2979: AI and the Future of Personalized Education with Memrise

[00:00:01] [SPEAKER_01]: Could Artificial Intelligence be the key to unlocking a new era of personalized education?

[00:00:09] [SPEAKER_01]: Well today I'm going to be joined by Steve Tolling and he's the CEO of Memrise, which is a pioneering company in the

[00:00:17] [SPEAKER_01]: Ed Tech space and they were recently ranked the world's second leading Ed Tech Company by none other than Time Magazine.

[00:00:25] [SPEAKER_01]: An understeves leadership, Memrise's harness day eye to revolutionise language learning,

[00:00:31] [SPEAKER_01]: crafting tailor-made educational experiences that adapt to the unique individuals needs.

[00:00:38] [SPEAKER_01]: A with over 70 million users globally.

[00:00:42] [SPEAKER_01]: Memrise uses advanced memory techniques and the latest in Euro science to dramatically enhance learning efficiency.

[00:00:51] [SPEAKER_01]: It's a big step away from that one size fits all approach to education.

[00:00:55] [SPEAKER_01]: So I want to learn more about how this technology works, what could it mean for the future of education?

[00:01:02] [SPEAKER_01]: I would say mean for both teachers and students.

[00:01:05] [SPEAKER_01]: The cost of hosting a daily show for 140,000 monthly listeners can be significant.

[00:01:11] [SPEAKER_01]: And I've like the take a moment to thank those who make it possible for me to keep delivering this content every day table.

[00:01:17] [SPEAKER_01]: And I also want to talk about the fact that legacy DRM failed to securely enable external collaboration on sensitive files.

[00:01:25] [SPEAKER_01]: And I think it's important to recognize organisations in this digital age face a somewhat risk, trust contradiction.

[00:01:32] [SPEAKER_01]: Yep they must share content with untrusted third parties while also protecting that day.

[00:01:37] [SPEAKER_01]: So it's time for a more modern DRM solution one that solves this dilemma but without compromising security and productivity.

[00:01:45] [SPEAKER_01]: So collaborators imagine editing files externally without losing control.

[00:01:51] [SPEAKER_01]: Stream zero latency video renditions to authorize users but without any actual file transfers needed.

[00:01:58] [SPEAKER_01]: The co-author can view it remotely while you retain full ownership.

[00:02:03] [SPEAKER_01]: Your files stay protected in cart work secure unclave, ultimately they never leave your environment.

[00:02:08] [SPEAKER_01]: So you can stay goodbye to data, leakage risks and experience seamless editing across all file types.

[00:02:15] [SPEAKER_01]: Not just native applications without any plugins required.

[00:02:19] [SPEAKER_01]: So say goodbye to deployment headaches, file transfer risk collaboration barriers and all those productivity constraints.

[00:02:25] [SPEAKER_01]: An experience a more modern way to collaborate on sensitive content sacrificing control or security.

[00:02:31] [SPEAKER_01]: And you could do all that by visiting kightworks.com to get started.

[00:02:35] [SPEAKER_01]: And with my thank yous out of the way, I'm now officially excited to introduce you to today's guest.

[00:02:42] [SPEAKER_01]: Well, buckle up and hold on tight because no matter where you're listening in the world it's time for me to be your ears all the way to London here in the UK.

[00:02:50] [SPEAKER_01]: Where Steve is going to shed light on some of these innovations and also the broader potential of AI in learning.

[00:02:59] [SPEAKER_01]: So, a massive welcome back to the show is 12 months since we last spoke but for anyone listening that missed act.

[00:03:05] [SPEAKER_01]: Do you just remind everyone listening with a little about who you are of what you do?

[00:03:09] [SPEAKER_00]: Sure well thanks for having me again and my name's Steve Torrey here and I happen to be the CEO.

[00:03:15] [SPEAKER_00]: But come be called Memrise which is a language learning app that to we believe is a bit different from the rest in the pack and that we do everything we can to use technology and to personalize experiences

[00:03:31] [SPEAKER_00]: And make the experiences much like going in country and surviving as it can possibly be from the you know comfort of the device in your backpack.

[00:03:42] [SPEAKER_01]: Well, so pleasure to have you back on the podcast. So that's you know every day I try to take a different topic get people thinking about how technologies transforming the way that they work and live at et cetera.

[00:03:54] [SPEAKER_01]: Are you like everyone's still going crazy about AI is a lot of hype behind it but I'd like to dig a little bit deeper into that and get people thinking differently about it so Steve how are you seeing AI transforming the way people learn and what other key benefits for creating a more personalized learning experience for students.

[00:04:15] [SPEAKER_00]: Well, I think the short answer is in that personalization word that we're throwing about here because that is the way that we we kind of learned our first language.

[00:04:29] [SPEAKER_00]: Is by being absorbed and immersed in the things that are going on in your world in your home with your parents with the things that matter in your life.

[00:04:39] [SPEAKER_00]: And it also happens to be the way that all the polyglots out there would recommend you learn a language which is pay attention to the words and the subjects and the things that matter to you not just words that somebody tells you you should learn.

[00:04:54] [SPEAKER_00]: And that's always been hard to do because we've always been in a mode where we're marching a whole lot of people to the same test or the same piece of content they have to consume is so on.

[00:05:07] [SPEAKER_00]: So we really just got to teach all the same words all the same people and all the same time and that's no longer the case with technology so.

[00:05:16] [SPEAKER_00]: The that's the long we did way to say the short answer is personalization is what AI brings to the table in on a number of different axes.

[00:05:27] [SPEAKER_01]: And that's one of the reasons I invite you on here today because memorize is a language platform that uses space repetition of flashcards to increase the rate of learning but you've been doing this long before AI so I'd love to find out more about how memory memorize leverage as AI to create these personalized experiences for users and also.

[00:05:46] [SPEAKER_01]: The approach that you take and how it can be applied more broadly maybe across the entire education sector to.

[00:05:53] [SPEAKER_00]: Yeah, well so I think here here's how we go about it first and foremost we do have the ability to help you memorize words and know the meanings of words relatively quickly.

[00:06:06] [SPEAKER_00]: But the way we've architected ourselves you get to choose what words matter to you if you want to say you're learning for the purposes of traveling we can gather the words that matter to you there and you can learn those words.

[00:06:22] [SPEAKER_00]: And because AI technology is so robust we can take any group of words that a user happens to know and start to run it into all of the content that's out there the YouTube videos the TikTok videos that are out there.

[00:06:39] [SPEAKER_00]: Because AI lets us filter those things by the word anyone individual knows and we can even pitch people into conversational practice with the words that they know so we can constrain a conversation to something closer to the small personal dictionary that a user or a learner has.

[00:07:03] [SPEAKER_00]: And that's super exciting that we can do that and now before I go too deep into it I'll pause there but I think that is the springboard for how it can be applied in other areas of education and life.

[00:07:17] [SPEAKER_01]: And I love how memorized was it an earlier doctor ever AI with your member tutor so how does member enhance the learning experience for your users particularly in real time situations I looked at bring that to life with some of the work you're doing there.

[00:07:34] [SPEAKER_00]: Yeah well I think there's two really really key things about member right it's easy to think about this as a chatbot or AI driven tool but really you know what matters most is the degree of fear and concern that people have in trying to speak a new language because they're so worried about human judgment.

[00:08:01] [SPEAKER_00]: They're just so panic that people will think that they're done because they sound like a sixth grader or five year old and they they just don't go ahead and use the language that they're learning early enough in the experience.

[00:08:17] [SPEAKER_00]: Where something like mmb that we provide to people allows folks to do the practice that one needs to do without any human judgment there's no human on the other side.

[00:08:31] [SPEAKER_00]: You know deciding whether you've done a good job or not are all planning on whether you're smarter not you can just get in there and practice.

[00:08:39] [SPEAKER_00]: And so removing the human judgment you know main benefit one to get to a way of practicing early in the journey which is main benefit too.

[00:08:51] [SPEAKER_00]: Is a great combination for getting people ready to use the language they're learning and they don't have to wait two years of memorizing 5,000 words before they actually start talking to somebody.

[00:09:06] [SPEAKER_01]: And as a spec will have a lot of people listening maybe that in the education space or used to be in the education space and hopefully we've set up a few light bulb moments here around the opportunities that it can present so.

[00:09:19] [SPEAKER_01]: How can teachers do you think utilize AI to create personalised lessons tailored to each student's knowledge level and and what impact do you foresee this having on traditional education methods where it was almost a one size fits all approach and some of the kids will get it and some of them weren't.

[00:09:35] [SPEAKER_00]: Yeah I think that the key thing is that these technologies can give teachers sort of infinite assistance or an assistant for every student that's out there when you are learning anything you can ask these technologies to.

[00:09:55] [SPEAKER_00]: Now spend some time in an exchange with a student about a particular topic and then these these these large language models can watch the conversation and decide where.

[00:10:09] [SPEAKER_00]: They need to guide the conversation in order to strengthen things that might be weak so it can apply to history lessons it can apply to language learning it can apply to math it can apply to all kinds of areas where.

[00:10:26] [SPEAKER_00]: If people are understanding things and getting questions right then you keep giving them more advanced things to work on and if they're not getting things right you double back over that and yes different questions or you provide different guidance and when you can constrain a subject the topic question.

[00:10:46] [SPEAKER_00]: And and pitch everybody into a large language model that is constrained by that topic you can do incredible things in terms of allowing the technology to walk people through their own path of learning.

[00:11:05] [SPEAKER_01]: So let's take a student that maybe doesn't respond to those traditional teaching methods in what way can AI powered to assist those struggling students with difficult homework questions and how does this democratize access to the quality of education.

[00:11:21] [SPEAKER_01]: And I think it's also important to point out we're not talking about replacing teachers here is an additional tool so they can spend more time with these students right.

[00:11:29] [SPEAKER_00]: Absolutely it's definitely not replacing teachers first and foremost any of these tools need to be set up and made ready to be helpful to the teacher about the subject that they're teaching so first and foremost.

[00:11:46] [SPEAKER_00]: The the teacher decides what's important and where they're taking the class so definitely not replacing teachers only assisting them.

[00:11:57] [SPEAKER_00]: But yes for sort of an example of how this can work well so for example let's just take something outside of language learning like math right and you're working on something say division long division.

[00:12:13] [SPEAKER_00]: And you can ask people questions and those that get the question right can move on to the next more advanced thing.

[00:12:22] [SPEAKER_00]: And those that don't get it right can go through a sort of step by step discussion about how you do long division for the problem at hand.

[00:12:34] [SPEAKER_00]: And in this way the people that understand it and are racing forward can continue to do work those that are confused or are not quite getting it just yet can work on the problem at hand until they do get it.

[00:12:50] [SPEAKER_00]: And the teacher can you know preside over all of that with with their sort of LLM assistance sitting with each person staying focused on their particular moment in time each of these LLMs can be present for the student at that moment in time which is.

[00:13:11] [SPEAKER_00]: Very important and one of the toughest things for a teacher who's got a class of thirty thirty five people and can't give individual attention to each one of them.

[00:13:21] [SPEAKER_01]: That's I said at the very beginning of these episode it has been twelve months since we last spoke but one of the other things that sat off my tech spied his senses was when I was doing a little research.

[00:13:30] [SPEAKER_01]: Quickly learn that memorise is recently ranked number two on times top-ed tech companies of twenty twenty four so a belated congrats for me on that but.

[00:13:41] [SPEAKER_01]: What is it that you think sex memorise a par in the competitive landscape of educational technology and how was that news been received there's set off your communication like a vague a slot machine was happened since that.

[00:13:55] [SPEAKER_00]: Yeah it actually has been quite a nice recognition that we received and it has resulted in lots of inbound both from new users signing up to you know people in the ed tech space given it's a call asking what it is we're doing so successfully with an artificial intelligence for the purpose of teaching people a language.

[00:14:19] [SPEAKER_00]: I think we wound up at the very very top of that list you know number two in the world and number one in Europe for sure.

[00:14:30] [SPEAKER_00]: Because of how deeply and how completely we've embraced technology to truly give people a personalized experience.

[00:14:41] [SPEAKER_00]: To truly give people their own dictionary that they can run into all kinds of features whether their grammar features or pronunciation features or you know viewing video features or chatbot features all of these things can happen so I think that's how we wound up on that list.

[00:15:01] [SPEAKER_00]: And it really has been quite a boon for us so we are grateful that whatever mathematical equation they ran that gave us the score that we got.

[00:15:13] [SPEAKER_00]: We're grateful we'll be recognized for it.

[00:15:15] [SPEAKER_01]: I was really excited about this. You don't just use a. It's memory techniques, noro science and combined with a.

[00:15:23] [SPEAKER_01]: So question of God, I was confused. I have all this fit together. I have to say a.

[00:15:28] [SPEAKER_01]: And Hanslow's methods of memory techniques, noro science and your entire approach to language learning because it feels like there's almost a secret recipe there of how will these things come together and converge.

[00:15:41] [SPEAKER_00]: Yeah, well so maybe I can answer that best with a little bit of an example.

[00:15:47] [SPEAKER_00]: You refer to space repetition earlier right and that's because we know from you know work that noro scientists have done and teachers have done and you know all of us have done and learning things that we've learned in school.

[00:16:04] [SPEAKER_00]: That you need to do some level of space repetition to make something stick.

[00:16:09] [SPEAKER_00]: But we don't actually know that the space repetition in time is the same for all things for all words, for all concepts and so on.

[00:16:21] [SPEAKER_00]: So that space repetition that we once had to just program in a soon like hey well gonna give you a.

[00:16:28] [SPEAKER_00]: We're going to ask you a question if you get it right we want to ask you for a day if you get that one right we want to ask you for a week if you get that one right we want to ask you for two weeks where we just form you lay a complete program it.

[00:16:38] [SPEAKER_00]: Now we can let AI start to say well we'll just start to run you into all kinds of experiences.

[00:16:47] [SPEAKER_00]: And when we notice that one of those experiences don't go well doesn't go well because you might not know the meaning of a word.

[00:16:57] [SPEAKER_00]: Well then we can start to send you back to do a little quiz about what that word means.

[00:17:03] [SPEAKER_00]: So space repetition can be come something less formulaic and more a result of the input of the user into the system.

[00:17:18] [SPEAKER_00]: And I think that allows things to be much more useful so it's the same principle of space repetition you got to touch it every once in a while for it to cement there we can use that same principle but we don't have to.

[00:17:33] [SPEAKER_00]: And you know, code it as just a formula that's the same for everybody if that makes sense really tells I'm a interest at where all this is heading how it evolves etc.

[00:17:44] [SPEAKER_01]: So if we appreciate you can't it's almost impossible to predict the future of a minute with the pace of technological change but how do you envision AI to shape the future of education and also for any educators that might be listening to us still maybe sat on the fence or struggled to see where it fits into what they do.

[00:18:02] [SPEAKER_01]: What steps do you think educators and institutions should be taking right now to maybe integrate AI effectively if not now but in the months and years ahead.

[00:18:12] [SPEAKER_00]: Yeah I've got I've actually got a lot a number of answers to that question but first and foremost it is absolutely positively crystal clear that these technologies are going to be a thing for ever more going forward they're going to get better and better and better.

[00:18:31] [SPEAKER_00]: And being well versed in what these technologies can do is absolutely essential for anybody and everybody in the world let alone teachers who are teaching people how to learn the things necessary to navigate the world.

[00:18:51] [SPEAKER_00]: So that's number one but number two is this technology if understood even a bit can make teachers lives easier.

[00:19:06] [SPEAKER_00]: So even if there's not the you know shared belief that this is going to be an important thing in the future.

[00:19:15] [SPEAKER_00]: There is a world where you can see that your own life is made easier because you can consider it these technological assistance to help you do the job that you need to do more effectively with less friction

[00:19:36] [SPEAKER_00]: and more ease and bring more joy to the job.

[00:19:40] [SPEAKER_00]: Get a better result for what you're trying to accomplish so I think that you know to answer the question of where is this going in the future.

[00:19:51] [SPEAKER_00]: I think it's going in a direction that it becomes a seamless assistant in the space of education that can very quickly sort of replicate out a given teacher to help a large group of folks with more attention to detail and personalization and less friction and work in effort for the teacher.

[00:20:21] [SPEAKER_01]: Well it's been an absolute pleasure having you back on the podcast we talked a bit about your origin story in our last episode where you've gone from everything from eventually to EY to founder at just a bi better and obviously CEO and memorise and I suspect if you look back at that career you have more than a few stories that you can share maybe some of you can share but is already funny or interesting story has happened to you in your career to to leave on a light heart moment.

[00:20:49] [SPEAKER_00]: I think there's one interesting funny one it's funny with hindsight but I remember when I was working at IAC running a company called Apple on that was originally a company founded in mincebellarouse and I was running that company in spitting a lot of time in mince but at the moment of this story I was in New York and the main person running the operation in mincebellarouse gave me a call.

[00:21:19] [SPEAKER_00]: With a sense that I never really thought I'd ever hear ever in my professional career and it was this Steve the KGB just left the office.

[00:21:29] [SPEAKER_00]: That in hindsight is quite humorous but at the time it's an interesting category.

[00:21:40] [SPEAKER_00]: That would have been 2019.