This week on the Business of Tech powered by two Degrees Business. Our education tech companies are going gangbusters, but as artificial intelligence a massive disrupted coming for them, or the technology that will make them even more successful and valuable.
We talked to CAMI co founder Henji Wang following a major new investment in the company that values it at three hundred million dollars, flush with cash and already forty million students and teachers using its online tools. How will AI factor into the future of education and Cami's business.
We are certainly at the tip of the iceberg, if you will around what's possible in this technology. There's a lot more that you can do to purposefully take AI, take genitor of AI, make curriculum more accommodating to the students' needs. I think that's the dream, is to be able to personalize it for every single student. But doing our way that is purposeful.
Now. This is the first in a series we are doing on how AI is impacting various industries and the opportunities for Kiwi businesses and entrepreneurs that AI represents. We'll be looking at healthcare and manufacturing in upcoming episodes. But given Ben's a former school teacher. We wanted to start somewhere close to his heart.
Yeah.
I love education and I really enjoy talking to Henji from CAMI about where it's going, and so my interview with him will be coming up shortly.
First, let's take a look at the big dump of product related news out of Apple this week, with the launch of the iPhone sixteen lineup, a few interesting health tech editions, and the debut, of course, of Apple Intelligence. What did you make of it, Ben, Tuesday's big announcement.
Yeah, I think it was the definition of iterative in many ways, but I think some of those iterations were pretty cool. I know the online reaction that I saw was pretty underwhelmed in general, but I thought there was some interesting stuff in there. The little camera button that turns the phone into a more of a focused camera tool. I think like while most people might not find that incredibly useful, there's going to be a good chunk of
people who do. Might open up more of the camera's options to more people, And it really is working to turn it into an industry tool, which I think is super interesting. And if you look back earlier this year, I went and jumped off the sky tower because Apple had done a deal with aj Hacket Bungee to use iPhone fifteen's in all of their replacing their cameras to
capture customer experiences. So people are starting to see the value of using the iPhones in place of some of these more expensive camera rigs, which I think is an interesting trend that probably shouldn't be overlooked and isn't something that I think most of the consumers maybe watching that glow Time event would have noticed, but really stuck out me.
Yeah, I mean people are shooting entire movies on the iPhone now, particularly the iPhone Pro and Pro Max or whatever it's called, So the quality level is there, so to have that tactile to be able to really control that is really useful. But the key sort of takeaway for me actually is the pricing's actually decreased for the New Zealand market, and that's stayed the same in the US.
So the iPhone sixteen will go on sale here for one five hundred and ninety nine dollars, and that is down from one thousand, six hundred and forty nine dollars for the iPhone fifteen, and that discount or lower price is reflected across the range the sixteen pro, for instance, is nineteen hundred and ninety nine dollars that used to be two thousand and ninety nine. They're looking at the New Zealand dollar compared to the US, going how much
can keiws take really in a recessionary environment. So that is good for consumers, But for me, the other big thing that caught my attention is really not on the iPhone. It's on the Apple Watch and on the AirPod Pro.
These health features. I mean, it's super exciting turning the AirPod pros and I've been talking about this for a long time, but they finally nailed it, turning it basically into what they're calling a clinical grade hearing aid, so not necessarily replacing hearing aids that are usually very expensive and finely tuned to the user. But if you have sort of low or mid level hearing loss, being able to use these to basically boost the volume, the clarity
and adapted very much to you. You can tune them using the app on the iPhone.
It's pretty incredible, and you can do the hearing test and the Apple Health app as well. So it's like a beginning to end solution for people with like you say, mild to moderate hearing loss like that is probably something that I don't think many people would have had on their Bengo cards, you know. But it really just goes to show the potential for medtech in people's lives.
The only thing that does give me pause.
Is that the cost of these devices and the fact that they are still very much in the high end premium, does start to see a little bit of a diversion or further diversion in healthcare for different ends of the social spectrum. Absolutely, you know, it's not going to be maybe the people that need it most who are getting these things strapped to their wrist, perhaps, Yeah.
And if you need a device set runs iOS eighteen for the hearing aids, you need the AirPod Pro two. I think it is it's four hundred and seventy nine dollars that they're going on, the mircrop in New Zealand, a high end Apple Watch I think nine or ten series, and the Ultra, so you're looking at five or six hundred bucks for one of those. But they are the not subscription based, which is good, so credit to them for innovating these things. Neither of them have had FDA approval.
Or the right regulatory sign off. Yet they say that's coming soon, so hopefully they get that over the line. It's curious that they launched these things before they've actually got it all signed and sealed, but.
Pretty confident about it. Yeah, exactly that they will.
I can't see it being too controversial. But I guess the other thing, which is also not here yet the phones have debuted, but Apple Intelligence.
Yeah, and that was my next point as well.
It's just like, why would you why would you do that? You've got this big flagship event and it's like, oh, that will be later this month. These features, these Apple Intelligence AI driven text editing and in box summaries all sounds great and we've been hearing about it for months. Do it in a big bang when the actual phones are unveiled at your biggest of end of the year.
Yeah, I mean it just I have to.
Like, it does get me really excited, the contextual personal context stuff, being able to go through and have that really smart assistant and all the upgrades to Siri. It gets me really excited for that. But at the same time, it's it's kind of like how many times can.
You can you get away with that?
Like I feel like I'm being I've been burned with the Surface laptop and this like burned a little bit with the originally with Samsung when they came out saying they were going to have all these great AI features and they were a bit milk toast and and so come on, Apple, like, my hopes are up, so really need you to deliver on this one.
Yeah, yeah, you're a business desk. A review of the Microsoft Surface Laptop was was pretty pretty scathing, and really that was down to the over promising on the AI function. I was the same I reviewed it, and you know, there's that button on keyboard which is your launch pad into co pilot, and co Pilot's a good a good service, but really that is it. All the other functions that they were really touting, like the recall function wasn't available. So I think Apple is in real danger of doing
the same thing. They obviously are making last minute tweaks. It's very complicated to get these things working properly. But if we don't see them having a really good impact by the end of this month, I think they're going to be in trouble and that may impact iPhone sixteen sales ultimately.
Yeah, absolutely, unless the people, the early adopters get these in their hands, and even the iPhone fifteen Pro people get the Apple Intelligence in their hands and go this is awesome and really generate that buzz online, it is not going to have that pick up impact that they're really hoping for for the iPhone sixteen because they're putting a lot of eggs in this iPhone sixteen basket in terms of their you know, dropping sales over recent years.
So yeah, and look, they're not really promising that much. It's pretty basic, you know what you'll be able to do in the initial run anyway. But what will be interesting is all the developers who got access to Apple Intelligence six months ago, what they will be coming up with and what will appear on the App Store or in their existing apps that draw on that neural processor in the iPhone sixteen and which you can use on the iPhone fifteen Pro as well. That is going to
supercharge the apps that we use every day. So that's when the real value I think will come, both for the surface and these co pilot PCs as well as phones that have AI on them, when all the third party app developers jump on board and make use of them.
Interesting that Apple had its big splash and literally the next day on Wednesday, news from Europe that they've been slapped with a twenty three billion dollar fee for back taxes because of their arrangement with Ireland where they were paying for a long time a very low effective tax rate. And you have this weird situation where islands have got
rich off companies like Apple. We had a guest on a few months back who was talking about the Celtic Tiger and how rich Dublin is because of the tech sector, and you had Ireland basically saying to the EU, leave us alone. We've got this great deal going with Apple, and the EU said, no, Ireland, you were part of the EU. We will all have uniform tax policies. Apple's going to have to pay. So they will have to pay.
They've said, I think in the fourth quarter they will have a ten billion dollar US impact on their books to settle that. So they'll have to sell a lot more iPhone sixteen's to make up for it.
Absolutely, yeah, yeah, and I mean it goes. It does speak volumes of the fact that they have capitulated to a lot of the EU's demands in terms of USBC in terms of opening the NMC chip and all these other things. The Europe is clearly a very important market to them. So yeah, I can't see them. Well, there's probably a little bit of grumbling in the background. I can't see them, you know, resisting too hard at paying their their fair share over in the EU.
No, and very you know, suddenly the Irish government has a twenty three billion dollar that's New Zealand dollar, sort of thirteen billion euro windfall.
Imagine that.
Imagine what that would mean for the New Zealand government to have that sort of money to put into infrastructure or something. So you're trying to figure out what to do with it. They want to set up a sovereign fun there, one hundred billion dollars sovereign fun And that's really because they're taking in so much corporate tax. Now, they got all of those companies there at paying low tax, and then the EU said no, they have to pay
a fair amount of tax. And the companies have invested so much in Ireland are staying there.
So it's just worked.
It's just a masterful plan really.
By the I fantastic.
Yeah, it's a pity we's so far away that it wouldn't quite work for us.
But no, right now, onto our featured interview with Henji Wang, co founder of CAMI. Last month, The Herald reported that BV Investment Partners had taken a controlling stake in CAMI. This is a Boston based private equity company that's been around for decades and invested in a lot of ed tech companies. That deal valued the company at three hundred million dollars.
Now.
It follows another US private equity company taking a big stake in Dunedin's Education Perfect back in twenty twenty one that valued the company at four hundred and fifty five million dollars. We've also got Crimson Education, which raised ten million US last year and a funding round that valued it it supposedly US four hundred and sixty million dollars.
Those are incredible valuations for education related startups and they're doing very well overseas.
That's just three of them.
There are several other fast growing ones, and it's that that that doesn't really get the same level of bars as like fintech or agritech, but which New Zealand seems to do very well in.
I'll tell you who has done very well from Boston Ventures buying CAMI. The New Zealand Growth Capital Partners, the government's venture investment vehicle. It's a spire fund will get thirty seven million dollars back from the sale of CAMI. That represents a seventyfold increase on its investment. But Ben tell us a bit more about exactly what CAMI does in the education tech space.
Basically, at its core, CAMI is a document sharing platform, so teachers can easily share PDFs with students and they can be annotated and they can be easily marked if they know the kids write their answers on the sheets and things like that.
I mean, that's that Its very core.
But it has evolved now into this really comprehensive assessment and learning product. So you can take a pack of resources and you can give them to students and you can create assessment it's based on them, and you know, it's got remote learning aspects to it. And it's basically like a document management tool on steroids specifically for education
for teachers and students. And obviously in the digital age, when all of your resources are online on your computer and they're going to different types of computers to chromebooks, to PCs, iPads. Being able to have something that can work across all of those platforms is really useful for education worldwide.
Yeah, and I've had a huge cut through in one hundred and eighty countries, big in the US. It's been the focus of them. Founded by students so very much, people who are feeling the pain points of learning in the digital age and decided to go about and fix that. So here's been talking to CAMI co Founderhnji Wang about a startup journey and the role AI is likely to play and the type of educational platforms Cammi is developing.
Hi, Henji, Welcome to the Business of Tech podcast. Thanks so much for joining us. Thanks for having me, Ben So, you are one of the co founders of a New Zealand's success story in the tech scene and especially the ed tech scene known as Cammi, and Cammy's been around for just over a decade now. I think you just said to me basically selling a really powerful and useful piece of software into schools. Do you want to give us just a very brief background about how Cami came to be.
Absolutely so. Cammy is this wonderful platform app that's used by over forty million teachers and students around the world. But before we got there, we started this business as just an app that we were as best friends at university. We wanted to solve a problem that we had ourselves, which was how do you effectively take study notes in class?
And this is over decadedgos So when devices were just starting to show up in every classroom, when technologists just started to show up more privalently in the class, prominently in the class brother and we knew that we had a look and there wasn't great software and we needed to do something about it. So that's the solution that we came up with. And then of course this trend
around teacher time and resource availability. Teachers are increasingly cling more and how do you have a way to still be able to personalize and make learning accessible and make it exciting, but do it in a way that uses technology in the most purposeful way possible and embracing that I think has always been the biggest challenge in education.
I think that kind of leads us quite naturally to the main topic of the conversation that we are here to have, which is, of course artificial intelligence and how AI is increasingly playing a role in the classroom, and I think probably the first thing we should do is split out AI as we're thinking about it, into what
we might call now classic AI. And then we've got this other public facing side of AI, the new wave generative AI, LM based AI, whatever you'd like to call it, that involves interactivity between dynamic language based interactivity between kind of the user and the software. So, yeah, let's go back to the early days of how Chemi was using AI. I'm assuming you've had some kind of algorithmic presence in your product for several years.
Yeah, So if we were to zoom back a little bit and just think about that implementation of technology in the classroom, Cammy, one of our proof points is we save teachers. This is teachers telling us we save teachers in the average of eight hours a week. Wow, So that is huge. That's over an hour each day that we're giving back to the teacher to enable them to focus back on their students, their learning outcomes, making things
more accessible. And if you look at how we arrived at that point, we did it in a way that takes technology that already exists today and put it together in a way that really drives activity in the classroom. We do it by, for instance, saving a clique for the teacher or saving a clique for the students so that they don't have to think about it and worry about it. So anything that we can do to simplify technology in the classroom is actually a huge gain for
the teacher. Just as an example, when we did all of this, we achieved this eight hour a week on average saving for the teacher outcome before we had any of the new generation of LAMAI implemented an product. So that tells you something about how we as a company know how to implement technology in the classroom, and we've done it using your traditional AI. There is a lot of One example would be how we take an image
that teachers have and digitizing that image. So take a scan of a document and we make it as seamless as possible to digitize it into a format that you can start using. You can start interacting as a student. So I, as a teacher, take a scan of a piece of curriculum, I send it off to a student that might have special accommodations or special needs, and they can start immediately using it with their speech to text.
So it allows it to read out loud the text or sorry rather text a speech not speech to text. On the opposite end, Cammie also has tools that allow the student to speak their answer back to the to their curriculum and record all whether that's a video, whether that's sort of a text written dictated down. And that's perhaps what you would consider your more traditional AI implementation of CAME, but we do in a way that is
as little friction as possible. And so yeah, that would probably be a few examples of how we've gone on to purposely implement this in a way. And for the more tech savvy readers out there, it's just a very seamless integration of what is known as the OCI technology solution.
Right, So it's just kind of any time that you see that there's a way to improve a process basically to make it less manual using these the implementation of these algorithms, you can kind of do that because you are a software product, and speech to text must have been kind of revolutionary for a lot of people when that was first able to be rolled out on mass Having that same kind of experience with the new generation the LM based AI.
In short, yes, I think one important thing I should note before we get into that is whenever we develop product, our philosophy is to speak to the customer directly. And if you look at every single feature, every single change that we've rolled out in our product, we can tie that to a conversation we've had with a feature with a parent, with an administrator, perhaps sometimes with a student, and so that is extremely important to us being able to link it back to this will help this person
and why. And when you look at how we've started to implement LLMS, we've done it in a way that is extremely purposeful. So our first enhancement using LLMS and Generator of AI is to take an existing piece of curriculum and you enhance it by bringing out or teasing out all the sort of quizzes and questions that typically a teacher would spend time typing manually out for the
students to answer. We now pull that out. If there's an answer key, we'll use take advantage of that answer key, and it pulls it all out so that the teacher can then organize it on their document, on their template, rather before they send it out to the students to answer. And one of the things that we've done is we can also select using l and MS the right answer, so it saves teachers time. They don't have to say
they don't have to grade it manually anymore. They can simply let it auto grade because at the time that we generate these quizzes or these questions, we've already got an idea of what the answer will be, and we have the teacher check all of this before they send it off to the students, which is an important bit around the limitations of these LMS.
So you're automating. So when you say a curriculum, that's what talking about is like a stack of resources, right, So a bunch of PDFs or word documents or that teachers gather over years to create essentially a pack of
resources that they use to teach a particular unit. So for example, if it's a history class, there might be a French Revolution unit, and so they'll have a whole bunch of documents, scanned pages from books, website resources they printed out, and then stuff they've made themselves, quizzes, like you said, all these kinds of things, and with the LLM tools, because of that contextual understanding, you're able to go in and take out just the quizzes and say, okay,
here you go. Here's the ten different quizzes that you've got in this curriculum. You can kind of use them how you want to now, and here's the answer key. And then on top of that, you said there was a kind of automatic marking aspect to it as well.
Yeah, so as it calls out these questions from the resource, it can also use its knowledge base in the LLM to select what is the right answer for you. But nine times out of ten that resource already has an answer key be able to leverage that or the answer to that is already from the curriculum that you're teaching.
So you can simply go and we'll tease that out and auto allow you as a teacher to auto grape, so that when you send it off to the students, you have the option to say, great, it automatically for me, But don't show the student the answer or show the student the answer so that they can learn from that and perhaps learn from why they got the answer wrong, because there will be a hint field inside a camy that allows the student to be able to see that
hint and understand, oh, I got it wrong because of this or this is the hint towards the right answer that I can try and reread my resource. So it's very flexible, and it's designed specifically because checking for understanding as a teacher is so important that we want to be able to provide the teacher with all the tools that need to check that sort of that what they're trying to teach is actually being comprehended by the student.
That's that's really interesting. I guess that's the one of the strengths of these generative AI models is that they are really good at taking unstructured data and unearthing specific things that you want from it, and you know you can make these requests. So what is what are the what are the possibilities you mentioned character based AI? So like a Marie Antoinette comes on and as a as an AI and perhaps the students can talk to her. That seems like an interesting use case.
Is that is that.
Something that you've had chats with about your with your customers about.
Out there in the wild. We have seen a lot a lot of different platforms that do that. We have seen some success in the classroom where a teacher might take a particular resource and essentially generate artifacts of different reason side resources supplemental resources from that core curriculum resource. These generated resources are done in a way that is
more exciting and engaging for the students. So an example would be hate this particular piece of core curriculum and generator in the voice of Lightning the Queen the movie Cars and catch out exactly, and you can have that sort of personality sort of explain the content back to you.
It reminds me of that meme of a parent who said that the kids wouldn't eat frozen vegetables, so they stuck some poor patrol stickers on and said they were poor patrol frozen vegetables. Like, is that, you know, figuring out a way to meet kids where they where.
They are, where they're interested exactly. And I think if you can do that, you can you've If you can get them interested in the curriculum and the content and the way that you know you're able to meet them in the middle, that's transformative. And I think ultimately those are some of the great successes that we don't often hear about in the class out of the classroom rather, but that's something that is you know, really helping the
sort of engagement and interactivity of that material in the classroom. Yeah, it is.
It is interesting, like we have this sense that for education to be valuable, it has to be kind of formal or it has to be dry, right, But you know, the reality is if you go to the best teachers classrooms, they're always covered in colors and posters on the wall, and you know, trying to try to walk that line between meeting kids where they are with interest and not being cringed, which is always a tough one to walk.
But the idea that you can use these kind of artificial intelligence tools to enhance that because not every teacher is going to be hit with the kids. So if you can kind of say, like, I want to take this resource and what are some ways that I could make it relevant for an audience of eight to nine year olds, what are some of the things they're interested in, and how can I bring these together?
That seems like a really powerful use case.
Absolutely, and I think this is why long term, the teacher's role in the classroom remains the same, which is to guide the students and help them along the learning journey. Learning journey, and AI is simply here to facilitate, to support,
but never to replace. You always need a teacher to be that initial seed of I think creativity in the classroom of how do you do this but in a way that is intentional and purposeful around all the learning theory that is out there, all the scientific evidence that is out there, around how do you best have good
useful curriculum and the implementation of that curriculum. So I don't think the role of the teacher is going in a way in times certain AI is simply there to help facilitate and really give drive meaningful productivity gains in the teacher workflow.
Are you sure that we will not have robot teachers coming in the future where kids all put on their VR headsets and they go to the virtual classroom. I mean, that's obviously an exaggeration, But what about the possibility of virtual tas right teachers assistants who can provide that kind of tier one support for kids where the teacher. Like you said, class numbers are growing, Teacher time is really hard.
Quite often a teacher in a classroom of thirty maybe five students will take up the majority of their time. So is there a way of providing an AI support for teachers to be able to actually help students with those first level questions before they go to the teacher. Is that desirable or is that kind of using technology
to justify having worse teacher to student ratios? You know, that's a big question to ask, But I'm just thinking about, like, as we move into the future of education where it doesn't look like we're going to be exploding numbers of teachers while kid numbers are going to go up, like, how do we is there a way that we can help to mitigate that?
That's an interesting idea. I haven't given it too much thought, but I think as a general thumb, if we're able to help the teacher, you know, do their job job more effectively, AKAA have a tutor that is able to advide let's say, first line support to any sort of questions about the correculum, then yeah, that could be a useful feature of a platform, Like do you do.
You see there being a space for And I'm not saying that Cammie's going to do it right now, but I'm just kind of curious whether you think that that is a future that we're headed towards, or is that that's just so far in the future it's not even worth thinking about having actual chatbots that can help students in the classroom.
I think there are already companies that are trying to do that today. I think there haven't hit mainstream yet. You know, I've cluely seen quite a few of them.
It doesn't take a lot to implement something like this, But the real question is like, how do you do this with the right amount of controls and visibility and transparency and safety so that it's actually easy for your non icky teacher that doesn't have a lot of time to go out and train about you know, all the biases and limitations of any I and understand that fully before they roll it out in the classroom. How do
you do how do you do that well? I haven't seen that done yet, but I can see a future of five teen years from now where that's a possibility, and I can see how that's going to be super beneficial, I think for the classroom. But you've got to do it right, and I think that's the bet that I haven't seen done well yet.
You mentioned personalization as well, using artificial intelligence to be able to contextually understand the way that a student answer different prompts, you know, using prompt might be the wrong word, but be able to answer different questions in class like can look at their writing, how they answer exam questions, how they answer short answer, how they approach solving math questions, and learn about the best way to help to guide them.
Can be fed information about their learning style, about their you know, any learning support they might need, and can help to amend curricula in a way that will be
supportive of students. We talked about it in terms of you know, lightning McQueen voices and things like that, but is there is there potential for it to actually more specifically be able to address personal needs or help teachers to address students' personal learning needs throughout the development of a curriculum plan for a classroom.
Yeah, there's a lot there. I think there is a lot of opportunity to realize with AI that we haven't realized yet. And a lot of it comes down to how do you personalize this curriculum to the interests of the child, to what they've mentioned or said in the past that might be of interest to them, and then how do you do it in a way that still provides an instructural value based on all the learning sciences.
So you know, we are certainly at this sort of tip of the iceberg, if you will, around just scratching
the surface of what's possible in this technology. There's a lot more that you can do to you know, purposefully, take AI, take genet of AI, do it in a way that can make curriculum more interacting, interact of engaging, accessible, more accommodating to the students' needs, and do it in a way that means that you don't necessarily need a special needs accommodation or special needs sort of special needs
flag in this student system. To be able to have that support offered to you, I think that's a dream, is to be able to personalize it for every single student. But doing in a way that is purposeful. I you know, I con certain I see all those sort of opportunities out there, and I think that's what's really exciting. Over the next five ten years of just working in this space, you know, we're able to realize a lot of things that we've been thinking about and dreaming about for such a long time.
What is the things that you're hearing most concern I mean when chat GPT first came out, you know, it was all about plagiarism, kids just asking chat GPT to write essays for them, and then there's the battle of the AI detectors and is it written by AI.
Or YadA YadA.
Is that still the biggest concern that teachers are having or is there something else that's coming along in the in the realm of AI and education that's starting to displace that.
Yeah, that's a great question. I think there are certainly there's certainly a huge market of AI detectors, but if you actually if you look at the efficacy of those detectors, it's actually very low. I think one of the dangerous things to do in education is to say to a child, we're forty three percent sure that you've laid rised as text, and even if it gets to eighty or ninety percent, you never want to have ten percent of doubt in there to then go on to you know, go through
a disimilary process for the student. So unless you're one hundred percent certain you I don't think I personally don't think you should go through a process like that. So what's important to educate as in my opinion is and we've implemented this throughout throughout CAEMI, is giving the teacher of visibility of how this work was evolved to where
it is today. And if we can show the teacher, hey, the student spent this amount of time on CAMI, and they edited this piece of work a few different times and we went through a few different iterations of answering this, then you can see that the proof of work is there. And I think that's the bit that's important, that visibility of how the student came about to this answer, And ultimately my hope is that everyone's able to get to
that point throughout every single product. But yeah, this is why I think it's so important to you know, when you're implementing technology, you're not just jumping at the next API that's available by chat reput but actually do it in a way where you're thinking about how does what is the impact of this in the classroom, and how do I actually help the teacher. You know that famous quote people don't know what they want until you show
it to them, right by Steve Jobs. And I think that's the thesis that we've always come tried to try to have at the core of CAMI.
You can see that working really well. You know in primary, intermediate high school where you can get kids to use a specific software where you can kind of hover over them as they do it by pen and paper if.
You really need to.
When you get to university, the conversation changes though, because you're sending kids away to go and do work. And then back in my day you're printed it out and you're stuck a cover paper and put it in a slot. But these days it's a bit more advanced. You hand
in a word document. Are you expecting that we'll in that realm start to see that you might need to submit something that will give you a record of changes throughout and then can you use like a souped up air detection to look back at those changes and say, oh, okay, actually we can recognize that this work has been done legitimately and give it a green TECH or a red cross saying it was just copy and pasted from Wikipedia or a chat GPT.
Yeah, I think you're onto something. I think it's important to have some sort of way to have the teacher review how you came about, you know, arriving at this answer, and if it means you know, being able to have some sort of supplementary material that comes along it alongside it to say Hey, I visited these ten websites and this is how I arrived at this answer, And those are my references, digital fingerprints, if you will, that's attached
to my work. I think that's the sort of thing that ultimately, I think everyone should go towards, you know, and if you take a step back and think about, well, what is that in twenty years ago? Yeah, it was a videography, was us putting references next to our to support our answers. That was showing that you've done that research. And ultimately, fundamentally that hasn't changed. And today's world, Yeah.
It's very interesting, isn't it.
Because you can also go to a product like Perplexity and you can say, give me a quote about the French revolutions or you know, the impact of the French Revolution on the European economy and attribute, you know, find a good source for it. And yeah, I mean it's still a bit of a way away from just being able to then say and make a whole.
Essay out of it.
So that augmentative side of AI seems to becoming more prevalent in all all areas rather than the replacement side
of AI. Everyone was so concerned about is that a trend that you're seeing as well, that you kind of talked about it a little bit, but that teachers are starting to say, oh, I see how AI can augment students to do research, How AI can augment students to give them a base line of writing that they can then you know, edit and evolve into something that is that is their own work, rather than the fear of well, it's just gonna kids are going to lose all their skills of being able to do anything.
Yeah, I certainly empathize with that point of view. I think, you know, some people might disagree with this statement, but AI is sort of having its calculator calculator moment inside of education where once upon a time, and I think we were all required to know how to do basic arithmetic. And you can sider of see in some education systems that requirements go on simply because everyone can pull out
their phones or a calculator and do that arithmetic. So they've moved on their curriculum to perhaps things that are more creative or interesting about math and stats, where it frees you from having to learn arithmetic to you know, you still have to have an understanding of math, right and numbers and have a sick feel of how these numbers play together. But to actually get the right precise measurement or calculation, you don't necessarily need to know that
you can punch it into a calculator. And I suspect, you know, if you think for five ten years, if we can give everyone that calculator but for writing, and everyone can write great. But it's then moved on to how do you formulate ideas and integrate it into an argument about something about your thoughts or perhaps a point
that you're trying to make. Then that could be really interesting, right because it frees you up again from writer's block, from all those things that you know are spelling and perhaps you can I'm just sort of spitballing here, but then you can really focus on the things that the crux of you know, what it makes writing great, and perhaps we can generate a whole generation of great writers. And you know, I don't know. I think there is the world's are oyster here and we can do a
lot with this. But I think to take that baby step, you have to really think about how do you bear some poment ai in the classroom and do it in a way that's purposeful and try not to experiment with these ideas because you ultimately it's these kids future at risk, right, It's that's what's at stake here. So you can't just play fast sinelos like what we've done with AI and
the corporate world. I think you have to be really careful about you know, what are the safety guard rails, back to my original points and doing them the right way. So I think there's a lot of opportunity in front of us. We just have to spend the time to do it right and realize all the all the sort of possibilities.
What do you think, Peter, do you think we're going to be revolutionizing education with AI in the coming years.
Well, Henji's very cautious there, and we've seen this a lot. You know, education when Generative AI really came out of the starting gate, that was the one industry where everyone had a lot of angst about and it was all about plagiarism. And it's interesting his comments there around you know, to what extent do you sort of start persecuting students around plagiarism? Are they forty percent plagiaristic or one hundred percent?
You know, there's consequences for starting to make those claims against the students, and these are not perfect, these plagiarism checkers. So that's where it started, and that seems to have died a death really over the last six months. I think there are genuine concerns about plagiarism, but it's sort of flipped around now where I think a lot of administrators in tertiary and secondary education in particular are looking
at it and going, actually, we're stretched for resources. Teachers have high teacher to student ratios when it comes to the amount of time they can give a student. If these tools can actually help us and shave time off, like he was saying, even the traditional AI stuff was saving in some cases up to eight hours a week of admin for teachers. So if the next step of that is reducing that admin burden and improving the experience
for teachers, fantastic. It is about augmentation of the teacher's role rather than replacing aspects of what they do.
Yeah, it's amazing, like as a former teacher to think that, you know, having eight hours a week back would have been just incredible. And I think there is still a huge amount of potential for AI in other areas as well, like for example, report writing, Like I'm sure parents all love to think of their teachers sitting down and spending twenty minutes contemplating every student in their class and crafting the perfect report, but realistically, we just didn't have time
for that kind of stuff. It was very much a cut and based exercise where you consider based on results, based on strengths, based on weaknesses, and then you just kind of put together as fast as you can, but in a way that and then you add like a personal comment, you know, for each one, and if you can use AI to kind of look at the student's history and then prepopulate something, and then add you know, a personal comment on top of that, which is basically
what was being done already. That's a huge amount of administrative work. I know that's not really exactly in Cami's wheelhouse, but those are the kinds of areas that I think, you know, could really have a lot of impact.
You were talking to him about the generative AI stuff, in particular automated grading, which which is great as long as you said that you still have those personal comments. You know what I sort of feared but sort of loved as well. When I was a kid at school. Was the comments in the margins from my teachers, all the red lines and the underscoring and the crossing out and the exclamation marks, that sort of stuff that showed that a teacher had read my essay and was passionate
about me understanding how it could be improved. So making sure that that human touch is still there. Even if a lot of the basic grading is automated.
Yeah, it'll be much harder on like long form essays, but even then you could get some basic stuff to be like, well, this looks like it's probably going to be in this range. Here are some areas, some sentences you might want to look at. Here are some issues that you might want to address with the student around spelling or grammar or sentence construction or word use choice,
and those kinds of things can be automated. The general feel of an argument is always going to have to be a teacher, I think, because although you know AI could potentially take some of that load, you probably don't want it to because the margin of even if it's a two percent margin of error, that's still too big
when you're dealing with education. I also really like this idea of having like a little personal assistant, a little you know, AI assistant that knows a lot about your topic that students can access to ask first tier questions like that, first tier support using chatbots for that.
Like, yeah, I could see the barriers to entry for that to be really really low. I mean, if you feed in a database of all your curriculum notes and that, I think, you know, a co pilot could do quite a good job of that now in terms of just giving you answers to history questions or geography or whatever.
But I think that the real breakthrough, and no one really is talking too deeply in education about this year because it is difficult to do well, is a genuine co pilot for the student that goes on them throughout their entire learning process, potentially over a decade or longer. Where for instance, I was terrible at maths at school. I really struggled, and I got coaching in maths. I hated going after school to do another hour of maths.
But if I had a copilot that was looking at my test results, looking where I was struggling, giving me suggestions and tuition that I could do in a nice, easy digital format, that would be really useful and so that involves probably a little bit more complexity. There's all the curriculum information, but also analyzing my performance and giving me suggestions to improve it. There's a whole pedagogy that needs to be developed in the AI driven world to
support all of that. But I think that's going to be the big game change of for students.
Yeah, and you know, kind of getting used to using these tools in a way that is supportive of how you do your work. I think that's going to be really important as these students move into the workplace. So these may not happen in primary school or even intermediate, but as they start to get into high school, having the ability to know when to ask a chatbot something and how to ask, and what the limitations are and
you know what their strengths are. Learning those through education I think is going to be super valuable as you then emerge into the working world and may have to rely on one to actually do your job. If you're not familiar with it, it may actually, you know, hold you back a little bit in the long run.
Yeah, So well done to Hinji Wang and team staying on at CAMI at least for the time being, as this big US company takes a controlling stake in it, so they seem to seem to be poised really for excellent growth, particularly in the US, which this Boston based investment company I think is particularly interested in the US market. So it seems like it's onwards and upwards for them, with a dose of AI thrown in.
Absolutely, congratulations came and all the founders there. That's our show for another week. Thank you very much to henji Wan for taking the time to talk about CAMI, AI and education.
Show notes on Cammi's big payday. Some background reading on AI and education and coverage of those Apple announcements are in the show notes. You'll find them in the Tech section of the Business Desk website, and.
The Business of Tech is of course on iHeartRadio and all major podcast platforms. Leave us a rating in your favorite app to help boost our exposure and get in touch with me with your feedback, ideas, topics, and guest suggestions.
Email me Ben.
At Businessdesk dot co, dot z or find us both on LinkedIn and sometimes x.
Another dose of the Business of Tech coming your way next Thursday.
Catch you then,