How the UAE is closing the education gap with AI tutors
Episode Details
Transcript
What if every student had access to a personal tutor who could help them learn based on customized areas of improvement or subjects of interest? This was the ambitious goal of the United Arab Emirates’ Ministry of Education. To turn this into reality, they connected with Quddus Pativada, the 20-year-old founder of AI learning firm ASI.
Pativada had already developed and tested an AI tutor with the potential to transform how students learn, and was looking for a customer to implement his solution. Rather than trying to block the use of AI in their education system, the UAE is embracing it. Working with ASI and Microsoft, the Ministry is rolling out the AI tutor for all students across the country, providing one-on-one tutoring as a means to modernize and improve learning at all levels.
In this episode, we hear from both Quddus and Raghad Aljughaiman, Strategic and Future Planning Advisor at the UAE Ministry of Education, about their ambitious journey toward personalized, AI-assisted learning at scale.
Link to full episode transcript.
Produced by Larj Media.
Hayete
Welcome to Pivotal. Hayete Gallot, corporate vice president for commercial solution areas at micro so. I work with customers around the globe to transform their business through technology. At the center of every transformation are people who give technology, its purpose. And that doesn't change with the advent of AI. It's actually being accelerated. People spark visionary ideas for leveraging technology. The release of AI technology like Chat GPT this year is exciting, but it has led to big question as we grapple with the best way to harness those tools to enhance and support people behind the work. We like to talk about technology. I love to talk about it, but we often forget that technology is most effective when it supports people with purpose. This season will demystify AI by talking to the innovators using new AI technology to uplift their industries and augment their people from education, to journalism, to surfing, and it just illustrates what AI is about. Everybody thinks it's about tech. No. Everybody's using AI. That's what we're gonna show you on this season. Hayete
What if every student had access to quality education. Today, we're looking at a story from the United Arab Emirates that asked this question. Their ministry of education is leaning into technology with the ambitious goal of rolling out an AI tutor for all their students nationwide. This is a story with serendipity. The minister of education had a vision to invest in improved learning outcomes. And that commitment coincided with the release of Chad GPT and a fortuitous connection with an impressive twenty year old UAE based student who had already developed and tested a version of the tool the Ministry of Education was looking for. Ragad Abdullah Aljugaman, the strategic and future planning advisor at the UAE Ministry of Education, tells us how the ministry landed on one on one tutoring as a means to improve student learning outcomes. Raghad
We've been talking for decades about personalized learning, closing the learning gap, how can we provide access to quality education for everyone? How can we solve for teacher shortage, for teacher upskilling for creating pathways for teachers. We've been looking at our curriculum, our skills that we focus on teaching our kids, etcetera. And along these years and the conversations that policy makers, the education sectors, teachers, parents and everyone has a stake in the education system. One of these solutions that emerged towards the access to one on one support where students want to learn through the old school approach of just being with someone and this person will help them and mentor them based on what they need to learn specifically, which is going to be super personalized. There's a lot of evidence around how it helps students improve their learning outcomes, when you compare students who will have access to private tutors, you will see improvement in their learning outcomes for up to, like, twelve percentage point from the ones who don't have access to tutors. But, again, not everyone has access to these tutors. We at MOE are the custodians of the education system. We know that there is demand for private tutoring, but not everyone can get access to quality tutors and tutors who understand our students who understand our curriculum And we've been having discussions around this topic for for some time. And we looked at different ways of of solving for this Hayete
equitable access to education was something the ministry of education had been thinking about for a long time, and they looked at a lot of different possible solutions. Including policies to regulate private tutoring, all in pursuit of finding the best way to support learning outcomes for all students. Raghad
We had a few conversations with different partners within the education space, within technology as well. One of them was Microsoft or our partner at MOE. And we're like, this is a problem that we want to solve for. Right? We did our benchmarking. We looked at what other countries are doing. We look at what the startup scene is about when it comes to online tutoring. Hayete
As these conversations were happening, OpenAI released chat GPT in November twenty twenty two. Raghad
Everyone was talking about chat GPT. A lot of government entities around the world were like, hey, no. What are we going to do with this new tool? Let's ban it. It's like cheating. How can we assess if our students are learning or not? It's like everyone's dealing with it, like, a problem that we need to solve. Versus, like, this is a new technology. This is a solution that if utilized in the right way can help us solve for the different challenges that maybe not everything, but will will help us revolutionize the way that we teach and and educate our kids. I remember the discussions that we were having with our partners at Microsoft and and then the idea emerged into, okay, we know that MOE wants to embrace generative AI in our education system. Hayete
It may sound simple, but it is not an obvious step. It's actually quite extraordinary when you think about it. A nationwide education system supported by their government has decided not to fight it or at best tolerate AI as part of its education system, but to lean in and embrace AI to revolutionize all that they do. Raghad
That was the first principle that we had as the MoE and the his excellency, the minister of education talked about that as well. It was like, look, it's here. It's either that we we adopt and we change to help our students to be relevant or it's gonna be there, and we can't we can't avoid it. And we will be irrelevant for our kids, in that case. So If you look at, okay, we're embracing it. What does that mean? It means that we will find ways to integrate this within within our system. And because we were already solving for access to tour tutoring and looking at different ways, we're like, what is the best way to introduce AI as well? Into our system. We don't want to come up with the solution just for the coming up with the solution. We looked at our pro at our problems and we were like, okay. The first step is, yes, we endorse it. The second step is we need to review our policies, our guidelines in the education system and see how we can integrate it. And the third point was, okay, can we solve one of our key problems that we've we're facing right now with AI? And this is how the AI teacher emerged. Because it was all for a lot of things that still have restrictions and limitations even if you would have a a human based platform. Hayete
With those three pillars in mind, their projects started to take shape. Raghad
We want to support our students get the personalized learning that they need we want to support our teachers as well who are burdened with a lot of the admin work and also doing their best to support our students in the classroom. But again, when you're with a classroom, it's not the same as if you're doing it one on one. So how can we support our teachers know more about the students that they have in the classroom. And we started the conversations now around we know our problem. We have our tech partner, and now we we're looking for a developer who is going to develop this tool. And there was a strategic question of should we develop it in house as MOE? Or should we look for a partner that would develop this with us? And his excellency is a strong believer and us as the regulator is not getting into the business of what a tech startups are doing or supposed to do. So I started looking for potential partners and developers. Hayete
During her search, Regad came across a tool from a company called ASI, that picked her interest. Raghad
I started chatting with the tool. It was a WhatsApp based. If I remember correctly, I was asking about inflation, felt like a human talking to me, to be honest. Sometimes you get you get super impatient with with AI replies, right, in response We're like, no. I want to speak to a human. No. I keep repeating my question. It's like, it didn't feel like that at all. And this is what I love the most about it. Hayete
I can't believe it, but Regad saved our very first conversation with the tool. Raghad
Hi. I'm Myda, virtual twenty four. Twenty four seven tutor. An AI designed by the, DigestAI and can help you with with your homework essay outlines. Quizzes and more before you become a student, what's your name? And then I ask a few questions. And then it's like, by the way, I just work with MOE, and it's like, awesome. What do you think you will be needing the most help with and what are your biggest problems? I was like, no. I want to connect with the human who created you. Great. I'll do my very best to help you. And I will get better at helping you. You can send an email to hello at digest AI. And then I had the first conversation with Curtis. And I was like, is it just you, or is there anyone else joining the call? You know, it's just me. And then I heard this story, and I was, like, very impressed. And then I remember after the call, I was speaking to one of my colleagues, and it's, like, I I think I just found our Mark Zuckerberg for the Emily. It was like very clear that we found our match. It was ASI. It was Kudis. Quddus
My name is, Kudis Beth Yvada. I'm the founder and CEO of ASI. Hayete
By the time Kudis Pethida, and his company ASI caught the attention of the Ministry of Education at age twenty. Coulders had already developed three in market beginning when he was just ten. Quddus
I started learning how to code when I was about eight years old. It was mostly developing games. So when I was ten years old, I published my first game to the App Store, which was a a clone of, flappy bird. It was a really popular game at the time. I created sort of a more customizable clone about, of it ended up sort of building more games since then, continually building products, you know, throughout middle school, high school. And then right before ASI, I created a COVID nineteen contact tracing platform that allowed students to anonymously notify their friends that they had COVID. There was a lot of stigma at the time, and that was sort of my first community facing platform, which eventually led to my work on, ASI and trying to sort of fix more issues with an education. Hayete
Kudos COVID nineteen application. Won an award from the government, but he was already on the next idea. In early twenty twenty two, Curtis launched an AI based application called Cado, Which turned photos, documents, or PDF into flashcards? Quddus
There was an idea I had back when I was fourteen, but the the tech wasn't there yet. They serve thesis. It's a core thesis that holds true, even today was, educational content is inherently inaccessible to people my age. Right? The so called TikTok generation. How do you convince this entire generation of kids who spends three hours watching thirty second videos to then end up reading a three hundred page textbook in the same period of time. So that was sort of the the core thesis and still holds true today, and I expect that to hold true for the years to come. We started thinking about how do we make this educational content more accessible. So we looked at textbooks and one of the biggest things or biggest drawbacks of textbooks was, just purely a length. It was incredibly long that a vast majority of friends I had in high school didn't even use textbooks. They would just opt for, watching YouTube videos or finding another resource that was much less sort of information dense. I built a very simple app that could take a picture of the textbook, and then, initially, it could just create summaries for you. So kado's in in its first version was just a summary app. At the time, sort of summaries were divided into two things. They were either, abstracted or extractive. A large area of machine learning was focused on extractive summaries. So just picking out the most relevant sentences. Me as a student, when I started playing around with these models, I realized that they weren't really summaries. They were just, sort of, rebuttals or repetitions of what's already in the text, but in a much shorter form. I wanted to see how we could reword the text that a human would or how the the best teacher in the world would. Sort of personalized to me and a summary that helps you understand something immediately. So we ended up, you know, stumbling upon language models, and this was back when GPT two was popular, and it was still when GPT three from open AI, I was still in a wait list. So language models certainly weren't popular at the time, and we ended up creating a much smaller model like a great summaries. And then students sort of had this little reaction. I gave it to my friends in my economics class, and students were, you know, incredibly sort of blown away by the fact that, oh, this sounds like an actual human speaking to me. They were just summaries created by this language model. It's very similar to the technology that powers, you know, tools like chat GPT today. We ended up on that. And, you know, towards before I graduated, I ended up raising a little over a hundred fifty thousand dollars from, Mark Cuban who ended up becoming a a really early investor purely from a cold email Sean Patel, who was a founder of prep expert, and then my college counselor at the time who was helping me write my essays. Hayete
Yes. You heard that right. Kudos, cold emailed Mark Cuban, and actually received a response and a positive one. Quddus
I ended up taking you a a gap year to continue focusing on Kado. A lot of potential there and just the pure excitement of my friends and people around me was obviously very encouraging. So I started doubling down on the platform. We ended up creating another model that could create flashcards from the textbook pages. So took a page of a textbook, and then it would create flashcards with questions and answers for you instantly. So really trying to reduce the time it took for us to and to understand something. After a few months of working on that, we will we started to see was students had this incredible wow reaction, the same wow reaction that students had with chat to BT or the other AI tools you have popping up today, but retention was still a weak point. After thirty days, as valuable as the app was, students just couldn't get themselves to be code on code addicted to a platform that they'd have to sit down and take pictures five different types of pages and then wait for the flashcards to be generated and then study on their phone. That was in a behavior that was very primal to students. It was very sort of outside of what they were usually sort of doing in their study routines. Hayete
The next iteration of Keddle, which was the version of Regad initially used, came from an idea kudos had while visiting family in India. Quddus
I gave my cousin the the the platform to use, and, my cousin pretty much told me that this This is a horrible app. I don't understand it at all. I mean, in India, you know, flashcards aren't really a thing. So he he was very confused as to why this was even a platform. And it was a very well designed app and out that I'm still proud of to this day, but when I saw him using a tool like WhatsApp, he was very familiar with it. And keep in mind, this is a thirteen year old in rural India. He was using WhatsApp fluently. He was sending GIFs. He was creating group chats. He had no problems to whatsoever. And, I thought WhatsApp wasn't as well designed as Arab. But what I sort of saw then was people our age, the new generation quite hard to make them change habits. So we started looking at what other habits the students have in their study routines. A large part of their study routine was actually speaking to other people. Their smartest friend in school or the teacher they're really close to or a tutor they pay for. And we were trying to find a way to replicate that, but something that's successful twenty four seven is actually much more efficient as accessible for a fraction of the cost of a a personal tutor would be. So this was a few months before, Chad CPTa come out we ended up playing around with the chat based interface. So, essentially taking the work that we did, with language models, but, training them to be, you know, sort of, fine tuned for chat. And make it behave as a tutor would. So, that's how we started getting to work on on the AI tutor itself prior to the UAE project. The end of last summer, we sort of set out to say, what if we could actually commoditize such a a a a better factor in learning. And we ended up working on the AI tutor, ended up showcasing that Techcrunch disrupt, and then start being heads down in our development. Obviously, when Chad JVT had come out, the the team was sort of low worried. Now there's a fifty billion dollar company building the exact same thing. So we we really heads down an education, sort of building out really key differentiators. And eventually, the ministry of education and the UAE took notice of that and we were able to collaborate with the UAE National AITO project. Hayete
What happened with his cousin just reinforces the notion things need to be very familiar if you wanna drive adoption of an application. We've seen that time and time again. As we look at how people can get people to use their applications. In this case, his cousin was familiar with WhatsApp, and just couldn't understand this new UX that was presented to him. It is typical and it was so smart for kudos to actually change this approach accordingly. It seems the ministry of education couldn't have picked a better partner if they tried. Kudas is a recent graduate from the UAE as a product of their education system. He understand the ecosystem very well. Is very close to the end users having just graduated high school within the last year. And he already has developed a tool with proven interest and says with students. Raghad
They were not starting this from scratch with us. They have experimented already and they had developed a tool that thousands of students were using. Quddus
We sort of deployed our contact system on WhatsApp for students to play around with. That ended up unexpectedly going viral a little bit in areas like Africa and the Middle East, where we've been able to collect really high quality data and just really have user feedback to iterate upon. After Chachapiti come out, we really heads down on the technology. But we weren't really focused on productizing it. But we still needed fast user feedback cycles and, being able to see what students think, being able to see if the work we were doing was really helping students or not. So We pretty much took the entire system, and then we plugged it into WhatsApp. So students could add a number on WhatsApp and just speak to it like they would to a human being. Obviously, they were well aware it was an AI but students would pretty much have the number of WhatsApp go back and forth with it. What ended up happening sort of early this summer and and right before exam season, was that students start sharing it with their friends quite frequently. We only gave it to a few hundred students, but that ended up going to tens of thousands of students, around the world, and then started getting this really great usage data from areas like Africa where traditional AI systems would be pretty inaccessible just from a pure, lack of computers or not having wifi home, a tool like what's have allowed us to. Take this really powerful technology and then deploy it in areas where even if you're on a two g connection, you're able to sort of access it. Hayete
So, again, here, we're seeing that to get a lot of people to use or technology, you need to find a UX that they're familiar with. And in this case, it's all about WhatsApp on a two g connection. And it's very interesting to see the network effect here. It's not about forcing users to adopt the technology. In this case, users loved it. And they got other people to use it. And then everybody was using it. That's the best case scenario when you're introducing a new application. Not only was the application working and spreading. It was reaching communities with historically less internet access and bandwidth. This story is interesting because it reminds you of agriV that we discussed earlier. They also shifted to using WhatsApp to farmers because farmers were familiar with WhatsApp. This is how they drove adoption of their application. Quddus
Our north star or our mission is to give every student in the world, a a personal tutor. Some of the research that we do is around taking these large language models and making it accessible through, in a in a low bandwidth region where you don't even need a two g connection, but could you potentially access these really powerful tech with no wifi even? Could it be locally hosted on your device we're testing out some other methods related to that. So it ended up going slightly viral in Africa and Middle East. We started seeing some really good usage even in Europe, and that was one of the things that really helped us, win the UE contract because we were able sort of show actual students using it and actual students saying that they prefer this personalized system over even a very well known to in, like, Chachi BT. Hayete
Even in ASI's initial testing of their AI tutor app, this will promising initial results they couldn't have foreseen. For example, They found that over ninety five percent of students at a medical school in Ghana were using the app. Quddus
So we had this, like, bulk for medical school who were just using it purely for studying sort of things related to dentistry or surgeon studying for their exams. So that that was really cool and and super high impact that something we were really proud of. We also ended up seeing students in Lebanon who were using it schools were closed. So they were actually using it as replacement for some of the sort of schooling that they were missing out on. So, you know, most of the use cases surrounding this after school supplement a student comes home, and they're trying to study their text, but they just simply don't understand it. They sort of go to our tool as a first stop. Hayete
They also saw students who described it as a tool they use in transit. Quddus
Students in the bus or students in the car when they couldn't have a big heavy textbook with them or they just were sort of more satisfied with other stuff they were using. It was quite interesting to see how diverse use cases were, but at the end of the day, all of the students were trying to solve a very similar problem, and exactly the problem we intended them intended for them to solve, which was How do we understand this one thing faster? Most of the use cases were sort of around that, but mostly after school learning supplement, how do I learn this thing and then getting an answer as soon as possible, but also having fun as a companion of sorts. So we had these safety barriers in place, but a lot of students really just felt they had a really smart friend they could speak to. Hayete
Students were using it as a tutor, but also as a social outlet of sorts. Quddus
We were seeing some students who were using it upwards forty five minutes a day, which is very similar to the usage you see on social media or social networks. You know, we had a a large cord who was really just using the tool because they felt a little bit less lonely, which is also another high impact use case. I was sort of a little away from our mission, but, something that we were really proud of as well where students were actually learning more as a result of feeling like they had this companion. So lots of interesting diverse use cases, but most of it sort of tying back to this perfect efficient after school supplement that they don't have to pay for. So we really confident in the approach we were taking, and that ended up sort of having this really great ripple effect into all of the other stuff we were working on. Hayete
If I was able to bring all this work to Regad and her team, and now the task became, how do they create an AI tutor specifically designed for the students in the UAE. Raghad
A tool that speaks their language because we're developing the tool now where it It's bilingual. It speaks Arabic, Arabic, modern standard Arabic, and Amerati dialect, Arabic, and English as well. And it's trained on our curriculum. So a tool that understands our students, understand how they speak, what they're supposed to learn. And it's very adaptive. It remembers them. It knows what they asked yesterday and the day before, and probably the things that they need to learn now and the sequencing of our curriculum and the areas that they are weak at and the areas that they are excited about. So I don't think that you would be able to develop a person that's such a personalized tool without without generative AI, especially now that we're we're focusing on creating this PLM, which is a personalized learning model for our students. Quddus
Chachup Tea had come out, it was sort of a a huge wake up call and the fact that on one side, there was this what we thought at the time of master competitor, but ended up not really being a competitor. But on the other side, We saw the power of the technology itself. Hayete
Students were using chat GPT at scale and deriving real value from it, but chat GPT alone. Wasn't going to be the solution to individualize learning needs of the k twelve population in the UA. Quddus
What we saw was the OpenI models down the loan, we're unable to provide the value that students need it. Today, if a student in grade nine was to ask, Chad GBT, what is inflation, and a student in grade twelve is asking the exact same question. They're getting the exact same answers, or you could be doing a PhD asking the exact same question. You're getting the the same answer as well. Hayete
So ASI and the Ministry of Education set out to build a much more personalized version of the tool. Quddus
I think what really interested them was the fact that not only were we at the frontier of the technology itself. I mean, we weren't just another large language model company that was just purely developing on an API. We were really building proprietary systems that could understand students better than most systems could really help students in the Azure open AI service at at at its core, but building a contact center around it. The one that was able to sort of accurately tell the model know, the student's a grade nine. He's really good at these subjects, really bad at these subjects. This is how you're making better at these subjects. So we trained a profiling model in house. And really the entire context engine, we try to keep it as in house as possible. And then that sort of surrounds this core language model that today, you know, sits on Azure open AI service as a result for partnership with Microsoft. Raghad
I think my favorite part about interacting with the tool is that it gives you examples that are very relevant to you. Like, here, if if you are in Dubai and you ask, a math question and you will get You will get, an answer about, like, but Khalifa. Or if you're asking a question about inflation, you will get the local currencies as reference. It's small things, but it's small things that also matter to our students and bring the curriculum to life. Hayete
For the first phase of development and testing for this UAE specific AI tutor, they had to select two subjects to focus on. After some discussion, Regad and her team selected math and English. Raghad
Although math was the most challenging one to start with, but we're like, no. We want to we want to include math in the first phase as well. So there are multiple reasons. One of these reasons is that this these are two of the most important subjects. And also the two of these subjects that you'd have a lot of resources that that you can utilize that are mapped to our curriculum as well. And we wanted to solve for a real problem. So math was a real problem or a lot of students. And for LMS as well. So how can we how can we solve for that and start integrating this from the very beginning? Hayete
So how are they overcoming the challenge of incorporating a pedagogy klee correct math curriculum with a large language model? Raghad
We are working with our teachers as well. Like, we've come up with a way to assess the accuracy of the tool and the pedagogy as well, where we're asking our top teachers the same questions that we ask the tool to see what the answers are and how we can improve the engineering, the prompt engineering and the responses of the tool. And has two parts. One part is when the students are proactive and they go and they ask questions, the tool itself is also proactive. So if you ask a question like, a few weeks ago about something or the the day before, and then, hey, you learned about x now. Maybe it's best that you learned about y, for example. Hayete
As we explore companies on the cutting edge of what AI is able to offer, one of the most striking takeaway. In building an application on top of large language models in general of AI is the importance of investing in good data. We've heard over and over garbage in garbage out, but it is worth repeating. Good data is a core tenant of quality, technology built off AI and large language models. It is also important to train the model, to ground it and fine tune it. So that you can provide recommendations in the right context. For the ministry of education, this meant leveraging the context from students and teachers. Quddus
Of the things we spend the most amount of time on is data. So, you know, not just student AI data, but really data that goes into the AI itself and the language models. We've invested a lot of money and time into really figuring out the right data that these models need. But more importantly, what's the right data that translates to the best outcomes for the students? You know, our our Nordstrom metric is learning outcomes you know, we see certain types of data that may not be in a student's traditional curriculum, but is available online, and that sort of has a much better impact on the personalized model and how it's able to sort of curate responses. So spent a lot of time on that, but really the key metric that we use, some of the really high quality data that we collect is, from this contact center itself. So this contact center not only consists of personalization model that's able to sort of accurately personalize responses a student is learning about inflation. And they're, in Dubai, we're able to use Shawarmas as an example of increasing prices. But also it has other models as part of this context in it. So we've got a mathematical reasoning model that's able to supply context on, you know, really complex questions that a language model on its own wouldn't be able to answer today, but the combination of the context of another model plus the core LLM, is able to sort of answer really accurately. So there's the the the context engine allows us to collect really high quality data. And after that point, it's really a flywheel where we're using this really complex engine to, get the best sort of responses and responses to the students and the best manner possible, and that data goes back into the system to make it better. So, you know, we had that initial stimulus today. We have over five million questions that are being asked by students And we've seen that, you know, over the last few months, so that's over five million high quality data points that we can use to sort of improve the models rapidly. But the context engine is part of that sort of pipeline to create better quality data. And after that sort of initial stage, it's really a flywheel that keeps on going really in perpetuity. Hayete
You can't underestimate the importance of that context setting. As you apply the large language models, the more context you can provide, the more adoption you will see, because the more personal the experience will be. And in this case, for student, it is so important. If you're in Dubai and you're discussing inflation, it should be in the context of Dubai. If not, The tool doesn't seem very relevant to the end user. It is so important to think about what is gonna be the engine that you use to provide context as queries come through. You need to be able to interpret the context of a given situation or query so that you can provide more relevant and accurate responses or outputs. But the context engine can only take the tools so far. Quddus
One of the things we're doing in house right now is actually curating our own content based off certain frameworks or certain topics. You know, we we built a pretty solid team around that fact as well. We are definitely grabbing information out. What what we do really well is not just grab information out when the student but, really being able to predict what the student is going to ask and then finding the right context for it. So it's not like there's a team of thirty engineers who are just scraping content and figuring out a way to make this work we built a system that's able to do this on the fly. So before a student even asks the question, we know exactly the right sort of context to supply the system. And then when the student does ask the question, you're getting a response in, hundreds of milliseconds rather than tens of seconds, which is what you usually see if you took the approach of just purely grabbing context when the student asks a certain question. Hayete
For regard and the ministry of education, one element of immense value is unlocking a lot of new knowledge and insights on learning patterns nationwide on the back end. Raghad
What is different about this tool as well is that in the back end MOE has access to everything. We see every single student who has access to this tool and how they are performing. For example, if our grade nine students are all asking questions about, math concepts that are in grade twelve, You would be able to see that. You would be able to see where your students are at and how the sequencing is matching the level of the students and vice versa. If a student in grade nine is still asking questions about, concepts that they should have mastered in grade four, you would be able to see that as well. And you would be able to see it, to analyze it, and to, like, adjust as needed if it's, like, a a personal a problem or if it's a system wide policy insight that we get into our our curriculum, and that's at an at a high level for MOE with with policy making. And you could, like, generate insights about demographics, about different schools, about everything that you could you could think of. When is the best time for our students to learn? How do they ask the questions? How do they interact with the tool, etcetera? Hayete
Having unprecedented insight into student learning pattern is exciting enough. I mean, in this new world where educators are able to gauge proficiencies, could we replace conventional standardized tests? But another important element is the real time insights for teachers. Raghad
Even teachers now would be able to tell. They don't have to wait for the next test to know if their students are grasping the concepts or not. Are they with them in the classroom or not. You might have a student who's not who's not active in the classroom, but you don't know if you might assume that maybe they they're not following, or then you go to their data and see, you know, actually, the student is asking questions that are even more advanced than than what what I'm teaching right now. So that level of insight that we are providing the teachers with would help them even personalize the learning experience in the classroom. Hayete
Can you imagine how exciting and revolutionary this could be for educators? You do not have to wait for the test. To understand where your students are. You can adapt your teaching so that when they actually get to the test, they can succeed. Cause after all, it's about teaching and getting people to adopt the learnings, not to fail them on the test. What you see there is will the AI tutor replace human teachers? Of course, they won't, but that's the first reaction. That's the natural reaction. Is it complimenting me? Or is it aiming to replace me? You always have to deal with that kind of anxiety when you implement an AI system. And this is a great example here. Not only does this AI tutor help students. This tool and the valuable data captures helps teachers be better teachers. Ensuring that the tool is safe for student use and age appropriate is a huge task. And that brings us to where they are now in the development process. Raghad
In the UAE, we have a a teacher guideline or code of conduct for our teachers that we are applying on our eye tutor as well. We've done a few a few hackathons as well to challenge the tool, and turn it, and we're taking this to the granular level of details this is where we we have the real test. We're working with our partner ASI and and following these these values that are very clear to us within our education system to also make sure that it's integrated within within the tool. And this is where the hackathons are important because you put the tool in in a real test. But we're following the same code of conduct that we apply on our teachers, the same guidelines that we have on our education system. With our students to make sure that this is completely safe. We're taking this as a talent. We have another hackathon next week with our students, and their challenge is basically to hack the the AI tool. Ask their job is to ask the tutor questions that they can't answer. Or they might answer incorrectly because we want to make this tool as, relevant as accurate as safe as possible for our students. So we're seeing the impact we're adjusting. Then we will launch again, and then we go back, and then we we build our our research and dashboard along along with the tool as well. We see it as an experiment It's like we are building a a plane while we're we're flying at the same time. This is where we're at now. It's very exciting. A very promising tool, and I really look forward to seeing the results of the first phase and rolling it out on a national level. Hayete
Student safety is top of mind for kudos too. Quddus
That's something we keep for fun at all times, possibly. I mean, but as soon as we first use the demo, even the really broken demo that we had towards the end of last summer, we could immediately tell that this was a technology that could be dangerous if not just, you know, responsibly deployed. One of the key focuses that we add and and is really around security scalability and making sure that we adhere to all the responsible AI principles that are out there. So we we spend a lot of time on that before any student even touched the platform goes through weeks or or even months sometimes, of red teaming and making sure that it's it's sort of aligned to principles and aligned to the Greek limit, really make sure that's holistically as responsible if a deployment as it can be and having a partner like Microsoft allows us to focus on the other paradigm of security and scalability as well. You know, security, the really large part especially when you're dealing with students who may be as young as fourteen, thirteen, and eventually even younger than that, having a part of, like, Microsoft allows us to approach you from tech perspective and an infrastructure perspective, but, having a partner with the UAE Ministry of Education allows us to approach it from a pedagogical educational and really just a safety perspective where is this response to being deployed and and be a part of the pilot phase is exactly figuring out how to do that right at scale, as much as ASI believes in the responsible deployment of AI. And one of the things we also believe in is the deployment of AI is almost a must. It's scary as technology may seem at first sight, the benefits of the technology or incredibly immense. Me personally, I'm a big believer that we need to sort of accelerate the technology, maybe not to the point where it's taking over the world, but right up to that point where it's as helpful and beneficial as it can be. Nothing is used publicly until it's consistently red teamed and tested thoroughly. Very, you know, we we spend more time on testing sometimes than we actually do on development, responsibility, I principles, and responsibility as a whole is something that we take very seriously. We've taken very seriously even before Hayete
So with safety top of mind, they're building and testing this tool with a select number of students and clear impact measures. The process of red teaming involves assigning a team to intentionally misuse the tool and then report back about the vulnerabilities it discovered. It hasn't been rolled out at the national level. At the same time, they're focusing on research. Coming up with the right methodology, the right way to assess the of this tool, making sure they're selecting a diverse sample of school and students or joining the pilot or the soft launch of the tool assessing the learning impact and adding the measures required in the dashboard. And they're focused on answering questions from students and parents and most crucially teachers. Raghad
The first question that teachers asked was what's going to happen to us. What does that mean for teachers? If you are working on an AI based sort of teacher slash teacher, what does that mean for us, for our teachers. Right? And the key answer to this is this technology is supposed to support us it's going to support our teachers and students. It's not going to replace anyone. If anything, it will empower our teachers. We want to support our teachers. And one way to support them is to provide technology that would help them understand their students more and also support our students outside the classroom. Like, at the end of the day, They're not using the tool in the classroom. This is a tool that would be accessible for them after school. Not everyone is privileged to get access to a a tutor that would cost. I don't know between thirty to two hundred dollars. Right? Or not everyone would have educated parents that would be able to teach them, grade eight, grade nine, math, or science. Right? So this is what the tool is for. Hayete
For kudos, the best way to get feedback is to show not tell. Quddus
I mean, the the best way to do it is let them use it. Yeah. I I think using this best tool, like, when we've seen that hesency from teachers or from parents, we we show them the data. Right? We show them how it truly helps students at scale. We give it to their own students, or their children and and let them use it and play around with it for a while, and teachers, parents, schools. They can immediately see this change an attitude towards learning that the students sort of, over time develop and build. Now that's something that's really exciting not just to us, but to the schools as as a whole. Right? If you're able to make learning truly fun and engaging. You know, that that that in itself is really the hardest possible problem you can solve. Right? If if I was to name, you know, out of all my friends, five people that truly enjoy learning. I'd probably have to spend some time thinking about it. So the the best answer really I can give you is let them use it. In most technology, that's that's usually a pretty good answer. Let them use it. And and if the after they do use it, if they're still pretty hesitant, it's probably not the best kind of technology. Maybe you need to get back to the drawing board a little bit around within an experiment a little bit more. And when we do hear that bad feedbacker still continue the hesitancy, we never blame that on the actual person that's using it or the parent or the teacher. Completely our fault and our responsibilities. You know, the tool is designed and developed with students, teachers, and parents in mind. Right? It's not we're not just a company that goes away for a few years and develops it with a bunch fending years, and we come back, and tell them all to use it and force it upon them. Teachers are consistently part of the development from the beginning of the company. Students have always been a part of the company since the beginning. I mean, I was a student myself. When I first used it. So, when I first developed the company, so the end user has always been a part of the development process of the holes. Hayete
I just love his, let them use it confidence. It's kind of infectious. You can't help but smile and share him on here. His well placed confidence in pragmatism seems in perfect balance. Is confident because he's kept students and teachers in the Quddus
loop of the Hayete
development process from the beginning. Big change requires buy in. And this project is getting a boost from the UAE and their forward looking minister of education. Quddus
We saw an announcement by the minister of education. His excellent C face all. We, you know, and we, we really were able to see, that, you know, he was really forthcoming in his vision for AI. A lot of education systems in the world were very shy about AI. I mean, you saw, you know, sort of districts in the US banning AI in fact, but, the minister here was very forthcoming and very, you know, sort of outspoken about his willingness to embrace AI that they're possibly the greatest partner that we could find the greatest first government to to work with. There, you know, vision that that we think, like, not a lot of people have in the world, but their vision was very straightforward and and something that we could definitely assist in. And it was a vision that, you know, I had two years ago, two and a half years ago when I first founded the company. Raghad
It was that key message from his excellency once, charge was released and everyone was freaking out. I remember the tweets. Everyone was like, what do we do now? Students are going to teach. Our learning outcomes are not as good to begin with, what's gonna happen now, etcetera. And he was like, no. This tool is here. And even his first So during that time when we announced the two, the speech that, he made was written by tragedy, and that was, like, early twenty twenty three. So that sends a very strong message. And definitely, that has helped a lot. It helped us a lot. But at the same time, because we are a government entity, There is also this responsibility that if we are going to introduce this technology to our students, there is that responsibility of making sure that this tool is hundred percent safe for them, right, to interact with. It's, hundred percent aligned with our values in the UAE. It's it understands their their language, the way they learn, etcetera. So that responsibility is is also, like, comes with with that flexibility and that, like, pioneering approach of, yes, we will we will we will work with this technology, we will implement this. We will do this while keeping us grounded at the same time. And that's this is what you need. I think the the balance of both. Hayete
And they're already looking at impact beyond the UAE. Raghad
We're very conscious of the connectivity challenge that a lot of students face. We're we're lucky in the UAE that our students already have that bring your own device policy. So they have their own devices, and that's how the our education system thrived within COVID. It was very easy for us to to move fully online. But we are very conscious of this even with that infrastructure in place. Because we're thinking of everyone within the ecosystem. Right? And we're we're also thinking of what's beyond the UAE. We're we're pioneering. We're starting this initiative, and we want we want to help solve for global education challenges. And we're we're very conscious of that, spec specifically on the low band bandwidth requirement for us with our students and being able to access this offline even. That's in one element. Another element is we're we're still looking into how are we going to roll it out to our students. We've talked a lot about the cost of hiring a private tutor. This will reduce the cost, like, twenty times from what's on average a student would pay for a private tutor. And then you will have students who might actually need this and wouldn't be able to pay, for example, for the subscription fee. And this is what we are solving for as MOE right now. How can we come up with this model where students can have access. Is it going to be that everyone needs to subscribe? Are we going to secure subscription for all all all all of our students, some of our students who are who Quddus
are who are in need. That's still a discussion Raghad
for us, but for sure, it's it's going to solve for a lot of the challenges when it comes to lowering the cost on on everyone. So we are already in discussions with a few government entities and ministries of education who are interested in experimenting with this tool and introducing it to their own markets. Hayete
They are still in development of the tool, enrolling it out on a national level. But what has been very clear for a while and has become even clearer in the last few months since the introduction of Chad DPT? Is that the way students are learning is changing already. Raghad
If we want the education system to actually be relevant and help our students in their learning and development, we need to adapt. Maybe a lot of government entities are, and ministries are, like, afraid of what this technology would mean for them. But it's even even ministries and even, like, the education system as we know it now, needs needs to, in itself. Otherwise, we wouldn't be relevant. A lot of people are already talking about how relevant it is. In terms of what you're learning in school versus what isn't happening in real life. And this is the conversation that would we've been having for decades. Do you want to distance yourself more from your students and the people that you're supposed to serve or do you want to be relevant and and help them as well? Which is the purpose of of all educators. Right? Hayete
So what advice does CUDas have for education leaders interested in taking the plunge into this technology? Quddus
I would say just go for it. You know, I think one of the biggest barriers that people my age have or at least when I was speaking to friends is for some reason, they think that, you know, to build a company or to build a product, you need to be this super genius or you need to be, you know, Steve Jobs. But the truth is there's nothing in the world that's made by people any smarter than you are. Right? You look at the roads, you look at the cars, you look at the stop signs, they've all, you know, being developed and built by people that were very similar to you at one point. As much as you don't think that's true, it probably is true to a very high extent. So really just going for it and then just making sure you continue going for it. I mean, when we first started the company, today, what we do in the company looks very different to what we did when we first started, but we just kept going at it. We kept, you know, iterating. We kept listening to our users kept sort of iterating on based on what our user said, and we never sort of gave up and and we built the right team to do. That's the biggest advice I can give is just, you know, going for it and then continually going for it, you know, after the fact that maybe you've broken down once or you failed once, just keep going for it. And eventually, you'll succeed. It's very cliche, but, it is the sort of biggest barrier I see, and I try to convince all of my friends to start company, they're build products, and they're always like, oh, I need to be a super genius to create an app or I need to be a super genius to create a company. The truth is I'm not a super genius, and most of the founders I know, you know, as smart as they are, they're, you know, they're not really super geniuses. They're just people that are incredibly determined and and very, very enthusiastic about a problem they they wanted to solve. So, I think that's all the qualities you need. And if you go for it, you'll likely sort of remove any doubt that you may have. Hayete
It's amazing to see the possibilities with AI. And if you were to think about topics and scenarios, I don't think any one of us think learning, which is such a personal experience, would be a good target for AI, but If you look at the ministry of education in the UAE and the approach they're taking, they're really embracing AI, to enable all of their students and tackle a very big problem for the country, which is tutors is not affordable for all. But they wanna create equitable opportunity for all citizen in the UAE. So they're embracing AI. And you're not forgetting that at the center, it's about the humans, the students, the parents, the teachers, and they're ensuring that what they're building, this AI system, not only is providing great outcomes on the learning path. For all those students, but that people are comfortable with it and them being very responsible in their approach. Just testing it, They're ensuring each step of the way that this is aligned with the values that they've from the start defined. It is not about replacing humans. It is about complimenting and helping everybody at all level. This is a true way to make learning accessible to all Hayete
you for listening to pivotal. I'd love to hear your story in your pivotal moments, so don't hesitate to follow me and share on LinkedIn. Information is also available in the show notes. Our show is produced by large media. That's l a r j media. Special thanks to Lynn Yang and our partners at We Communications. 
