#212 Transforming Learning Landscapes: The Power of AI in Education with Dev Aditya

0:00:02 - Mehmet
Hello and welcome back to a new episode of the CTO show with Mehmet. Today I'm very pleased to have with me from London Dev Dev. Thank you very much for being on the show today. The way I like to do it is I keep it to the guests to introduce themselves. So again, thank you for coming to the show. Tell us a little bit about yourself and what you do. 

0:00:22 - Dev
Thanks a lot, mehmet, and first of all, I'm privileged to be here and happy that I could speak to you this morning. So I am Dev Aditya. I'm a co-founder at Ottomans Institute and one of the things that I do here is take care of all of our AI products, so I lead and own them, and our AI product department is called OIAI, so our main focus is on teaching. So what we provide is two way interactive teachers to anybody with a smart device, so that you can have access to a teacher whenever you need it, virtually in a pocket. 

0:00:59 - Mehmet
That's great, and while I was preparing it's a kind of interesting journey, dev right, because you were trying to, as they say, democratize the AI and teach AI to a lot of people. So what have inspired you to be on this journey? 

0:01:22 - Dev
Yeah, it's a very interesting journey. So back in 2020, we were providing our original digital training solution and COVID hit, so we had to go all guns blazing into the digital sphere and we released our content completely virtually, and this was led by human trainers from our team, and within about 11 months we were in 11 to 12 countries. So we were growing very rapidly and at that time we even had to take a volunteers to come and help train for us, because getting trainers on the fly is a very difficult problem to solve. The world has a shortage of 69 million teachers, etc. And from that time, we also saw and we've been sort of covered in research conferences etc About it is that the teaching element was one of the main things that really set our content apart, set our pedagogy apart, and the main question there was how do we scale this? 

So, back in the end of 2020, we started thinking how to scale the teaching element, and the only thought we had at that time because teaching is an intelligent activity, right? So the only thought we had in our mind at that time was artificial intelligence, and we started developing this journey. From then, in 2021, we had our first prototype, which taught a series of lessons in BC, fur and HCR refugee camp in northern Iraq, and from then we've had multiple versions until where we are today. And today is not only the language model that we have that teaches, but we also have digital human teachers, so we can map human beings and that human being becomes a teacher which is powered by AI, and that AI virtual teacher then teaches back, so it's almost like learning from a human being off of Zoom. That's sort of an experience we give. 

0:03:17 - Mehmet
That's amazing. Can you describe a little bit the technology behind the world's first AI digital human teacher? I'm very curious about to understand a little bit behind the scenes On a high level, of course, I know maybe there will be some stuff which you cannot share, but on a high level, and because education is something I discussed a couple of times now on the show and I believe AI can add a lot over there, so I just need to understand. You know, how would the process of building this model work for you? 

0:03:53 - Dev
Yeah, sure. So first of all, let me just break down the model for you. 

0:03:58 - Mehmet
So we have a language model. 

0:04:00 - Dev
So the language model itself does two things A, it has a corpus of information and knowledge that it can teach and secondly, it can hold human-like, human level conversation. So that's something we were doing earlier on, right from 2021, but now, with the advent of chat, gpt, etc. We know that, you know, at a larger scale, that can happen. So that's one part of the equation. The other part of the equation is we have a very good extraction system. So if you were to give it, let's say, a research paper so you wanted to learn a research paper, but you wanted to be taught a research paper you could give that research paper as a PDF, for instance, to our AI. It would synthesize that information and then teach that back to you within a matter of minutes. So that's the second aspect of it, and the third aspect of it is from the product front point of view. So we can stick this as an API to platforms. As a bot, for instance, if you work with her, all of your courses could have a bot teacher. 

But, also one of our main pet projects or our current focus is the digital human teacher, which I just spoke about. So we also develop, using AI technologies, digital avatars which look very human like and it's like learning of a zoom, one to one with the tutor. 

0:05:12 - Mehmet
So that basically, let's say I'm just trying to for the sake of audience. So let's say I am interested to learn about the let's take an easy thing chemistry. Right, I'm curious about a specific topic in chemistry, so I can feel, I mean, your technology can get the information and teach to me, as you know, as a teacher is doing this like, as you know, a bot, or maybe in a voice manner. So basically you can replace the teacher there, right? 

0:05:47 - Dev
Yeah, so that's a very interesting point. So first of all, about the replacing of the teacher point of view, I'll address that first and then give you a scenario. Our aim is not to replace the teacher at all. Our aim is that you have access to a teacher If you can afford a tutor or when you're at the institution. We are saying, once you're out of that setting, you can carry this teacher with you in your pocket. You want to learn? At 10pm you can learn for the first time in a two way conversational basis, not in a one way basis where you're just looking at a video and text and trying to learn. And it goes pretty deep, because it's not only that. You can ask the AI a question. So I'll give an example. So we have a lesson, let's say, on how to write a CV. You know one? 

0:06:33 - Mehmet
of the same. 

0:06:34 - Dev
So it's not only that you can ask it a question like you know how long should the CBB or what should be the structure of a CV While you're learning, it will also ask you questions. It will direct your learning and you can also ask it context based questions. So I'm just giving this example of a CV so you could ask it something like I want to apply for the job of a sales agent. 

0:06:58 - Mehmet
However. 

0:06:59 - Dev
I'm just finishing my PhD and my only experience that I have is in research. What should I include in my CV? And then it would say that that's absolutely fine. Even as a researcher, you build transferable skills. The skill of giving a presentation is something that could be valuable in a sales job, so you should highlight it. 

So that's the level of depth that conversational AI today, especially with you, know the generative AI capabilities we can do and we can really go to the next rung of ed tech, if you wanted to call it in that manner, because it's two way conversational and it's fluid, it's human like. 

0:07:34 - Mehmet
So it basically can ask me questions, it can try to see if really I understood the topic or not, which is fascinating. Now you mentioned, like when you started the development and you know, covid hit. So I'm sure this is one of the many challenges that you face. So, but what are like other challenges that you also were able to overcome during this process, because you know this is a huge project. It's not like something easy. So I'm curious to know how and what are these challenges where? 

0:08:08 - Dev
So one of the major challenges was obviously funding. So this is a deep tech product, so there's a lot of funding requirement, and we were very fortunate to be able to bootstrap it until reaching its very high level MVP stages, and the reason for that is we have one top quality grants from Innovate UK, which is the highest funding body of the United Kingdom, and we were also able to get some side income from our other teaching and consultancy work. So we were able to bootstrap right from $0 to $500,000 and really make this vision piece into a very functional MVP that is being used by people. So I think that was a big challenge that we were very fortunate to be able to overcome with the funding Moving forward. 

The two main challenges we see are a machine learning talent that's becoming a very, very high demand, low supply resource, as you know, in today's AI arms race, as people call it. So that's one, and the other one is today AI has finally become accessible. So we started it back in the day, but it's become accessible with the APIs of the open AI's of the world and NVIDIA coming out with their own models and so many other models. That's out there, meta in the loop, google, et cetera. So it's more accessible now, but it's still an expensive service to provide. So if you compare that with the traditional SAS, the cost of providing, let's say, an hour's worth of content is still significantly high, and I think we need to ride this wave and basically keep riding it until this cost of running the service reduces. And that is a normal sort of direction of technology anyway. And I think there is a race to really work on new chips et cetera that can give us higher compute power, which obviously the AI runs on at a lower cost. 

0:10:05 - Mehmet
Yeah, I believe I agree with you there regarding this point, and history showed us over the years that everything might start from a very high point price perspective and then it will go down. Meanwhile, I would say the capacity and the performance will keep also increasing, and more slow still is valid till today, and I think it's fascinating how things change. I remember when I started this podcast, chatgpt was just two months old and see all the things that happened just in a very short time, so I believe that's very possible. Now you mentioned something People it's like arm race and say hi and everyone talks about it. I'm curious to know because they are doing a revolution in education basically so what the feedback you received from both educators and students, like what was the main feedback you get from them? 

0:11:07 - Dev
Yeah, it's a very interesting landscape here. So some of our positive feedback is people are very excited about interactive two-way learning, especially with digital humans, and I'll be honest with you at this stage, part of that is also excitement because you haven't had that before. We don't really have any competitors out there as such which is doing this in real time. The closest one we have from a very big player is Khan Mingo from Khan Academy. However, they are still in the realm of a bot. They're not putting a digital human there, so there's a lot of excitement around it. 

We also did some pilots by plugging this into education platforms, as I just told you about, and two interesting data points from there I want to share is 73% of the users said that they would come back to the platform more because of that interactivity. And 74% said and this we didn't anticipate, it just happened to come that they now knew what to learn next on the platform. Now that's a very important piece, because stickiness is a traditional vein of tech platforms and we could be in the cusp of really revolutionizing that and increasing significantly, by orders of magnitude, the stickiness that a user can have. So that's some of the positive and very exciting feedback that we are getting. 

The other side of the spectrum is what you just said you could replace a teacher, and we always say that our role here is not to replace a teacher but to give access to a teacher 24-7, which has never happened before, and we have about 750 million people out there who don't have access to a proper teacher, or 1-ish 200. So for those sort of people, they still don't have access to proper schools or teachers. But the access to mobile phones and internet penetration is growing extremely rapidly and we've proven that We've taught in Afghanistan, we've taught in refugees' institutes in Iraq. We've even taught in Malawi, which is one of the poorest countries, simply because we can now reach them through the phone, and I think that's very exciting and that's something we are riding on and trying to leverage to also reduce this global inequality per se. 

0:13:18 - Mehmet
That's really fascinating. They have to hear that from you and I believe the reason why people liked it. This is my own assumption. I didn't see it, but my assumption is because I'm sure maybe you have, because of the models you used and the way you've built it, you try to mimic the type of a teacher that usually students love to interact with. So maybe this is why Because, for example, personally I just go back in the days when I was at school or university If the teacher or the professor was kind of boring, if I might say he's just reading the slides or reading the book without interaction, you lose appetite to keep engaged and while there were some brilliant teachers or professors who they can engage, they know, and I mentioned asking the questions and this take and give approach. 

I believe this is important To your point about the role of teachers. I see it as a tool. I don't believe because we were talking about my guest Kash a couple of weeks back in health state. We're not saying it will replace the doctor. It will actually give the doctor extra tools to, you know, to his job or her job, teacher base. I had Manisha with me like two weeks back also as well, and we're talking about future of education and general using tech and about homeschooling and all this, and we agreed that some areas of the world they are underserved and this technology it will serve them and this is a huge plus and it's fascinating with a small device, now knowledge is in the hand of everyone and really I get excited myself when I chat about these things. So amazing, fascinating. So I believe you work in multiple countries. You mentioned some of them, so how do you tailor to different cultures and education systems? How do you manage this? 

0:15:27 - Dev
So before I answer this question, Mehmet, can I just touch on something? 

0:15:30 - Mehmet
Sure, definitely. 

0:15:32 - Dev
So absolutely, you know, spot on on that fact that we have had boarding teachers and we've had teachers who really inspired us, right, and that's what we're trying to do. And I just give an example. So our first virtual teacher, we've called her Beatrice, right? So I'm just using that name. We are able to give a Beatrice to everybody at the same level. So when you get. 

Beatrice, you get the same level of teacher. When I get Beatrice, I get the same level of teacher. But as you learn more and more with Beatrice in the tech world, as the user engages more with the product, as you learn more and more with Beatrice. She fine tunes her teaching to you. 

0:16:12 - Mehmet
So you're giving everybody. 

0:16:13 - Dev
equal access at that. You know, equal playing field to begin with, and then it morphs to your needs. So that's again very personalized, and that's what AI allows us to do, and that's, you know. That's what I thought I'd share based on what you were saying. But absolutely. Thank you. And coming back to your question about you know the different countries, et cetera, our AI teacher as I said, we are focusing on it being a teacher, so it's content agnostic per se. 

So, for instance, if you had a different content with different sort of pedagogical influences in the UAE, for instance, if you had differences in India, if you had in the UK, that's fine, because that really doesn't matter, because it feeds into it, it synthesizes that information and teaches you, not something that's hard built into it. So that's again some flexibility that we have. 

0:17:05 - Mehmet
That's amazing, and how do you envision the evolution in the coming years for this? Like what are you know the other collaborations or projects that you know might be coming in the future? 

0:17:18 - Dev
So currently, we are talking with universities now in a couple of countries to pilot this within the university with their curricula, so that sort of touches on that question as well. 

Universities have their own curricula, et cetera. So it will be doing that and this will be coming as early as September, october of this year. So that's one. The other one is to really, you know, be able to be in a position to jack up the interactiveness and the stickiness of existing MOOC platforms and digital platforms. But the ultimate goal really is its voice first as well, our solution. So the ultimate goal would be, at some time in the future, if you wanted to learn anything, you should be able to, you know, open the OIA app and ask, let's say, beatrice, can you teach me how to do this? And if we are able to reach that position where you can rely on this tutor, coach, mentor that's on your phone, who you can ask something on the fly right, and they are able to teach you, at least give you those fundamentals in an interactive manner, I think the world will fundamentally be a different place 100%. 

0:18:27 - Mehmet
You know, and I, like you know the way you are passionate about it Dev Like we need like more projects like this. Honestly, and the reason is this is a true, I would say, proof you know that AI is. You know it's not evil, right, because a lot of time on the show. This is why I want to highlight, because I know like people they go by sometimes the stream media and you know like it's replacing the jobs. It's eliminating this doing that. 

But you know this is another example of how AI can be leveraged for something which is everyone touch, everyone of us, and it, you know it, will empower us. So I'm just imagining, you know. Again, coming to your point, If someone is not familiar with, maybe, programming, so maybe this tool can you know, Allow them to understand programming algorithms, these topic, in a better way and then maybe they will go and do something very useful With it. So it's amazing over there now, if you allow me, dev, you know, like, just shift gears a little bit and you know Little bit more about you know the entrepreneurship side of it, yes, and so you manage to raise the funds in any of you know, very short time. So what advice you, you know can you have for aspiring entrepreneurs looking for securing investments what you can advise them from your experience. 

0:19:52 - Dev
I Would say go with your gut and your vision. 

So there are two, two elements of this go with your gut and your vision, but ultimately it's the market that decides what's that final output that they want. 

And the reason I'm saying that is because if you go too early into, you know, trying to get let's event, your funding or, you know, angel funding, etc. Without really creating that MVP, without really, you know, talking to a few users who have interacted with this, you may end up not, you know, setting that vision goal. It may get skewed because too many other stakeholders have come too early and you are not yourself certain about what that end Goal is. So take a little bit of time, you know, build that understanding yourself and then go and push that in front of wherever you're seeking funding from, because you want, you know, whoever supporting you beat grants, beat VCs, beat angels, whoever supporting you you want them to back your vision, not just an idea. Because that's where this whole dynamic can go wrong. And if, if you are skewed from day one on that very solid trajectory that you know is going to succeed and obviously we all have to have that optimistic belief, but if you don't have that from day one. You're certainly not going to reach that end point. 

0:21:14 - Mehmet
That's amazing. One thing also about you, dev, while you know getting you know a little bit chance to understand more about you before this episode recording. So you have worked at Alan Turing Institute and Brunel University, london how this experience has shaped your vision for AI in education. 

0:21:36 - Dev
See, my beginning with Brunel University has shaped who I am. I started at the age of 18 there as a student and I have had multiple interactions with the university, so that has helped shape who I am, I think. Firstly, from an educational standpoint, it has helped me understand, you know, what else we could give out into the world. You were talking about underserved, etc. So that has really given me that foundational understanding From a research. You know, from a PhD side of things, my story is slightly different. I actually started OIA's AI work and released that first prototype for the learners in Iraq before I got into the university level of research. So it's not that my research led into this. I started my research because I thought there is so much more to do in research apart from the product itself. So it's slightly different from other stories you may hear. 

0:22:35 - Mehmet
Amazing and, you know, as we are coming to an end also like this is something what I quote my eyes, you know, like being a young global innovator and an under 30 social entrepreneur. How do you hope that you can inspire the next generation of tech leaders? 

0:22:53 - Dev
Well, I have an open door policy. It's not an open door policy, rather, it's an open linking policy. I do support a lot of young entrepreneurs. I always like to actively give them time. But one thing I'm very certain about given in the space where we are in, and at the end of the day, the work we are doing is a team effort. Right, it's our ML engineers, our product leads, dr Ottomans, their research, our entire team, marketing, etc. Who are driving this, and a lot of them are pretty young, as you would guess, in this new space of AI, and I'm sure we'll have a couple of other under 30s coming out of our team in the next three to four years. I'm very certain about that because we are for runners in this space and our intention for doing this is good, so I'm very hopeful for that. 

0:23:42 - Mehmet
That's great to hear. So people can find you on LinkedIn. This is what I understood, so I have your profile. I will put that in the episode description, if you don't mind. Before we finish, is there anything that you wish? I have asked you. Did I miss anything? You want to tell us anything? 

0:24:00 - Dev
No, not at all. I think I really enjoyed my conversation. It's always good to speak to people, especially as I could see At the moment we can see each other as well that you were excited by what I was saying. You could understand what I was trying to explain, so that's always good and motivating for me. So when I come out of this podcast and go to my next meeting, this motivation stays with me. So I would just want to be thankful that you gave me an opportunity to speak with you today. 

0:24:27 - Mehmet
My pleasure. So what is the website there for? 

0:24:31 - Dev
So you can find out more about us on oieducouk. 

0:24:37 - Mehmet
Okay, I will put that. I have that link. I will put that also in the episode description. So, guys, it's a very cool thing, it's a very cool technology, it's a very useful technology, because I'm a big believer there that tech should serve humans. It's not like about being fancy, so when I get excited because in my mind as we were speaking, I can see a scenario where it's really useful and helpful. So, thank you again for being with us today, dev, and, as usual, this is how I end my episodes. 

Guys, thank you for tuning in, thank you for the feedback and I'm happy that you are liking the content. Keep the feedback coming, whether it's positive or negative, no problem, and if you want to have any questions, you can reach out to me. You can find me also on LinkedIn, where I'm most active. If you're interested to be a guest same like Dev was today, I guess don't be shy, don't hesitate, reach out to me. We can arrange the time and we can discuss the topics also as well, and I hope that you enjoy the show, as usual, and we'll meet again in the next episode. Thank you very much. Bye-bye. 

Transcribed by https://hello.podium.page/?via=mehmet