouchard: I'm sure like there's. no reason to not try and not just have a fun discussion or at least worst case you just call some people and you have discussions and in the future you start recording like much later on if you want to do a podcast but you start practicing with talking with Someone that you don't know yet, I guess that would be a good way to start.

Ken Jee: I'm a big believer, whether it's with podcasting, whether it's with landing a data science job, whether it's with almost anything in life, the only people that fail or truly fail are the ones that quit. So the only way that you do not land a data science job or something along those lines is because you stopped applying.

Ken Jee: There's varying speeds to landing that job to finding success in podcasting or any of those types of things, right? How efficiently you do it, how much [01:13:00] you learn from each failure or like mini failure or like thing that doesn't go your way. Right? Those are the only things. That are truly bottlenecks for you.

Ken Jee: You know, I recently had a high school friend say that, you know, a lot of the guys that I went to high school with, they they're impressed with what I do, you know, they, they respect a lot of the autonomy. I have that. I was able to build a lot of these things out on my own and that, you know, I'm able to have an audience and, and these types of things.

Ken Jee: So it'll, it'll, it'll, it'll, it'll, it'll, it's like a very nice lifestyle. Right? But the thought I had is literally every one of you guys could have done this. I'm just the person that didn't stop or, you know, like everyone has those ideas. I first acted on the ideas and then I didn't stop whenever there was any adversity.

Ken Jee: I just kept going. So there's a great book called The Dip by Seth Godin. Essentially everything goes through this lifestyle where you get early momentum and then there's a dip, right? And then you see this giant exponential, [01:14:00] uh, essentially spike. And The people that are successful just get through the dip, right?

Ken Jee: They get through that down part where the excitement and the motivation and all these things are low. That's what creates success. It is always a war of attrition. As long as you are learning along the way, if you're not learning, if you're not iterating or not improving, you'll stay in the dip forever.

Ken Jee: You know, the idea. That insanity is just doing the same thing over again and over and over again and expecting the same result. But if you're observant, you're introspective, you're willing to make changes to improve. It might take some time, but you always get to the other end of that thing. So, you know.

Ken Jee: To me, that is the most important lesson. And, you know, there's dips that you're going through now. There isn't just one dip, I guess, but I got through the earlier ones. I'll get through any of the future ones that I have. And that, I think, is the philosophy or the mindset that is, you know, that maybe not everyone shares and [01:15:00] maybe it isn't right for everyone, but that's it.

Ken Jee: That's my perspective at the very least. 

Louis Bouchard: Oh, no, I think it applies to everything, like even working out, it's, I, I'm a strong believer that anyone that works out for at least like two months straight will get addicted. They will love it. They will like it. They will just continue forever. The dip after, I don't know, like one week to two, three weeks is where people stop.

Louis Bouchard: They just don't see progress immediately. And they, they don't like the gym yet. And they get discouraged, but I, yeah, I've, I don't know. I I'm just convinced for, for working out that if you get there, if you keep going at it long enough, you will love working out. It's just like so beneficial, so fun. So everything just positive net positives.

Louis Bouchard: And I'm pretty sure it's the same, as you said, for podcasting, for YouTube, for writing a blog newsletter, like you first, you, [01:16:00] you might suck at when starting, obviously it's completely fine and you will just get better as long as you are open to, to, to getting better. And if you get through this period where nobody sees your content, but still keep going.

Louis Bouchard: One day someone will see it and it, this person will share it with their friend and it will just exponentially grow. 

Ken Jee: Well, you know, a while ago I started something and, you know, I haven't been good with upkeeping it, uh, but I started this initiative called the 66 days of data. So basically the idea is every day for 66 days, you just spend five minutes a day learning data science.

Ken Jee: That's all you have to do. And you just create the habit of learning every day. So 66 is special because in some research, they basically found the average amount of time that it took to ingrain a new habit was 66 days. But 66 days, at least for me, was the amount of time that it took to get past the dip.

Ken Jee: For that learning and building that learning habit. And [01:17:00] another aspect of that is that you would share socially what you were working on, what you learned that day, and that creates an accountability factor. And it also gives you a log to look back on to see how far you come. I think one of the reasons why we get into that dip is we feel like we haven't made any progress is we feel like we're not going anywhere.

Ken Jee: But I If you really think about it, if you're producing a podcast, if you're doing work, like you have all the lines of code you've written, you have all the podcast episodes you've recorded, you have all the new people you've met, you've made unbelievable progress. It just doesn't feel like that because the metric you're evaluating it against is often the wrong metric for the time period that you're that you're pursuing it.

Ken Jee: So I think that, you know, doing things like you don't have to do that challenge. I haven't. Been active in that for, for quite some time. Unfortunately, I just haven't had time. I still think the idea is great. I was like, ingrain this habit, share with your friends, create this infrastructure for yourself to, to help you succeed.

Ken Jee: And that's what willpower is so fickle. Right. The [01:18:00] idea that I'm going to be motivated to do this, I'm going to do it, but creating habits, doing things consistently, having other people hold you accountable, seeing how far you've come. Those are what make it so you can achieve greatness. And I think that that's what I'm seeing more and more is that the special people are not.

Ken Jee: Some of them are more dedicated. Some of them have this immense fire, but a lot of the time it's just they have a better operating system that they've created or they've downloaded, uh, to be able to achieve that success. 

Louis Bouchard: Yeah, a very good conclusion. I've always also shared that just consistency is key.

Louis Bouchard: You there's nothing other than consistency. Consistency. I think other than like wanting to improve. There's, you don't need the skills at first, you don't need anything, you just need to want to get better and keep at it, like being consistent every day or every week or whatever the frequency, but just stay consistent.

Louis Bouchard: Is there anything you'd like to [01:19:00] share to the audience to check out about you, the KNN podcast or your own work? Anything you'd like to say? 

Ken Jee: Yeah. Feel free to check out the KNN podcast, Louis I'm going to have you on hopefully, uh, very soon here as well. Uh, aside from that, just, uh, the YouTube channel, um, you know, I've been trying to experiment on some, some different styles of content.

Ken Jee: I view YouTube as a creative medium where I can tell the stories that I'm interested in telling. I've cared a little bit less about pure viewership and those types of things over the last year or so. And honestly, that's been good for my mental health. I find it more interesting and hopefully the stories are better and more interesting as well.

Ken Jee: So if you want to reach me, find out anything about me, YouTube comments are by far the best place. I try to still respond to all of them. Um... And yeah, this was amazing. Thank you so much again for having me on.

Louis Bouchard: Yeah. Thanks to you. Thanks to all the value you taught me and just the [01:20:00] amazing discussions. It's rare that I can also talk with someone else doing podcasts.

Louis Bouchard: I don't know if this will interest many people, but I feel like it's, but yeah, we discussed it, but gas is super valuable and I think it's a. It's a relevant skill, just communication in general is a relevant skill to develop. And it was surely interesting to most people to listen to that. I hope so. Let me know.

Louis Bouchard: But yeah, thank you very much for your time and for all your insights.