AI Learning Companions For Complex Skill Development | Samreid Manez
Disclaimer:
The content shared is to highlight the passion and wonder of our guests. It is not professional advice. Please read our evidence-based research to help you develop your unique understanding.
💕 Story Overview
💕 #MAGICademy S3E4, we had the pleasure to meet @ Samreid Manez, founder of Valearnis. Integrating AI learning companions and non-player characters (NPCs) in a gamified learning experience can potentially enhance complex skill development.
Story Takeaways
AI learning companions and Non-Player characters (NPCs) have the potential to identify player strengths and weaknesses, provide targeted context-rich environments that simulate real-life scenarios, helping players develop specific complex skills through dynamic interactions.
Progress is essential for engaging players in games, creating a sense of achievement and motivation. Effective progression systems offer clear goals and rewards, preventing players from feeling their time is wasted, and fostering a rewarding cycle that sustains engagement and satisfaction.
AI can foster a more egalitarian world by democratizing skill development and leadership through AI companions and gamified learning. This reduces reliance on traditional hierarchies, promoting a diverse form of leadership based on knowledge and skills, and empowering everyone to contribute meaningfully to their communities.
Samreid’s MAGIC: Perseverance
#AIAvatar, #AINPC, #AILearningCompanion, #PersonalizedLearning, #GamifiedSkillDevelopment
-
00:00 Introduction to Valearnis
03:25 The possibility of skill development without tutors?
07:44 The RPG Elements and the Importance of NPCs
13:13 Progression is an important element of engagement
15:39 Making Interactions More Dynamic with AI
19:22 AI's Potential for a more Inclusive and Diverse Society
24:42 Samreid's MAGIC
-
AI learning companions can adapt without the burden of an ego: https://youtu.be/MJyAJGngud4
AI as Non-Player Characters can make learning more dynamic in a gamified environment: https://youtu.be/-qGF3zvHy2I
A scenario where AI NPC makes learning more dynamic: https://youtu.be/kT1uIulrdug
The engaging power of real time learning progress updates: https://youtu.be/SqnVeVR0jp0
Can AI become multi-cultural with a global citizen mindset? https://youtu.be/DT8-i5BxImE
Let's revitalize our hope and inspiration toward technology:
-
Pérez-Ortiz, M., Novak, E., Bulathwela, S., & Shawe-Taylor, J. (2021). An AI-based Learning Companion Promoting Lifelong Learning Opportunities for All. ArXiv, abs/2112.01242.
Hu, Y. (2024). Improving ethical dilemma learning: Featuring thinking aloud pair problem solving (TAPPS) and AI-assisted virtual learning companion. Education and Information Technologies
-
Samreid Manez is the CEO and Founder of Valearnis, an educational platform that integrates AI with traditional learning for a global, student-centric Education-As-A-Service. He bootstrapped Valearnis using technologies like GPT-4 for a virtual coach and learning companion. He was the youngest Councillor in the Shire of Boddington, advocating for sustainability. Born in Perth to Malaysian immigrants, Manez holds a Bachelor of Science in Applied Geology from Curtin University and qualifications from Waseda University and the University of Illinois.
-
Jiani (00:04)
Hi, hello, Sam. Glad to have you at MAGICademy podcast.
Sam (00:09)
Yeah, thanks for inviting me. I've been really excited to hop on the show.
Jiani (00:13)
Excellent. when I first came across the startup that you're building, it gets me super, interested.
Can you tell us exactly what this platform does? Like what?
Sam (00:23)
Sure. So right now we're at the, I guess, like the pivot moment. So initially our platform when it launched, it was for grade improvement, homeschooling support, and even like a tutoring alternative. So tutoring is, is a costly industry because you have to pay someone to one-on-one, you know, putting their time to teach your kids something outside of school. If you feel that your kid,
needs extra support, or that they're not being I guess, stimulated enough in school or they need to go higher in the same area. You might be you know, inclined to, to hire a tutor, but you're usually paying, you know, $40 an hour, $50 an hour, like it's expensive. And it definitely adds up for families. So the other idea is, you know, we
we've basically made an application that students can use after school and they can, they can learn those, from those areas that they might be lacking in, they can learn at their own pace and they can get that extra support for a fraction of the costs. So that's basically what we're trying to do there. But homeschooling parents also have, have trouble where, you know, they have to rely on really old textbooks or they have to rely on
other parents to give them resources on how to homeschool their kids. So we're just trying to also help them out a bit and make sure that they have the resources that they need so that their kids can learn in an engaging way.
it's to replace the idea that one on one tutoring is necessary. So it's an application that has lessons, quizzes, all sorts of little like bite sized content for the kids to go through. And it makes sure that, you know, it's for the year level, it's for their specific subject areas and the kids go through them. They're incentivized to participate because, you know, they're getting experience points, they're leveling up, they're moving through.
a platform and they, you know, they're, they have the characters that are leveling up. They have incentive to, I guess, engage with the platform. And we're really just trying to mix in the elements of what makes a learning platform effective in terms of education, but then also what makes an experience effective in terms of engagement. So we're trying to find a balance between
video game ideas and learning platform ideas and trying to blend it all in one so that students can have an actual engaging time with their education and they feel that they have self-determination, they feel that they're empowered and that's really important throughout platforms. So yeah, that's pretty much what we're trying to do.
Jiani (03:25)
I see. So the ambition of removing the need for tutor is very ambitious. So you mentioned about, yeah.
Sam (03:32)
It's definitely ambitious. Yeah, it's something that I guess, you know, obviously, tutoring will be around for a while. But if we can free up certain households from needing to spend that amount of money on tutoring, that would be great. But yeah, it's basically just the idea that we are that we're trying to focus on is that it's an application that
we all would have liked to have used if we were in high school ourselves. So yeah, that's basically what we're trying to achieve there. If we were a teenager and we were using an app, we probably wouldn't have liked to have used iXcel or Maths Online or something like that. So we were trying to make an application that actually has that incentive structure in it, that...
makes me want to engage without being forced to engage. So that's what we're really trying to bake into this that we want kids to learn and we want to make sure that they learning well. But we want that to be an enjoyable experience that they feel that they have control over. And they feel that they're getting a lot out of. So we have you know, yeah, experience points, characters leveling up on demand reporting, which is also a really interesting area where when you're in
school at the moment, they have a thing called, especially its summative assessment or summative reporting, wherein you do your whole semester of work. And then after that, you're kind of, you're given your report card. And then I guess like most students, this is definitely true for me, you know, you get your report card, you see what you got. And then
That's about it. You know, you get your A, your B, and then you go for your two weeks holiday, your three weeks holiday, you come back and you don't really remember what you got the previous term. So the score you got doesn't really change anything in how you learn. It might make a tense argument at home. It might give you a difficult, you know, time with your parents or something. But most of the time, it doesn't really do much. So
What we're really trying to do there is if a student can know how they're doing in real time, then they can adjust and they can make those changes before it gets to the stage of them failing or gets to the stage of them not understanding the content at all that their peers might be understanding and moving forward with. So if they can check that, Oh, okay. In real time, my algebra is not great right now. You know, my understanding of geometry, excellent.
But algebra, I'm not doing too well. And I know that right now that it needs help. So then I would start to focus more on that. My parents know it as well. Maybe my teachers know it as well. So that data is really important for everyone, the student, the parent, the teacher, and they can make those changes before anything significant happens.
Jiani (06:46)
Hmm. So it's kind of like, um, like learning in a full sense, like learning, make sure that everything that you learn, you have all the support right there. Um, when you need it, um, so you can move forward. So it's like a competency, mastery based kind of learning trajectories.
Sam (06:55)
Exactly.
Exactly. Yeah.
It's definitely mastery based. Yeah, because there's been so many times when even I in school have felt that, okay, I'm not I'm not understanding this and everyone is kind of moving forward with it. And you feel like there's a sense of dread when a whole classroom is moving forward with something that you don't feel that you've gotten yet. And that's really what we're trying to solve here that making sure that you know, every student's journey
is moving at a pace that they are comfortable with and that they are actually understanding the content before they move forward. So yeah, that's ethos there.
Jiani (07:44)
Very interesting. You mentioned about using, you know, we talked about using AI as a teacher or teacher's assistant. You also talk about using AI as a non-player character in a game or a simulated environment. Can you tell us a little bit more of how you are currently leveraging AI to be an NPC?
Sam (08:09)
Oh, yeah. So the I guess the like future outlook for our platform is yet to be that cool intersection between game and learning platform so that you know, the students are engaged, they're interested in the platform, they have the character, they're leveling up, they're moving through the platform, and they have an incentive to keep interacting with the platform. And in terms of you know, outside of the educational aspect of that, we could move towards the
RPG elements of that. So right now we already have RPG elements in the platform where you can level up, customize your character, you have pets and all that. So there's items to collect, there's those sort of elements. But I found that as a player of games, the most interesting part of an RPG is always the NPCs, the people that you interact with that give you your quests, that join you in your adventure.
that give you all those sort of interesting experiences, they're the most interesting aspect to me and putting them in a platform like this, obviously is important and something that we've been working on for a while, but AI could really change that experience to make it more dynamic. So let's say, without AI, I make the idea of an NPC. He's maybe a...
He's a blacksmith or something, you know, he makes armor and you approach him and he gives you some interesting quest to go and, you know, find him some special item to then he'll make you a cool sword or something like that. Now, I would have to write that entire experience and I would have to write all of the dialogue for that. But the issue with that as well is that for the player, I would also have to write the dialogue choices.
for the player. You couldn't actually say what you wanted to him. I would just give you three or four options for how you would interact with him. I would give you maybe some nice options, maybe some rude options. You could be kind of like, it gives you some flexibility there and then he would have different dialogue depending on how you talk to him. But it's not really your specific way of wanting to talk to him. It's just kind of finding like the best fit.
that I've already scripted for you, I've already written the story for you, and you're kind of just choosing which element of what I wrote is most interesting to you. That's the old way of doing it, that's how video games currently work. But with AI, I don't have to do that. I could just, instead of scripting anything for him, I would teach him as an AI that, okay, you're a blacksmith, you have these interesting quests to give out.
Jiani (10:48)
Hmm.
Sam (11:03)
you have these interesting incentives and these interesting goals as a blacksmith. And when you interact with the student, you want the student to be, I guess, telling you specific things, to prompting you in specific areas. But I'm not writing your dialogue, I'm just writing your story. You are this, and now when a student comes to you, he can say whatever he wants. Or she can say whatever she wants.
they could have a really interesting dynamic conversation with you. And they could say completely human natural language, and they don't have to choose dialogue options. They can say whatever they want and have a very personal experience with him. So that, you know, they can have a very real feeling conversation instead of it just being, Oh, here's some options to choose from. And you know, this is, these are your options and you can go from there. You can basically just say whatever you want.
and he'll react in, you know, have a way that he's been programmed to react in. But, yeah, that's basically what we're trying to do there. Trying to make experiences that feel real and that are giving students, you know, that sort of immersion in a platform and with a game that they feel that, oh, wow, okay, that's interesting. He's like a real person. He's like a real blacksmith. Maybe he's a bit grumpy and that's kind of cool. And maybe I could be nice to him or I could be rude to him.
And you know, if I'm rude, maybe he doesn't give me the quest, maybe he tells me to go away. It's a good learning experience with a kid. It's an interpersonal experience with a kid. Uh, you know, it gives them that experience of if you talk to someone in a specific way, they will have different reactions. And if you are nice to people, maybe they'll, you know, be nicer to you. And if you're rude or, you know, or not, then, you know, maybe you'll have different reactions, but
Yeah, that's basically what the idea around AI could do there is that you won't have to script interactions anymore. You would have more natural interactions that would feel a little bit more like real life.
Jiani (13:13)
Excellent. what are some elements that really engages players in a good game? How do you, how do you characterize a, how did you characterize?
Sam (13:22)
Mm. Progression. Progression is really an important one. You have to feel that you are moving upwards, that you are moving forwards towards something. How school and university usually works is that you start technically your semester on 100, technically, and then you get your first assignment and it brings you down immediately. You know, you're at 70% or you're at 60% or whatever.
and then your average just keeps changing from there and then you end up wherever you end up by the end of semester. Now, if you do an assignment and you get a 60% on it, you don't feel like you made progress, you just feel like you only understood 60% of something. So you don't really feel like you're making a sense of progression there. Whereas with an application like ours or in a video game, you do something and you gain experience points.
Now, if you did that thing really well, you would get more experience points. But if you did it mediocre, you still get some experience points. You're still moving in upwards direction a little bit. So you're incentivized to do better because you'll get more. But even if you don't do that well, you still feel like, oh, okay, I mean, damn, I only got you know, I only understood half of this lesson. But I'm still given, you know.
500 experience points or whatever. So at least doing this lesson was worthwhile. At least there was a reason for me to have done this because it's still moving me in an upwards direction. Whereas if you're in school, the incentive is you have to do the assignment because you're forced to, because the teacher said you had to. And if you got 60% on it, then okay, fine, I got 60%. And that is where it is. But you're not getting anything, you're not moving upwards, you're not building from there.
So there's no upwards momentum. And that's a really engaging feature of video games and of what we're trying to do as well. Where you feel that no matter what you're doing with a platform like ours, you're moving upwards, you're progressing, you feel like there's actually a point to this more than it just being, oh, I just had to do it.
Jiani (15:39)
Hmm. Do you, um, in terms of the quality, um, so if you prompt an AI to be a non-player character, the NPC, do you, do you think the quality of them improvising based on the students' responses are consistent? Have you encountered any particular challenge in that sense?
Sam (15:46)
Hmm.
Okay, so with in terms of the AI powered NPCs is obviously something that is not implemented yet. It's something that we're working towards. But the idea of the quality of it, that's obviously something that you have to put in the work to put in those parameters around to make sure that you still need to write an overall story, you need to write enough data for that character to base itself off.
And as long as you have that there, then the quality will still be maintained. But if you don't give it any of that data, if you don't give it any backstory or any goals or any world building, then it will feel a bit flat, a bit bland. And yeah, you would lose a little bit of that quality. So it's just about putting in the work to create a world around these characters and to create a backstory for them, and then let them go and play through AI.
But you need that guardrail around it. You need that, I guess, that structure around it. So it's like you make a structured world, you make a very well-defined world, and then you let these AI agents go and play in them and interact with the players, the students.
Jiani (17:30)
Interesting. The guardrails, what, how do you structure a guardrail? What are some key elements that needs to be considered in the guardrail document?
Sam (17:42)
Again, it's as if you're doing an employee handbook. I would say that's the easiest way of explaining guardrails. It's just treating it as a member of a team or a member of a company. And then you just have a very strict employee handbook rules around how it should be interacting and how it should be speaking to people and the sort of output that you want from it. You want it to talk in a casual way.
Jiani (18:02)
policies.
Sam (18:12)
Do you want it to talk in a very professional manner? You could even put in, you know, if the student is speaking to you in a very casual way, then you can speak back in a casual way. Whereas if the student is speaking to you in a very professional setting, then you should match their way of speaking. You can then output, you can put parameters, obviously, no swear words, no, you know, you can't give the student any URLs, no links to click on anything.
If the student asks you anything that is inappropriate or uses swear words in the question, then you would then say, hey, sorry, this is an inappropriate to talk about. Let's move on to something else. So all that stuff is important and that has to be put in.
Jiani (19:00)
I see. So it kind of like a policy document, like it's highly customizable and situation-based. The deeper, the more thorough the thoughts are putting into it, the better, the stronger the guardrails will be. Excellent.
Sam (19:04)
Exactly.
Exactly.
Jiani (19:22)
Hmm. Do you think that would help with leaders to help leaders to become better leaders?
Sam (19:30)
I think so. It may end up driving us to a society that's a little bit more egalitarian and maybe a bit less leader focused actually. It might just lead us to a society where people are kind of all on a similar level and they're all respected and we may not, I guess, idolize specific people anymore. There may not be, you know, cults of personality anymore when you have, when everyone has an unbiased companion that can...
sort of tell them the raw truth that they might need to hear rather than depending on leaders and stuff like that. But I think it will definitely help society function in a better way. But yeah, so I'm very excited for the potential of it.
Jiani (20:17)
Excellent. Do you think AI can truly be multicultural? I know currently the AI's database is heavily biased or one culture centric. Do you think AI has the possibilities to be kind of more cross-cultural, kind of global citizen mindset?
Sam (20:37)
It does. It definitely does. It does. It depends on which AI you talk about. So open AI definitely was trained on a very Western data set. And the data that it was trained with is very Western centric. But that is just a data problem, not an AI or LLM problem. So the problem isn't with the technology. The technology there has infinite potential.
the issue is with what it was trained on. So as long as other companies out there or other existing companies start looking at, Global South and start looking at different cultures, different countries and training their data on, train their AI on that data as well, will definitely be in a much more equal place. And I think also more, the AI would be smarter
If it wasn't just trained on one specific cultural data set, it would be a lot smarter if it could get in, you know, all those sorts of different ways of thinking from different cultures around the world, so that it can then adapt to every difference is not scenario, and that it could take into the considerations, things that people may be in the West just don't take into consideration. And if the AI was only trained on Western data, then it would only think like a Western person. Whereas if it was trained on.
the entire world's data and data from all sorts of different cultures, different languages, it would have logic and understanding and reasoning from so many different diverse cultural sets. So it would not only be a more inclusive AI experience for different people, even if you take that out of consideration for a person in the West using it, it would actually be a better experience having an AI that was trained on more diverse data sets.
because it would technically be smarter in how it could think and how it could reason because it would have a larger knowledge base and other ways of thinking. Now it learns better when it can have opposing views, when it can have other sorts of ways of formulating ideas baked into it that it wouldn't have if it was only trained on a specific data set from a specific culture.
That is the issue with the English side of the internet is that it's very Western focused and it's very first world. But there's a whole host of data out there, in Asia, Africa, South America, there's all sorts of data sets out there that's available that we need to start training the AI on, different AIs on it. And we need to encourage companies in those areas of the world, in those countries,
developing their own AI, and then we can work together from there. But that would be the best case solution that we train it on as much data as possible and on as much diverse data as possible. There's no point training it on a million Reddit pages from, you know, all from people that are in California and in Texas and stuff like that, because there's a very specific way that those people think, and there's nothing wrong with that, but you're not getting.
other ways of thinking, the way that people in China or India or Indonesia formulate ideas are all completely different from each other. The way they use their languages are different. The way they debate is different. So it's really important that if we're trying to make a super intelligence or some sort of intelligence that could be the collective of the human thought and human ideas, that we actually give it
the resources to do that and we give it the potential to do that by feeding it as much diversity as we can.
Jiani (24:42)
a super brain, multicultural super brain. Excellent. Great. Now let's kind of get into the final part of the conversation, which is the magic. We wanted to, the whole reason behind it is we wanted to kind of ask people to look within rather than.
Sam (24:48)
Exactly.
Jiani (25:11)
out, like projecting out and looking for, like you mentioned, a particular leaders who dominates the culture, um, either economically or politically, um, find their own inner light. So my question for you is when you were, um, let's, let's kind of jump back in time. And when you were like 12, 11 years old, what did you enjoy doing? What were you having fun with?
Sam (25:40)
I really enjoy playing video games. That was always a really big thing for me. And I definitely still do. So I'm happy that I've managed to do something that incorporated the elements from that industry that I found so interesting, that I found so engaging. And I've managed to then take that and put it in a different industry in a way that...
Jiani (25:44)
Hahaha
Sam (26:09)
engages people in a way that enhances education for students. So I'm very happy that I managed to take in specific lessons from how those games worked and why kids love them and move them to something else. So yeah, I'm very happy that I managed to do that.
Jiani (26:29)
Hmm. Yeah. Excellent. Um, do you have to overcome some sort of challenges in your lives, in your life? Everybody has one life.
Sam (26:39)
in my life? Yeah, sure. I mean, I would say that, you know, going straight into this straight out of university, I didn't work in geology as a geologist. I could have been making, you know, a lot of money straight out of university. And I didn't go into that. So I didn't save money to start this. So it's been a very long journey to get it where it is now.
Um, I would say that, you know, my magic could be more perseverance, um, and just sticking with something and really just, you know, no matter what, you know, we, we tried without an app at first and we were told no, and then, you know, we, we made a demo and we went from there, but it was a really long journey, uh, and, and it was a really long process to get to where we've gotten now. Uh, and I think without perseverance, it, it wouldn't have, it wouldn't have reached that.
Jiani (27:38)
Excellent. Thank you very much, Sam, for sharing your stories. Is there any particular thing that you wanted to share that my question hasn't been able to draw out yet?
Sam (27:55)
No, I mean, this has been great. I think, yeah, the main thing I wanted to share is just that if we could, I think in the 60s, there was like this idea that science and technology was fascinating and inspiring. And you know, we're going to land on the moon and we're going to have, you know, the world of tomorrow and we're going to have robots in our houses. And there was so much hope and inspiration tied to science and technology.
that I think in the last few decades we've really lost. So I would say that it would be really nice if we could start moving towards that again, having that hope and inspiration, optimism for the future and for the potential that technology and science can bring us, rather than just focusing on the negatives of it and actually being inspired and inspiring others around it would be.
the ideal way of thinking from my point of view.
Jiani (28:59)
Excellent. Thank you, Sam, so much for sharing your insights, your wisdom, your thinking and your optimism for a more diverse, equal future with the help of AI. So such a great chat. I hope you enjoy. I hope you enjoy the chat as I did.
Sam (29:17)
Thank you so much.
I loved it. Thank you so much for having me.
