Welcome to Season Seven of Edtech Insiders. The show where we cover the education technology industry in depth every week and speak to thought leaders, founders, investors, and operators in the edtech field. I'm Alex Sarlin.
And I'm Ben Kornell. And we're both edtech leaders with experience ranging from startups all the way to big tech. We're passionate about connecting you with what's happening in edtech around the globe.
Thanks for listening. And if you liked the podcast, please subscribe and leave us a review.
For our newsletter events and resources go to edtechinsiders.org. Here's the show.
Per Borgen is the CEO and co-founder of Scrimba, an interactive code learning platform with over a million users worldwide. He's also a web developer and coding instructor. Per's passion is to help people change their careers and their lives through learning how to build software just as he did at age 29. These days, he leads scrim as AI engineering initiative, the company has partnered up with hugging face and Lang chain to create tutorials and courses that help web developers learn how to
build AI powered apps. Through these partnerships Per gets a front row seat for how the AI infrastructure companies lay the groundwork that the rest of the industry builds on top of Per Borgen. Welcome to Edtech Insiders.
Thank you. Great to be here.
It's great to have you here we are coming off of the AI and Edu conference where you talk about some amazing things happening at scrim bar, and how you're using it to teach coding. Give us a little bit of your background of Scrimba and your story in edtech. Before we get started, I know you have a really interesting backstory. Sure. So
Scrimba was actually the brainchild of our CTO Syndra, who back in the days created his own programming language called imba. So that's the name similarity. And he really wanted a way to teach in butter people, though, he found that writing documentation was really boring. And that recording videos was a really tricky process, like creating a good five minute tutorial and put it on YouTube takes like an hour of your time to make it polished. And also, there's no
interactivity and videos. So he was like, How can I make this better for me want to teach my programming language and for students or learners who want to like, interact with the code when I'm teaching it. And then he did invent this format, which kind of mixes a code editor with a video so that the learning experience for learners are much better since they can interact with the code whenever they want. And it just so happens to be that the act of creating so called scrims is super simple.
It's just opening up the editor hitting record starting to talk as your coat. So he showed me that prototype to me back in 2016, it was, and we had done a startup before, which failed. But we knew that we worked really well together. So he showed it to me and was like, What do you think about this, and I was like, This is amazing that he shouldn't just use it to teach him by should use it to teach JavaScript and web
development and everything. And I just happened to be really into teaching coding at that point as well. So we kind of teamed up yet again, and hit the ground running from around the start of 2017. And yeah, now here we are, like to seven years later and got a million users and lots of people who love our, our product and employees around the world. And yeah, just trying to make the best cold learning platform possible.
Yeah, I mean, your point about interactivity is so key. We are also used to video as a means of learning these days, and you there is a lot of power and video, it's visual, you can obviously, you know, pause and rewind and stop and follow along. But the difference between a passive video and an interactive experience is huge. And you know, decades of learning science says that as well actually doing is by far the best way to learn. So I think that's sort of baked into what
you do. It's grandpa. Exactly.
And the fact that we can use the same medium kind of the audiovisual video ish thing as an interactive component as well, that actually flips around on the entire teaching approach. So when we get new teachers, we have to, like Teach them to like, no, don't just explain your code, challenge the student again and again and again, and ask them to jump into the code. So it's kind of it flips the go from being a monologue to being much more of a dialogue where you throw the ball back and
forth. And then it took us actually a couple of years to figure that out ourselves. Yeah.
You say you have a million users tell us who are these users? Are these individuals learning to code on their own? I know that you have bootcamp models. Do you work with b2b like give us a little bit of it over view of how the business works.
Yeah, so it's mostly b2c for us. And these are self motivated learners, for the most part from all around the world. So India is our biggest market in terms of traffic. The US, of course, the Big Gay European countries, Brazil, like all over the world, Nigeria is a big one as well. And mostly b2c, so people pay out of their pockets.
But of course, we have a lot of people using us who are just happened to be hired at various companies as well probably pay or their employer pay, and also a bunch of students who go to colleges and universities.
Yeah, it makes sense. It's a really exciting approach. And recently, you've really, really embraced the AI revolution more than I think most companies I've talked to, which is a lot of companies at
this point. And you've been talking about this sort of idea of learning to code with AI, as a way for non coders or, you know, beginning coders or people, children or people who just don't want to sort of slog through the beginning aspects of coding can use AI to experience the more fun and engaging parts of coding directly. And I find
that super interesting. Tell us a little bit about your approach to using AI to making the act of learning coding more interesting and more useful for a wider variety of learners.
Yeah, so what we're doing with this AI coding for non coders course that we've launched, and we're working on updating it, and adding more to it, is to actually kind of front load all the fun, and delay all the boring parts, because that is now possible. Because imagine that you're a fresh developer, or you're starting out at learn how to code, and you want to build like a blackjack game, for
example. So in order to do that, you have to learn about all of these quote unquote, boring things like loops, and conditionals, and data types, and all these theoretical things that people might not be interested in, they want to play blackjack in a game they've created themselves. So with AI, we can kind of skip ahead and jump over all of those theoretical parts and get the AI to create the first version of
the code. And then you can kind of give them a challenge, like, Okay, so here, the AI has created this blackjack game. But what if you want the message that appears when you win to be a little bit different, where in the code, you look to actually do that. And then the student has a motivation to actually look at the conditional and figure out where that console log or print statement or
whatever exists. And then gradually, they can learn as a side effect of these small quests about all of the theoretical, quote unquote, boring parts, which I think are fun. And eventually, most developers do end up thinking that the logic is actually part of the fun. But for beginners, it's not that easy to see that. So yeah, I'm really excited about this way of learning. And I think it applies to not just coding but also a bunch of other subjects as well, I
couldn't agree more, it is such an amazing insight to see that one of the things that AI does, especially because AI can write all these copilot programs and AI is actually very, very good at creating programs, the idea that it can take that learning process and really turn it on its head rather than having to build up from you know, fundamental building blocks and these sort of abstract concepts, you can jump right into the interesting parts and then learn the fundamental concepts as you
need to to actually be able to do what you want to do. That just changes the motivation enormously. It makes it instead of having to have this lots of motivation at the outset and say, someday, I know this is going to pay off, but I'm going to try to do it. You got to play right up front, when it's like it makes it feel almost more like a sport or something or you can just pick it up and go. And then you have to learn how to actually get good. Exactly.
And I think even further, like we don't even have to stick with one specific project for the course like for example, Blackjack, because might not be the students interest in blackjack, they might rather be interested in dinosaurs, and thus you create a little dinosaur game and get the AI to create that. And that kind of personalizes the learning experience as well. And yeah, we haven't done exactly that. But that's one of the things I'm thinking about as we we build out this course further.
When I hear you talk about new coders being able to develop things based on their interests. It reminds me of a concept that I've heard you talk about, which is the new role of an AI engineer. Tell us what an AI engineer is.
Yeah, so it's a pretty new role and putting you name AI engineer. And most people think it refers to the people who train these neural nets and write the code for them. But it's actually not that's more of a machine learning engineer. So are the AI researchers. The AI engineer refers to more of the hacker builder types who take these
tools. For example, the chan GBT API or the Paul two API or all of these open source API's from hugging face and all of these new tools that have popped up over the last few years, and use them to build new products, because it's in that implementation layer that the world essentially changes is when you take these powerful API's and implement them into all of the world's products, or companies and industries, because that is, without doubt going to happen and must happen.
If you don't use AI in your product over the next few years, your product will become obsolete. So analysts projected and a lot of people think that this will be the most in demand engineering role of the decades. So we are going all in on this and are soon going to launch a full AI engineer path I'm screaming out.
I love this idea. So tell us a little bit more about what are the specific sort of skills or mindsets that you're thinking about when you're trying to create an AI engineer with your Scrimba pathway? Yeah,
so the first thing is that you need to be a coder. In order to be a good AI engineer, you can't of course, hack together things with chalky beauty as well. But the better developer you are, the better have an AI engineer you'll be. So it's for people who already know how to code. And what we teach them is, for example, how to take the chance completions endpoint from open AI and use that in your app to, for example, create an internal knowledge database for a
company. And to do that you very often have to use a new set of tools that most web developers today haven't been exposed to yet. For example, there's the concept of embeddings. And vector databases, which is something you use in order to give your apps specific knowledge. Because TPT doesn't know about the internal documentation of the company or building the app for you need
their data. And you need to transform that data in ways so that it works together with the large language model underneath that creates this chat interface. And another thing is to build so called agents, which are essentially programs that can make decisions for you and also interact with the world for views that is called other API's. For example, booked tickets for the user or search on the worldwide web for the user or close a support ticket
for you. Yeah, you get it. But there's a bunch of things here, which normal web developers who quote unquote, don't know. But that AI engineers really need to master. And this is a really a jungle these days, there's not a lot of best practices, you can spend hours and hours bashing your head against the wall. So we are trying to kind of condense the wisdom into a course that takes any web developer and turns them into a competent AI engineer.
It's so interesting to hear you talk about this, one of the things we heard at the conference was the idea that AI raises both the floor and the ceiling for developers and for all of us. And when I hear you talk about the two different things, you've been talking about raising the floor, meaning that a novice coder can walk in and say, I want to create, you know, Pokemon, but for dinosaurs, and they can literally tell the AI to do it, and it can create a
program. And then they can use that program to learn how to code that's very much raising the floor, you can do amazing things with that. And then when you're talking about what this AI engineer is really about raising the ceiling, taking existing coders and engineers, people who understand development, and saying, there's a few concepts that you really now need to know like the ability to take new data and embed it into a machine learning model or to use agents to access
external tools. And those are not in your normal toolset. As a developer, you didn't learn that in college, you didn't learn that when you were doing your own training online. But now you need to know it, because this is what you know, really is going to happen. So I love this
combination. And I think scrim was really out front in both of the sides of this, you know, using AI for novices, and then taking existing developers and turning them into these super developers, these AI engineers who can really be very, very adept at using new tools in their work.
Yeah, I totally agree with that, that raises both the floor and the ceiling. And another metaphor I've heard used is, it's a tide that raises all boats. So if you were at 0.1x, engineer before, and you just knew a little bit, well, now you have 10x, and you're a 1x. Engineer. So some people may think that yeah, well then what's the point learning to code, you can just learn a little bit of basics and then
boom, you're a 1x engineer. And you got to remember of the tide raises the 1x engineers as well. So they are suddenly now 10x engineers, and can do much more than you. So some people kind of think that AI makes core technical skills and coding skills obsolete. I think it's actually the opposite. There's never been a better time to learn to code because you're going to get so much back from those skills because you have kind of an amplifier at your
side. So once you learn a new technological subject, you can accelerate even faster with these new AI tools.
Absolutely. One thing that I find very interesting about Scrimba and I think it's maybe hard for people to envision this, if they haven't seen it yet is that these scrims that Scrimba creates these sort of combination, interactive videos that use coding and video together can actually be embedded, you know, they're in Scrimba, but they can be embedded elsewhere as well. It's almost like a YouTube embedding, you can just have a scrim somewhere and learn to code in
line. And you've been using that technology to forge partnerships with really amazing industry players like Lang chain and hugging face. Tell us a little bit about what you've been doing with these leaders in the industry, and how have they impacted your business and your platform? Yeah,
so both Langchain and Hugging Face, and now either link to scrims or embed scrims in their documentation. And we have other partnerships as well. So we're using this to kind of sharing our secret sauce with other partners, we think that's a net win for both them and us because it's more awareness for
us as well. And we've been doing this, especially in the AI field, because when we're creating this curriculum for the AI engineer path, or for other AI engineering courses, as well, what we found is that this world
moves incredibly fast. So it could be that we start with a course, and we work on it for three weeks, and boom, suddenly, a new version of the API comes out a new version of the SDK, a model changes and just or new new pattern, or just like a best practice emerges, it's so hard, we got to stay at kind of the bleeding edge. And the best way to do that is to actually have partnerships with the players who create who host the models, or create the libraries. So
that's why we've done that. So these partnerships help us ensure that we're actually teaching the latest and greatest and not are falling behind, which is really important these days.
Yeah, one of the other things that I've heard you talk about that, I think is a classic issue in instructional design is that, you know, there are some subjects that they call
the Evergreen, right? You teaching Roman history, you pretty much know Roman history, there may be new things once in a while, but it's pretty much evergreen, this is the least evergreen topic, I think I've ever seen, you know, on a weekly basis, all the tools change, the language is changed, the techniques change, new companies
come out. And as you just mentioned, it moves so fast, and mentioned how you have sort of adopted some new ways of creating courses to make sure that you're not caught with a course that is too quickly deprecated. And made, you know, out of date. Tell us a little bit about how you've done that, because I think others listening to this are probably experiencing the same problem with AI, it just moves so fast.
Yeah, we actually have to adapt our learning approach quite a lot. Because normally our courses they are very intertwined. So one section and a course takes one project and teaches you how to build it from A to Z, and actually forces you to build it yourself as well. But that becomes tricky when kind of what you do the for loop, or the the method you used in lesson four is repeated or effects the code in lesson 17.
When it's intertwined that like that, you kind of paint yourself into a corner, because you can bet that that method you used in lesson four is going to change.
So we had to do what we call, like more compartmentalization, where we actually step a little bit away from the overall hey, let's build a cool project from A to Z. That's the motivation there have to go still maybe a little bit more theoretical than that and, and kind of pull out the things we think will be more or less evergreen, and teach those in separate, and then have many projects here and there. Essentially just separating out compartmentalizing everything a
bit more. And that's something we would never do with a regular JavaScript course. But for now, we have no other choice,
right? I mean, the fact that JavaScript at this point is considered so stable compared to this stuff, it makes sense it is some of these coding languages have been tried and true, and they don't change very much anymore. And that is certainly not true of all of these AI, toolkits and languages and the API's, it's really quite
amazing. So one of the things that I really admire about what you're doing at scrim but is that you're not content to just sort of integrate something with AI and then move on you're really really staying very cutting edge on the entire industry and who's doing the most cutting edge things even if they're not in education. So you we mentioned some partnerships with people like hugging face, it's like the leading model repository and does all sorts of things. Lang chain is a very
cutting edge framework. Tell us about how you stay on top of the most interesting things happening in AI generally, and then how you build these partnerships. I think it's something that very few other Ed Tech's have really done.
Yeah, so staying on. Top is Twitter is of course that important or x is an important source there. And actually quite funny is that I got both of those partnerships with Lantian and hugging face. They started via Twitter, just by interacting and being part of the community
and chatting with them. And actually a used thing for blockchain, I created a tweet about how easy they made it to create an AI agent in the browser, which I just found super fascinating, like 10 to 20 lines of code, you can do that. And that just put me in touch with them because they liked that content. And then we just got started chatting, and we saw Okay, actually, you guys want to have scripts in your documentation? And then be like,
Yeah, of course. And, like, yeah, we'd Of course love to create that, because that gives more awareness around scrim, but it's a win win. And then we also started working on a course together. So that's coming out right now, actually, this week.
And as we go forward, I think we'll just continue to work with them and actually help them improve their documentation, because that's also something we do, we try to learn their technologies, and then we give them feedback on Oh, actually, this was really clunky explained, or this should actually be before that section. So and even I know that they've taken our considerations and look the implementation of the
library as well. So I'm gonna, we're not going to take credit for for any of the clever technical solutions they have done. But it's definitely a dialogue there. And we're actually kind of closer to the metal than I would think beforehand, I wouldn't have expected they would be this close with the maintainers. But it's been so cool, just get a front row view at how some of these people develop these tools that affect so many developers and, as a consequence, affect even more industries and end
users. 100% You know, the way you're describing this, I think, is so powerful for our entire community, frankly, for the Ed Tech community, it's like, you know, you think of some of these big tech products or these very sophisticated, you know, development teams that are doing things infrastructural things for software as being, you know, sort of on their own island, but really, they have as much incentive as anybody probably more than anybody in making sure everybody learns how to use
their product. Totally. Right. And as education experts, so you know, I mean, you have been doing edtech, for seven years, you've taught a million people how to code more than a million, you know, for you to come along and say, Hey, by the way, we're, I'm really good. We're really good at teaching people how to use new languages. Here's some ideas about what you might want to do to help people learn it. I mean, I'm sure they're all ears
to that kind of thing. And, you know, I think that we sometimes I don't know, maybe I'm speaking for myself here, but sort of have a little bit of an inferiority complex in edtech, we think that we're sort of off doing our education thing, but there are other people doing this really intense, you know,
world changing work. I hope that story you just said, it really makes me feel very positively about how we can all work together and make these incredible technologies actually usable, and actually, even as you say, maybe even affect how they're developed themselves.
Yeah. And in my experience, we very often always get like a positive reply, when we come with good intentions, like, hey, actually, this wasn't pretty hard to understand, can you update this and that? And so I think like, that's something teachers all over the world should like, just remember, if they struggle to, to learn a new tool, you can bet there other people struggling to do it as well. And then you're doing the world a favor. If you reach out to the company and help them improve it.
You're doing the world a favor and doing them a favor, because they're having all these people.
For the most part. Yeah, exactly.
So let's talk a little bit more about the hugging face partnership, just because hugging face is a really fascinating, you know, world, it's almost like the new StackOverflow or the new GitHub, in a way because it really just sort of maybe you do a better job than I would tell our audience a little bit about what role hugging face plays in this AI landscape and how it sort of is enabling all this open source models to be available.
Yeah, I think you said it. Well. It's a new GitHub, it's GitHub for AI. That's where most of the big players host their open source models for anyone to use. And hugging face takes their models and exposes them through API so that anyone through connecting with the hugging Face API, the other SDKs can build products with these open source models. And it's been fascinating to see how quickly open source has caught up with the closed source
models. Even though I'd say that the closed source models have a lead, it didn't seem like that for like, at the beginning of this year, it didn't seem likely that the the open source models would be as good as they are now. So maybe it even surpasses the closed source models at some point that I'm not sure. But I think it's important for us to teach both because we don't want a world where all developers just know how to use closed source AI model. We want them to be able to use open source
models as well. That's been the kind of the mantra amongst developers for decades. Now. That open source is the Way to go. And that's how we share knowledge and share value with each other without charging for it. Yeah,
and then without charging for it is a big part of things right now because you know, we have this, these closed source models like GPT, or palm, or I think Gemini is coming. And you know, all these ones where you can use them. And they're super powerful, and you can build right on top of them. But you're basically paying the piper, you're paying, you know, open AI or Google, you're paying for every API call. And then you have these increasingly powerful open source models like calkin.
And there's hundreds now at this point, that are cheap to free, I think they're free, basically right to use, you
gotta pay for you, if you want to third party to host it for you. And if you've course, got to host it somewhere. But if you have your own server architecture, and you can do it a lot cheaper than if you just query external API that host that it does everything magically for you today, a lot
cheaper. And actually, what's interesting is that you can even get it 100% Free by offloading it on the user's device as well, which is something I haven't faced us to super interesting transformers JS library where you can download models in the browser and run them and actually do quite sophisticated
things. So of course, far off, like the chat GPT capabilities, but you'd be surprised what you could do in terms of like, get text and object detection and images, computer vision and with audio transformations, and many cool things. And then there's no cost for the developer, because everything is running, the inference is happening on the user's device. And
that is so powerful for an educational use case, because I mean, you've been making Coursera courses or Coursera projects, and anybody who's worked in trying to make sort of large courses for lots of people you always hit up with against this problem of, Well, should we asked them to use the, you know, official tool, even if it's proprietary, which means either they have to pay for it, or we that you know, the EdTech has to pay for them to have
access to it? Or is there a really good proxy, something they can use that they can learn a ton from, they can use much less expensively or even for free. So the idea that you can you know that these options are already both available, I find it very exciting for as an instructional designer, just because it means you have lots more options about how to get
people to learn. Exactly. Do you see the potential in the future for education companies to use hugging face and these open source models to be able to teach AI engineering at a less expensive way or in a different way than accessing GPT? Totally,
I think anyone who teaches AI engineering in AI should have this kind of asset in the back of their heads from the get go, like what close source models and API's should we teach? And how can we integrate some parts of open source as well, just to raise the awareness amongst the learners? Yeah,
it's a really exciting world. And so you mentioned that there are some other partnerships either in place or that you're starting to think about with some more of these, you know, cutting edge AI companies? Are you comfortable talking about any of them on the podcast? There's no, no promises either way. But I'd love to hear how you're thinking about it just because you are so in the weeds in a good way of knowing what sort of coming next. Yeah,
so we're talking with a bunch of companies now, and we're getting very positive results. So I'm confident that we'll get more partnerships over the next few months. I don't have any names to share. Unfortunately, I got to keep it to my chest until we gotten the agreements set. But we're going to share it on our blog and on Twitter, for sure. Once we get more partners.
That's exciting. Let me even zoom out and ask you, you are somebody I talked to who I really feel like you have your finger on the pulse of the AI space. And I know this is that's a relative thing. There are so many people trying to follow this very fast moving world. But I just want to ask, before we get to our sort of final questions just in a broad way. You know, where do you see this field going? Who's doing incredibly interesting work? What do you think is the
next big paradigm shift? How is this AI world developing in a way? What could we do to sort of take out our telescope and see where this is all going? Yeah,
so I think this is covered in one of our modules, one of the most important modules of our upcoming AI engineering path, which is what I mentioned earlier, AI agents. So generative AI is all about like creating things for your producing texts, producing images, producing audio and all that is great, it's going to continue, I think on an unprecedented speed and gotta blow our minds every quarter
going forward. But also, it becomes interesting once the AI can start interacting more with the world on your behalf, like figuring out what are the I really want to go surfing next month, but I only have $300 to spend, like what are my options and then and it can research how to travel to places what the weather is there if there's available hotel space there and Like, give you maybe even make
the decision for you. And like, Yeah, I think the AI agent space and where the ad becomes more interactive like that it's super, super interesting.
That makes a lot of sense. And it is really exciting to think about it this way. You know, I've been thinking a lot recently, just as this generative AI sort of world has been thrust upon us over the last year. You know, I feel like there's this anchoring effect, right. It's like, the first things that came out were Chechi Beatty, and then it was sort of incorporated into into being and then Google started to do all
their stuff. And we sort of anchored on okay, this is what we think of as generative AI or and, you know, mid journey and these audio developers and things like that. It's like this is what we think of as generative AI, you sort of ask it a question. And it looks into its own really, you know, complex and trained model and generates an answer, but it's somewhat isolated from the rest of the web in the world. That's
not as true of Bard. And now that it's starting to develop, and we're starting to get our heads around this more, I feel like that paradigm very well might not be where we end up at all, it might be considered the very Stone Age of this. And what you're saying about agency, I mean, that question, hey, here's a realistic question. I want to go surfing, you know, next month, or I want a new job, and here are my skills. What should I do? Or I want to go to
college? Where should I go to college, give me all these ideas, like, these are real world problems that cannot yet be answered by AI, or not well yet, but they're headed there. And I think they're really, I think we're gonna look back in a year or two and think of this first, you know, just sort of chatbot conversational based model as incredibly old fashioned. And it's just like, amazing to see how this is all gonna go. Yeah, I
feel that the kind of the blocker very often, for me to use an AI Chatbot is that I have so much context in my head, it'll take me longer, it's more work for me to actually explain everything to the AI, as opposed to just doing it myself. So that's how you can kind of the AI get that context and
know, know me well enough. And what I've done on my computer lately, and the hustle that has the up to date context, I'm not saying we should integrate AI and like too deeply into our lives so that it spies on us. Night and day. It's it shouldn't be careful here in how we added to our lives. But it's definitely room for a lot of improvements there. Yeah. And
I remember the first time I saw somebody talk about how you should ask Chad GBT to ask you questions before giving you an answer. And that's sort of just I had a little bit of a mind blowing moment there. Because hey, of course you should. But how would I bet most people never think of that. And that gets to some of the exactly
what you're talking about. It's like, you can ask it and assume it'll, it'll come up with a decent answer without knowing you without knowing the details of your question without knowing exactly what you're really looking for. But if it doesn't know, and it can ask you, it almost feels like that's a more natural way to get to what you really want to know is it it's looking at your data, if you allow it to, it's asking you
what you really want. And rather than it being on us to sort of figure out us humans, you know, to figure out exactly how to phrase it all this prompt engineering stuff, it should be on the AI, which knows this stuff very well, to try to get out, get to the right answer to get to what we really, really want. And it's really interesting to see it evolving in that way. Totally
agree. And I actually had the same realization when I watched the course or some of the lessons for the course, we have on prompt engineering for web developers. And our teacher there did exactly that, like ask, she was going to build an app with AI and ask the AI to ask questions back so the see could improve her prompt. And it just shed light on so many dark spots in the initial prompt and like, unanswered questions. So for me, it was like, wow, I gotta use this more. Yeah,
I've done it for like writing curricula or syllabi for things. You say, Oh, write me a syllabus for a class on this, ask me a bunch of questions. And they'll say, Okay, what's the target audience? Tell me the learning objectives, you know, how much experience? Do they have all this stuff? And it's like, these are great questions is exactly how I should be thinking in my own head. But it knows a lot. So there's so much we could be talking about here. And I look forward to continuing these
conversations. I have to ask you this question, because it's just I know, you're really on cutting edge on this. What is the most exciting trend you see in the Ed Tech landscape? This is not the AI landscape specifically right now, but in the EdTech landscape right now that you think our listeners should keep an eye on what's what's changing in edtech.
So to me, the most exciting thing is what I talked about earlier, how AI when you teach something can front load the fun and delay the theoretical hard parts, because that I think, allows for a whole new segment of people to, for example, start coding people who wouldn't bother learning about data types before or building a small game because they just don't have the patience or just not naturally interested enough that they could perhaps be interested with in it, have they
gotten the AI to help them get the first version off the ground, because suddenly, then they have like, they have an emotional connection to what they've built. That's how we humans are when we build something we really cherish it. So how AI can be used there to introduce hard subjects to new audiences, and doesn't necessarily have to be coding, it could be writing, it could be design, illustration, filmmaking, you name it, anything that takes a lot of
time to master. Now, it's able to give them that spark of joy from the get go. Amazing.
Yeah, I love that anything that takes a long time to see the sort of fruits of your labor. But you want to think of architecture, right? It's like, imagine you wanted to be an architect, you got to go to school for so many years, and learn all the structure at all the thing, which is important if you want to be a real architect, but at the same time, if you could say, wouldn't it be cool if a house looked like this, or
a building look like that? And I'd love to walk through a building where the walls were all like that. And he goes, Yeah, okay, here it is, walk through it. I mean, that's such a completely different way to learn than having to hold on to that idea for decades, until you have all the technical skill to actually make it happen. Filmmaking is a great example, as well, all your examples are terrific for that. I love that.
What is a resource that you would recommend for somebody who wants to dive deeper into any of the topics we discussed today? You keep really abreast of all this stuff.
Yeah. So we've talked a lot about AI engineering. And there's a great blog post about that called The Rise of the AI engineer at the latent space blog. And if people are interested in that subject, I would definitely recommend to check that one out.
Perfect, we will definitely include that link in the show notes for this episode. That is the rise of the AI engineer post on the latent space blog, which obviously AI engineering is something you've thought a lot about, and it's a term that I had not heard, until I heard it from you. And now I think it's it's going to be in my mind for a very long time.
Per Borgen of Scrimba making scrims that teach coding for over a million people worldwide, especially in India, Nigeria, you know, all over the world. I'm really excited about all these all this work. You're doing all this partnerships, and just the way you think about this space that it makes me very optimistic about the ability of ed tech to help, you know, the AI revolution really work the way we want it to work. So I really appreciate you begin with me here today.
Yeah, thank you. No, likewise, it was it was a joy to speaking with you.
Yeah. Everybody should check out Scrimba. Thank you for being here with me on Edtech Insiders. I really appreciate it.
Thank you. Great to be here.
Thanks for listening to this episode of Edtech Insiders. If you liked the podcast, remember to rate it and share it with others in the edtech community. For those who want even more Edtech Insider subscribe to the free Edtech Insiders newsletter on substack.