← Edtech Insiders
Edtech Insiders
Week in EdTech 6/10/26: Claude Fable Launches, NAEP Recovery, Code.org Becomes CodeAI, AI Techlash Grows, Canvas Breach, and More! Feat. Yi Ng of Curiosities AI & Dr. Rob Abel of EdTech Futures
01:35:34
Loading summary
Transcript123 lines
- [00:00]
A
This season of EdTech Insiders is brought to you by Cooley LLP. Cooley is the go to law firm for education and ed tech innovators offering industry informed counsel across the pre K to gray spectrum. With a multidisciplinary approach and a powerful edtech ecosystem, Cooley helps shape the future of education. This season of EdTech Insiders is brought to you by Starbridge. Every year, K12 districts and higher ed institutions spend over half a trillion dollars. But most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts and strategic plans and then helps you turn them into personalized outreach. Fast win the deal before it hits the RFP stage. That's how top EdTech teams stay ahead. This season of EdTech Insiders is brought to you by Cooley LLP. Cooley is the go to law firm for education and ed tech innovators offering industry informed counsel across the pre K to gray spectrum. With a multidisciplinary approach and a powerful edtech ecosystem, Cooley helps shape the future of education. This season of EdTech Insiders is brought to you by Starbridge. Every year, K12 districts and higher ed institutions spend over half a trillion dollars. But most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts and strategic plans and then helps you turn them into personalized outreach. Fast win the deal before it hits the RFP stage. That's how top edtech teams stay ahead.
- [01:46]
B
If you look back in the 1930s in the New York Times, they interviewed a headmaster from a school in Boston who was talking about any problematic pupil. You can trace it back to a radio being in their home, right?
- [02:00]
A
Exactly.
- [02:00]
B
And you can see these exact same things happen with TVs, with movie. We have this cycle of outrage that is just like every aspect that touches the creation of radio, like the making of movies, those are having a negative spillover effect. And I do think we're having a bit of a moment of that with AI that anything remotely touching AI is bad because it's AI and AI is bad. And we're going to be outraged about that for right now.
- [02:26]
A
Welcome to EdTech Insiders, the top podcast covering the education technology industry. From funding rounds to impact to AI developments across early childhood, K12 higher ed and work. You'll find it all here at EdTech Insiders. Remember to subscribe to the pod, check out our newsletter and also our event calendar. And to go deeper, check out EdTech EdTech Insiders plus, where you can get premium content access to our WhatsApp channel early access to events and back channel insights from Alex and Ben. Hope you enjoy today's pod. Welcome to The Week in EdTech from EdTech Insiders. We are here with a special guest host today, Thomas Rogers, head of K12 communications at Whiteboard and Advisors. Welcome, Thomas.
- [03:17]
B
Thank you for having me.
- [03:19]
A
Yeah, we're excited to talk to you today. We're here in the heat of summer is starting up. It's mid June and you know, it feels like Everybody's out of K12 school, universities are out, but like things are still sort of steamy. There's a lot of discussion happening, especially around the AI space. Just people trying to figure out exactly what to do over the next year when it comes to all of these hot button issues. AI in schools, AI in jobs, learning outcomes and scores. New NAEP reports that you were just calling out. We'll talk about that today. But I wanted to start with just a pure AI topic and we can keep this pretty quick before we get into some of the real ed tech pieces, which is we can't go too far without talking about Claude Mythos and Claude Fable. Basically. Anthropic's you know, Fable model came out this week. It's the version of Mythos that they held back because it was cybersecurity risk. But it is apparently, you know, a truly step change. It's actually being offered pretty widely to Anthropic users, but it can do even more incredible things. Full projects, 10 hour coding projects, make video games, do all sorts of things. And it feels like it's a yet another big step towards incredible AI capabilities. What did you make of the Mythos or the Fable launch this week, Thomas?
- [04:32]
B
Have you used it yet?
- [04:34]
A
No, I haven't.
- [04:35]
B
So I have. I tried it out earlier this week and just like a very simple task. I actually did it without it and then did the task over again. It was just, you know, writing a briefing doc for an interview. And what I thought was really interesting is that when the Fable version of the document, it was absent a lot of the what people say are like the AI tels, an abundance of EM dashes, the using groups of three to describe something. It's not this, it's that those weren't there. And so I think it is going to force this change as the capabilities evolve of like, was this AI generated? Was it not? Because I do think people are like, oh, if this is AI generated, I'm suddenly not paying attention. I think that will come over to photos, videos. We'll increasingly have those moments of like, is that AI generated. Is that actually a video of, you know, whatever I was watching. And so I think that will increasingly be important. I also think one of the challenges in the AI conversation, and we'll get into Tech lash a little bit later, is we're making policies for technology that is evolving so fast that the policies can't keep up. And I worry that the pace of evolution of these AI tools, policies. We're making policies based on what the tools were two, three years ago.
- [05:46]
A
Yes. And we're making the assumptions based on what the tools were even two or three months ago. I think at this point, because you're starting to. It's so crazy because as you say, as soon as people start to hone in on, hey, this is where the AI tells are. This is how you know that this picture is AI. This is how you know that this essay or, you know, this article is written by AI. Well, you know who's paying a lot of attention to that? The AI Frontier Labs and the AIs themselves. And they are very quick to close any kind of tells. I usually use Claude almost every day, but I actually have not used it since this came out. But I've been reading a lot of the coverage of it and it feels like it. As you say, education policy already doesn't move that fast. I mean the speed at which AI is moving is faster even than regular technology evolves, right? It's way the amount of change we've seen between the capabilities of the different Claude models or the different GPT bottles over the last three now four years, it's unreal. I saw somebody had an amazing post. It just reminds me of what you just said about they asked Fable about the Strawberry question, right? They were like, how many hours were in Strawberry? And Fable's answer was like, are you kidding me? You trying to get me with like trick questions from 2023? Like, come on. They're like, you think I can't figure that out by now? There's a super intelligence is all you
- [07:03]
C
could think to ask.
- [07:04]
A
It is how many Rs are in strawberry? Yeah, there are three Rs in strawberry. Like, let's move on. I was like, oh wow. I was like, that is an answer.
- [07:11]
B
It is much faster. At least in the test I did. Now it does burn through a ton of tokens. So I would use it wisely because you'll hit your usage limit quite quickly.
- [07:21]
A
And that, that I think is going to be an increasing, I think token usage. The sort of fuel of AI is going to become more and More and more part of the discussion now. Ben and I have talked for a while now about the sort of triage model. Like it should be that if you ask Fable a question like how many hours are in Strawberry and it immediately says I'm not going to answer that, I'm going to pass that to Haiku, right? Because that is a silly question. And you just wasted, literally it said you just wasted $5 on a question like that.
- [07:49]
B
I mean repl. It tells you how much a task cost. I think like if Claude did that, it would definitely be helpful. You know, I recently was talking to someone about like don't upload PDFs because it has to convert into an image and then run OCR on that image. Like upload the Word document, drop the link into the Google document. One of my other duties at Whiteboard is overseeing it. So I oversee our Claude subscription. So when people run out of tokens, I get the request for more. And so I see who the high token usage folks are. Right.
- [08:20]
A
And that could mean very sophisticated AI use because you're doing so much with tokens. Or it could mean very unsophisticated AI use, which is you're asking it to alphabetize something and it costs several dollars. No, I totally agree with you. I think that's going to be more and more of a thing. I also think there'll be a layer sort of put in. I mean this is going to become such a pain point. For now they' just minting money, right? Anthropic has literally made more money. I think it's the most revenue any company's ever made in the time period this. I read some stat like that because of this. It's like everybody's using it, Everybody's starting to become reliant on it. And the cost is almost like almost invisible for. For a lot of the end users. So it becomes just this IT issue. But I think over time the cost is as this becomes just like electricity. The electric bills are going to be become more transparent, more competitive and they'll be more baked into the system to sort of lower the cost. The sort of circuit breakers where you ask it something and it says, I really don't think you should do it this way. Like I'm going to give you the 50 cent answer and save you the $50 answer. But it is, I mean, I think to your original point, which is really key here, the speed at which this is changing and the speed at which education policy, both higher ed and K12 policy, is trying to figure out how to use this and what's possible with it. It's so tricky. It's so tricky right now. And I get, I get in some ways, even though I don't agree why, if you're in the education system, you're like, I can't keep up with this. And if we're giving kids or even teachers the ability to do things that are like this powerful, it makes me nervous.
- [09:51]
B
I mean it definitely comes back to the key message of like teaching kids how to understand AI, what it is, not just how to use a singular tool.
- [10:00]
A
Yeah. And how the systems work. Right. How it's a stack, how there's different, there's costs involved with everything, how there's a token cost, how there's, how the different models do totally different things and how to choose a model before you even start thinking about what to do. That's all really interesting. One of the other things that in the AI world, but I think this dovetails with EdTech in an interesting way. There was an announcement this week from Meta. So you know, there's been this, this sort of widespread pushback. We don't cover it very often on this show because it's sort of out of the education world. About data centers. People have been worried about, you know, placing data centers in different places. It's often in low income areas. They're like, it's these massive centers that are going to use a lot of electricity and water. There may be jobs involved, but people just don't really know what to make of them. They sort of don't want them in their backyards. And Meta put out a, an interesting launch this week where it's a, basically a trade training program for trade jobs, welders, electricians, plumbers with job guarantees. And they're doing it in, let me see, the cities, Baton Rouge, Houston, Indianapolis and Columbus, Ohio. So places that, that are probably pretty ripe for, for data centers. And I think the real thinking here is if we're going to build data centers, let's create, you know, meaningful job training around them that is free and open. It's $115 million program. I think it's totally free. I might have that wrong, but it definitely guarantees jobs. And the idea, it's called America's Workforce Academy. And my reaction to this is like on one level, okay, great, this is something that is meant to really help people. It's meant to combat job loss. It also feels, it has a little bit of a whiff of desperation to it of this idea of like Meta feels like it has to get way out ahead of the AI data center backlash. And of all the places to get way out ahead of it, I don't know why Meta feels like they have to do that more than some of the other Frontier Labs. You might have opinions on that, but it's a little bit funny. I mean on one level, you know, I'm all for job training, I'm all for skilling, especially low cost, especially with job guarantees. All of that sounds great. It also feels a little bit of a whitewashing to me. What do you make of it?
- [12:00]
B
My initial reaction and I you don't know this about me, but I lived in Baton Rouge for eight years so I have some added perspective there. You know, if I think about my parents who have lived in the same town in Mississippi their entire life and I asked the how familiar are you with OpenAI are anthropic. They aren't, but they know who Meta is because they have a Facebook, they have an Instagram and that type of name recognition. And when. So when they start associating that with AI and oh, are they building a data center here? I think it was in Nashville recently that there was a firestorm about a data center being built next to a zoo. And like what would the long term impacts of that be? So I do think this backlash is real. Currently live in D.C. right outside of D.C. there was a big Loudoun county is tons of data centers, always has been. There was a bit of a interesting news story I think that two or three months ago where the George Washington University in D.C. entered into an agreement to sell their Loudoun county campus to someone to build a data center five years from now. So they have done a five year transition on the promise that this land will become a data center one day. And so that's not an entirely industrial area where they're now building this data center. And so I think a lot of people are like, you don't want to see a data center next to your neighborhood. And I think there are increasing concerns about like how it's impacting utility prices. So this is Matta's attempt to get ahead of that and be like, you know, we're investing to make data centers equal jobs. You know, I lived in Louisiana, went to college, lsu. A lot of refineries, a lot of oil related chemical processing plants. Those were always framed as, you know, those are jobs for people. And I mean similar to even like when you see like private prisons being built in some of these like more economically disadvantaged areas, it's because those are where jobs that are coming in for community because of, you know, the small town I grew up in has no manufacturing anymore, but it was a manufacturing town for a large portion of like my parents generation's lives.
- [13:59]
A
A lot of really interesting takes in there. Yeah. Definitely do not know about your Louisiana background, but it's true that the name recognition of a place like Meta and also I neglected to mention this is being done in collaboration with Mike Rowe and his organization. So he's very well known as a sort of the dirty work, the blue collar hero guy that that combination may. I can see what they're going for there. And, and if you sort of zoom out and say, as you're saying, if you zoom out and say societally, if we're going to a place where data centers are going to be needed, where AI becomes just part of the landscape for everybody and data centers are one of the absolute core aspects of that, then it does make sense. I mean some of these jobs, like being able to be an electrician for a data center is a pretty good blue collar job. It's a pretty sustainable blue collar job. It's one that is unlikely to go away if you're trained with data centers as your sort of core use case. So there is something future oriented about it. It still makes me a little uncomfortable. Maybe it's just sort of the inherent inequality in it. The same reason that, you know, building private prisons in lower income areas to build jobs makes me a little uncomfortable. It's not actually.
- [15:09]
B
I mean it's a. When you start thinking about it like and digging into like the long term impacts of that. Yeah, it is, it does make you feel a little, a little icky.
- [15:16]
D
Yeah.
- [15:16]
B
I also don't think we're not talking about like some of the broader like impacts of data centers. Like long term whiteboard just moved into a new office. We had delays on getting ethernet cabling because data center companies are just spooning up all the bulk cabling and those types of impacts are down like with construction materials are going to start impacting, having downstream effects.
- [15:38]
A
Yeah. I always like to try to sort of think back to some of the other huge technological revolutions. Like the moment when, you know, telephone lines started spanning across all of the country or railroad tracks or big electric plants that, you know, started springing up which were very ugly and very, you know, and not super safe to live near, but it was necessary for the infrastructure for when suddenly everything became electrical. And I feel like we're just like in the next generation of that. I just. And you know, job training is a good thing.
- [16:07]
B
There is a researcher in the UK and her name is escaping me right now, but she has written about the cycle of outrage that comes with any type of new technology. And I think that's a really interesting way to look at it because if you look back in the 1930s in the New York Times, they interviewed a headmaster from a school in Boston who was talking about any problematic pupil. You can trace it back to a radio being in their home. Right?
- [16:32]
A
Exactly.
- [16:33]
B
And you can see these exact same things happen with TVs, with movies. We have this cycle of outrage that just like every aspect that touches the creation of radio, like the. The making of movies, those are having a negative spillover effect. And I do think we're having a bit of a. A moment of that with AI that anything remotely touching AI is bad because it's AI and AI is bad. And we're going to be outraged about that for right now.
- [16:56]
A
Yes. And I think that makes a phenomenal segue to something that is a real. Has been sticking in my craw for the last couple of weeks. And I just want to do a very mini rant about it. But you're a communications exper. Will have so much more to say than me. But I'm just gonna. I'm just gonna rant for a second and then let's talk about the. What you're calling the tech lash. I usually call it the EdTech backlash. It's the. The longer version, but basically. There was an article in the New York Times a couple of weeks ago from. Jessica Gross is one of their opinion columnists about visiting the first AI high school. I'm sure a lot of people listening to this have seen this go around. It's not. It wasn't just now. And. And she talks in. In the article about how she went with these dark fantasies of all of the horrible things that it. Like where everybody was going to be totally isolated with an AI tutor is going to be really horrible. And she walked out being like, you know what? That's actually a really good school. It's in Georgia, Gwinnett county in Georgia, but near Atlanta. She's like, that was a really good school. They did all sorts of amazing things. And you know what? It wasn't because of the AI. It wasn't the AI. It was the teachers and it was the. It was the curriculum and how it was, how they organized it. The AI had nothing to do with why that was a great school. And like, she did this ridiculous cognitive dissonance. I found it insane. All about because it's so entrenched in journalism right now to think AI, especially when it meets education, is just negative. This has switched enormously in the last six months. And you're seeing this tech backlash, this AI backlash. It's against lots of things tech related, but really coming for ed tech in a lot of different ways. I know you know a lot about this. There's also something from the city council in New York this week. Let's zoom out. Tell us about your perspective about the tech lash, why it's happening and what we should be doing about it as an ed tech community.
- [18:41]
B
I mean, it's the topic that everyone's worried about. It's the number one question I get from clients, from reporters, like, who sorted this out? You know, and I think every, I think now every major outlet has done at least probably two stories on the outrage, whether that is interviewing a parent who went to a school board meeting to talking to, covering LAUSD and their changing approach to ed tech. The New York City letter, which I know we have on the list to talk about. But I do think there is this conception that if that ed tech means a student is sitting on whatever learning technology, looking at the laptop all day, every day, the teacher's just there, sort of facilitating. I imagine that Jessica Gross expected to walk into like Alpha school where students were sitting on there and had like an AI tutorial or an AI teacher. And that was not what she saw. The idea that you can learn about a technology, learn about AI without using technology seems to be something that people are struggling to wrap their heads around. But we even see this, like when we looked at some of the, like Senate hearings around the use of technology and kids and the impact that it's having, the way technology is described, you would think that kids are just scrolling TikTok all day in classrooms, like teacher assigned TikTok. And I think, I think there's also this, like anecdotes are becoming the headlines. And I think that happened with YouTube. Like kids are spending 90 minutes on YouTube. And it's like, well, if kids are spending 90 minutes on YouTube, I think we need to look at what's going on in that classroom. You know, Tammy Winkup at Securely, who I think you've had on before, she put out an analysis of their data and they showed the numbers were more about like five minutes on average of kids on YouTube a day, not the nine. Like, we're letting the extremes become the headline. And I get from the journalistic pressures of like, what are your click rates? What are your audience like those are the headlines that get traffic. It gets the outrage on Facebook and Reddit and drives traffic to their articles. But we're missing the nuance and I think we've simplified the screen time and tech conversation and in most media to a simple quantity argument because that's easier to explain and understand and not a quality argument.
- [20:55]
A
We'll be right back. Innovation in Pre K to gray learning is powered by exceptional people for over 15 years, edtech companies of all sizes and stages have trusted higher education to find the talent that drives impact. When specific skills and experiences are mission critical. Higher education is a partner that delivers, offering permanent fractional and executive recruitment. Higher education knows the go to market talent you need. Learn more@higheredu.com that's H I R E edu.com. We'll be right back. Innovation in Pre K to grade learning is powered by exceptional people. For over 15 years, EdTech companies of all sizes and stages have trusted higher education to find the talent that drives impact. When specific skills and experiences are mission critical. Higher education is a partner that delivers, offering permanent fractional and executive recruitment. Higher education knows the go to market talent you need. Learn more@higheredu.com that's H I R E edu.com. Yes, I would even say one more thing beyond quality and quantity, which is, and I'd love your take on this, I've been, I've been talking about this for quite a while on the show, so I hope I'm not repeating myself to frequent listeners, but to me, part of the problem is yes, there's pressure for click rates. That is a big deal. There's a pressure for story. Right? People need to understand the story. This is a very complex subject and education in general is complex. Every classroom is somewhat different, districts are different. There's no one through line of what, what education, you know, looks like in, in 2026. It's just not. That's not really true. It's so different in so many different places. And I think we're getting to a place where in this complexity and confusion, the clearest headline, as you say, the clearest way to talk about it, is quantity but also bad. It's like AI and tech are bad. Just like you said. They equated to YouTube, they equated to entertainment or to isolation. It's partially our own fault. A lot of the pictures of ed tech in class, you know, if you have a picture in a, in a newspaper about ed tech, it usually shows lots of kids looking at their computers alone. It doesn't often show kids working together or doing something with a screen free device or doing something with a teacher and a tech. It's. We sort of did it to ourselves in some ways. But I think there's some feeling of like, okay, it's easy to understand that narrative that tech has not improved scores, that tech has gone to the classroom, has invaded the classroom, and that kids are now isolated on their devices because parents see that at home, right? It's familiar to them. They see their own kids on their iPads, on their phones, and they're like, that's what's happening at school. That's horrible. No more screen time. And that narrative, it's anecdotal, but it's also easy to understand, has just absolutely infiltrated the entire. There's an article that you called out today. I thought it was really interesting about the city council members. About just over half of the city council members in New York are now urging the new bear, Zoran Mamdani, to basically put a moratorium or stop its AI education push. They were going to do an AI high school. It was put on pause. They're doing all this, they're saying, no more AI. But one of the things that was really interesting is the new chancellor there was basically saying that the public opinion on AI had drastically changed just in the last few months. He became chancellor in January and suddenly it feels invasive, it feels scary. But I feel like it's the card leading the horse here because it's a story that is so easy. Jared Horvath came out of absolutely nowhere and he's like, I'm a neuroscientist and kids are using YouTube and kids are using this stuff and the research isn't very good. There's not enough evidence and everybody's like, got it. No more stuff, you know, no more tech. It's a scary moment, but I also think it's our fault because we don't have the really catchy narrative on the other side. We should, but we don't.
- [25:03]
B
A few points there hit me on the digital Delusion book. It definitely came. I was last week I was at the Education Writers association national seminar. Came up a few times. I am eagerly awaiting someone doing like a deep dive into the actual research behind that book. One missing piece from a lot of this coverage of that book. It was originally self published and then it was reprinted by a major publisher. But it was a. It started as a self published book. And so when I've talked to folks with more academic, like several economists who Study education. They've raised questions about, like, the correlation versus causation of the research presented and that it's just not rigorous research. But it has been an answer similar to the way the anxious generation was. An answer on student mental health is an easy answer. It's like, oh, if we pull tech out, that is the silver bullet to fixing all of our woes in education on the like, narrative side. And I think this goes to why I think the science of Emily Hanford and her Insult a story were so successful in driving a narrative on Science of Reading is that a parent who has limited in the general public, who have limited understanding of education and the nuances and pedagogy can understand phonics versus not. My grandmother has really strong opinions on phonics and science of reading. You listen to Sold a Story. I think that is the only piece of education journalism she has consumed. But it appealed. She understood what it was talking about.
- [26:30]
A
Totally.
- [26:31]
B
I think that's why we won't have this. A sold a story of people always ask like, when's a sold a story of math? I don't think that will happen because it's not as easy to explain and just bring everyone along. And I think the easy story is tech bad.
- [26:44]
A
Yes.
- [26:44]
B
Take it out of classrooms. And so that is what's driving the headlines.
- [26:48]
A
The other thing is Sold a Story had a villain. That matters. Right. People are trying to find villains in this era. And I think that's a big part of. They are very suspicious that the rich elite, often tech in this world, are out to sort of get them. Right. I mean, if I had a nickel for every time somebody has said to me in some context, context, you know, if you're not paying for a product, you're the product. I'm like, okay. But it's like, it feels really good to feel like you're seeing through a conspiracy, you're seeing through a scheme to sort of have something happen. And when it's. And when you're combining tech, which, you know, there have been real issues with the. With the tech world, real issues with the tech world being immoral in various ways. And then you're worried about kids in schools, it's a very toxic brew because it's like, well, who's the tech people are coming for our schools and our kids. Oh my gosh, we got to fight them. It People love a villain and Sold a Story had a very clear villain to it. And you know, it's not a real villain, but somebody who was. I mean, but yeah, I Mean, you
- [27:50]
B
know, there are some profile pieces of Lucy Calkins that she is. Is the villain of that story.
- [27:56]
A
Oh, no question. She's a villain in that story. And I think that's part of why it resonated. So I'm not saying we need to find a villain. Like, I don't want to make Jared Horvath into the villain of ed tech or Hyde or anybody or Randy Winegar. I'm not saying we need to find a villain, but I do think we need to find some way better, clearer, simpler stories for ed Tech. And why it's amazing. And I've been talking about this for now, like three years, I think. But, like, I want there to be a story that is just undeniable that, you know, the. The kid from rural Kentucky who cured his grandmother's Alzheimer's because of. Of AI. Like, a story that people are just like, what? And we. I just don't see them anywhere.
- [28:37]
B
Yeah, I think there's definitely a hesitancy to stick your neck out on sharing some of those stories. And oftentimes when you think about all of the layers removed from being able to find those stories. One of my first questions when we onboard a new client at Whiteboard. And so I work with many K12 ed tech companies, nonprofits for profits, associations. It's like, who are your best stories? Who are your best customers? I want to get on the phone with them and talk to them. And that is a process of finding the right people, getting connected. Because you're going through different layers. Even the smallest teams will have, like, trouble, just like getting people on the phone, like schedules. It's crazy. You have to go through the district comms office if you want to tell that story. And who knows what's going on locally and you don't want to, like, run afoul of what's going on. So it is. There is like a sourcing story. It's hard. And I think reporters are struggling to find those stories too. Like, they're struggling to find the people willing to talk. I think my biggest fear with when you read the letter in New York City is that it is lumping together using AI tools or AI chatbots, both those designed for consumers and those designed for education, with teaching students about AI. And when we say AI as a topic is bad, we stop teaching students what AI is, how to use it, and ultimately that's going to harm them far more than using an AI tool in the classroom.
- [29:57]
A
Yes.
- [29:58]
B
And so the instructional team at New York City is great. And I'm sure They are having wonderful, thoughtful conversations. I worry that the narrative is going to throw out the like, it's going to be like, oh, teaching about, you know, how to use AI literacy is the same as teaching students how to use ChatGPT. And I think that goes back to what Jessica Gross was so shocked that we were that AI high school was teaching about AI without students being on screens all day.
- [30:23]
A
Yeah, we only have a couple minutes left and there's a couple things I want to get to. Another related story that everybody should look up because I think it's a really nuanced and interesting is there's a story from NBC News about a reporter who basically goes to the Google training for teachers and sort of is trying to evaluate what Google is doing with its training and how are they trying to treat AI in the classroom and then talks to iste. It talks to a lot of different people in the space and tries to sort of make sense. And it's a good article. It sort of has both ways. I worry a little bit that the takeaway could sort of be seen as, you know, Google is trying to sort of fool teachers into using AI or sort of like turn them into pawns in its AI. It's a high game. Which is like, I don't think is remotely true, but I think it's an interesting article. It's worth looking up and I, I think it's maybe a little bit of the antidote to the, the pure tech backlash where people are just like, they're out to get us. But a couple things I wanted to talk about as well. One is the, the new NAEP scores came out. Give us your take on the nape scores. It looked like different for different age groups. What, what do you think is the big headline for the NAEP scores? It just came out, I think a couple days ago.
- [31:26]
B
Yeah, I mean, I think it might have surprised a few people that there was some improvement there. It wasn't the negative story that everyone has become accustomed to. Nine year olds are climbing back, notably our youngest, most screened native kids that we've ever tested. But then 13 year olds are reading about where their grandparents did in 1971. And so we are seeing recovery. It's real, but it's narrow. I do think the story that no one's really leading with is that this will, because of cuts to ies and test cancellations. This is the last long term Trend test until 2033. So we won't get another look at this type of data for what, seven years?
- [32:05]
A
Seven years, which the world will be very different in seven years.
- [32:09]
B
I mean it'll just all everyone's flawed agents will just be talking to each other.
- [32:14]
A
Exactly. Yes. Pretty flat scores for 13 year olds in both reading and math. Nine year olds have a little bit of a bump up especially in math and especially for boys and for economically disadvantaged students. Some, some silver linings, especially for the younger generation or the younger, the younger set of the younger cohort in the NAEP scores. One thing that jumped out to me is that the percentage of students who reported reading for fun continues to pretty much plummet. It hasn't gone down that much further, but we're talking about, you know, 14% of 13 year olds read for fun. One out of, you know, seven. That is quite scary to me. That's down from more than double that in the 90s. That doesn't make me feel great. Even more than double that in 2004 we still had 30%. So we've really. And you know, people blame tech for that and that's probably not fair.
- [33:05]
B
Is it the fault of tech or do we need to go look at the data on how schools are assigning full books to students? Because I think there's been a lot of conversation in the reading space around that recently and the decline of assigning passages versus a full book and I think that's concerning. And could that be impacting students ability, like interest in reading full books? Yes, I would say my one caution on the NAEP scores and I haven't seen this yet, but maybe I haven't looked in the right LinkedIn post. I'm waiting for someone to be like, see, we banned tech and now we've seen a recovery. There is no way that like these test scores reflect any type of tech. Cell phone ban, any type of like, that's just not the way the testing works at the time windows. But I, I'm fully expecting to see that LinkedIn post saying that, you know, now that we've banned cell phones, NAEP scores are on the rise.
- [33:57]
A
It's a confusing time and people are very desperate for answers. I really think that that sort of underlies so much of the fuzzy logic that we're all doing right now in various ways about why this is happening and why that's happening. And it feels so good to have a clear explanation, especially if you're somebody who does, like you said, who doesn't follow this stuff closely, who's like that doesn't know what open AI and anthropic are going public and they're competing for talent like he was just like, I just want to understand what's happening in my, in my kids or my grandkids school. And to be like, oh, you know what? There was a pandemic and everybody got all this technology and they realized it was too much tech and kids were spending too much time on YouTube. And now we're, now we're fixing that. Like, that is such a simple story. And it is. So as somebody who's interviewed hundreds of edtech founders, you know, we. Tens of thousands of edtech tools. It's like, it's like misses the entire industry, literally misses the entire. It doesn't even mention EdTech that story. But it's getting brought down with the ship.
- [34:55]
B
Oh, I know we're in time. Thank you so much for having me.
- [34:59]
A
Last thing. Code.org, rebranded this week, and I know that this is something you care about. Code.org is now code AI.
- [35:05]
B
Yes.
- [35:05]
A
Big shift to AI education. Tell us about it and then we can go.
- [35:09]
C
Yeah.
- [35:09]
B
I mean, it's, I think, a natural evolution of the work. You know, earlier it was last year, Hour of Code became hour of AI and really reflected that shift of how AI and coding are related. I think increasingly seeing code.org put that emphasis on teaching kids not just how to use AI, but how to understand it and how harnessing that understanding of AI can empower them and to build their feature.
- [35:33]
D
Yeah.
- [35:33]
A
And AI is increasingly the entry point to coding now, which is, I think, really exciting for a lot of people who may not have been, who might have been boxed out of it in the past, which is exactly the audience
- [35:42]
B
of Code.org as someone who vibe codes quite a bit, you know, I enjoy it.
- [35:47]
A
Exactly. Thank you. Thomas Rogers from Whiteboard Advisors. Appreciate you being here. Let's go to our guest segments. We have amazing guests today. We'll see you all next week. If it happens in EdTech, you'll hear about it here on Week at EdTech from EdTech Insiders. We have amazing guests this week. For EdTech Insiders, we are speaking to Yi Ng. She's a product leader and founder focused on AI learning and human centered design. Specifically, she's the founder of Curiosities AI, which is creating AI ethical AI tools that help children build curiosity, communication and critical thinking skills. Very important right now. Her work advocates for AI that empowers young learners while keeping human values at the center. Iung, welcome to EdTech Insiders.
- [36:33]
D
Yeah, thanks. Thanks, Alex. Lovely to be here.
- [36:36]
A
Absolutely. So, you know, we got a chance to chat and meet at the ASU GSV summit, and you are doing something really unusual and I think pretty unique in the AI space. So I want to make sure that we understand it before we even jump in. Give us the elevator pitch. What is Curiosity's AI and how are you you working to keep curiosity and critical thinking at the center of the experience?
- [36:58]
D
Oh, what a great question. I think, first of all, I started Curiosity AI because of my kids. I'm a mom and a founder, and it's all about how do we keep those curiosities alive. My kids are super curious and they're super motivated to learn. And one of the things that we're seeing is that when the learning is not really taken account into the stuff that they really care about, they don't learn as quickly. So that's the main reason. And when we think about designing Curiosity's AI, it's all about how do we design an AI that really helps kids think rather than think for them. So I think that's probably the most important thing.
- [37:42]
A
Yeah. So when you say helping kids think and helping foster curiosity, what does that mean for the actual design of the app? How do you design in a way that instead of the AI sort of answering everything or taking on the thinking skills, that it really keeps the child in the driver's seat?
- [37:59]
D
Well, first of all, kids are super curious, and we know that as parents, and they ask questions very differently. So they ask questions about animals and Pokemon and six sevens and stuff that's so weird. And, you know, it's very interesting. And today's AI is actually built for adults and gives answers, and I think that's actually really harmful for kids. AI has to be designed differently. And what I mean differently is when we think about what AI is doing, when they respond to a kid, it really needs to come back not with the answer, but with a question. And that's what matters too, kind of the interaction. And I think when you think about designing, it's also in technology today, we're designing a lot of for streaks and for kids being engaged in the app, and that's actually really harmful for kids as well. I think in our conversations, we talked about what's the right thing for kids. And so we shouldn't be designing for engaging in the app as long as we possibly can, which I think as parents, we don't want that. And as a founder, I don't want to see that in the technology that I'm building, but instead it really should be about how do we design something that is all encompassing in terms of helping them learn. And then create the love of learning. And without streaks, how do we get them engaged? How do we get them to want to be in here? And I think that's the key in designing. I can't say we figured it all out, but one of the things that we're actually starting to think about is how do we get to build breaks into the app. And I've never seen that in any of the apps out there. How do we create more human connections? By building parent dashboards, which we have, but more about connecting to go talk to a parent, go to the library, go read a book and do a science experiment. And that's the kind of things that I haven't seen in technology. And we have to start building that for kids because otherwise they're just playing with streaks. And the streaks takes over and the love of learning kind of subsides and the curiosity subsides.
- [40:15]
A
Yeah, it's a great point. The only tool I've ever seen that sort of encourages you to step away from it, if I remember was the Wii, the original, the Nintendo Wii system. I thought, yeah, it would say, why don't you go outside and take a break? Right now you're playing a lot of the game. And I was like, that's a really interesting idea. But it makes sense, especially with AI and especially with gamified apps. The type of, you know, you're mentioning streaks or sort of eng engagement tactics that are designed to keep the kids sort of the gravity pulled back into the app and turning it around and saying, having the kid be going and exploring, going outside, going and doing signs, it's already a very different relationship. Right.
- [40:53]
D
And, you know, I think one of the things is, like, what if the best AI, that session ends with the child running back to the parents and asking better questions.
- [41:03]
A
Yeah.
- [41:04]
D
And then going to, I want to go to the library. Let's go to the park and explore the ladybug. I just talked to Ace about the ladybug. We have an app that's a voice AI, and so it's all about not using screens, but using communication, using conversation to really enhance that human connection.
- [41:23]
A
You mentioned Ace, and I think one of the things that we've wrestled with as an ed tech and education community is the role of AI companions or sort of always on tutors or the Aristotle to your Alexander the Great. You know, people are thinking about, what does it look like to have a sort of brilliant mind that's alongside you? And I think you have some really interesting ideas about the concept of an AI companion. And what you're trying to do, it's not a traditional AI companion. It's tell us what it is instead.
- [41:52]
D
First of all, as a parent, I wouldn't invite a stranger into my house and have conversation with my kids. You know, like, if you kind of think about, like what the real world is like, so what is it when you invite somebody actually into your home talking to your own kids? It's about building trust. And it's about building trust at the deepest, deepest level. What does that mean? Right. And so I think it comes in two parts. One is parents are in control. So we are building ace. ACE is an octopus that loves to teach kids and love to learn with kids and explore every knowledge in the world. But we need to give parents control. We need to give them that sense of safety, that sense of guardrails and limits like we talked about breaks things that are really helpful as a parent unit, as a family unit, so that parents feel safe. So I think that's the first thing. The second thing is we can't do what regular companions are doing in terms of keeping them engaged with the companion and letting them getting drawn in as though this is like what they only cares about. AI is actually really good at being positive, really good at telling you you're right about things. And we can't have that because what do parents do? They're always telling kids you're wrong or here's how to think about the world, or here's how I would shape it. So we have to find a way to get parents to help shape the experiences with their kids. And so this sense of trust, not only do we need to give parents control, but we also have to make sure the conversations in which we're providing is not a therapist companion based, but it's more about critical thinking and learning. And that's something that I haven't seen out there. I've tried all the different AI apps before. I designed this for the kids and I couldn't bring any strangers into my house. So I built ace. And ACE is our little, you know, octopus that we trust. And I think if we can trust it, I feel like, you know, other parents hopefully can trust it as well.
- [44:06]
A
Yeah, I mean, the trust component has just risen again and again in the context of AI. I think people are parents, educators, even students, you know, have to really figure out their relationship to this type of technology. I want to double click on this concept of parental control because I think that is it can take many different forms. And I'm curious how you're thinking about parental control. Is it transparency so that the parent knows what the student is talking about? Is it the parent sets the constraints or guardrails or time limits? What type of controls are you giving parents?
- [44:38]
D
Actually, both and more so. First of all, I think the first principle is transparency. So if the kids are having conversations, you have the history of the kids, the we call them wonder cards, so that you can see what the history is. You can set guardrails, you can set sensitivity based on the kind of topics that kids are having conversations on. Most of the topics are already guardrailed, but you can set the sensitivity of medium high, medium low to help understand, like, well, how do I want Ace to talk to my kids? Do I want the conversation to be stopped right away and then guide to a parent, do I want it to just say, you know what, this is not a topic we wanted to have a conversation on. Let's shift and then talk about these areas. So a good example, my daughter was asking about Judy Blume and the books and then the teenage years and just her body. And I think it's really good to talk about her body and help her understand, you know, her body. And when she was having conversations about the book with Ace, Ace was able to kind of explain to her cultural relevance of, you know, the author and what that means and how to think about what those topics means. And that really helped her understand how to engage in her mind about her body. And I really appreciated it because as parents, we're kind of like, I don't know, how do I have this conversation? So it's actually really interesting. It's less about just not having a conversation, but it's about how to appropriately have the conversation in an age appropriate way. In a way where parents, when they have either seen the transcript or for us, we were in the room when my daughter asked the question. We were very comforted in how AI was able to answer that question or Ace was able to answer that question and then felt like, oh, here's someone we can trust. And I think it's not just in the guardrails, it's not just in the transparency, but it's in the conversation. The age appropriateness of the conversation and how you carry that conversation and what kind of weight you put on the topics. And then it's actually quite nuanced. And that's the stuff that we are designing around.
- [47:05]
A
Yeah, no, it is nuanced and it's complex when you have this sort of open space where anything can be asked and it's driven by curiosity and wonder, then everything's on the table at first. But the AI has to be smart enough to know how to de risk conversations or redirect conversations or when to just escalate. And it's interesting because I think this is something true of many different open ended AI tools that young people use. But there's no sort of standard across the board. There's not a rulebook for how this works. So I admire you for delving into it so directly.
- [47:40]
D
And I think we have to start building that playbook, you know, for our parents. And it's not something that I think anyone has been able to done well or right. But if we don't start working on it, our children are the one who are not getting the, you know, the knowledge, the access to information, the access to knowledge, the ability to learn. Because we are saying no, no tech, no this, no that. It's less about no screen time, it's about the quality of the screen time. And I think that's something that parents are also just juggling with because technology out there are not building for that. They're building for streaks and engagements and being on the app and they're not building for thinking and asking better questions.
- [48:33]
A
Yeah. And sometimes the ecosystem just sort of throws kids. Even if you're watching a safe YouTube video, it'll then recommend a different one and take you in a funny direction. Or an ad will come up in the middle of a kid's app. That's inappropriate. And it's a tricky environment because everything sort of bleeds into everything else. I wanted to jump back. You said something that really stood out to me and I'd love to dive into it, which is that concept of, well, AI is sometimes very sycophantic. It'll tell everybody. Oh, you're right. What a great question. What a great idea I'll be. You know, and parents, often, that's not always the role they play. I mean, obviously parents are supportive, but sometimes it's about telling kids where they've gone off or how to think about something differently. And your whole company is called Curiosities. And I know you think about curiosity and wonder. I'd love to hear you sort of talk about how both the human and the AI can maintain that sense of curiosity and wonder in, in young people. Because it feels like, you know, we all start out asking so many whys and so, so curious about the world. And then over time through, sometimes through school, sometimes through parental involvement or friends or life. Yes. It sort of gets knocked out of us and people Stop asking why after a certain age. But I love that you're focused on curiosity, and I'd love to hear what that looks like to you, what kind of things kids ask and how these systems can maintain that sense of. Of wonder.
- [49:56]
D
So I think, first of all, we're designing for actually elementary kids and middle school kids. And at that stage, their values are still very much influenceable by parents. So I think that's the first thing to remember as parents. You're strong and you can help them shape what's the right thing. So that's, I think, from a parent side, from a kid side, at five years old, I think it's the peak of curiosity. On one hand, that's a really positive thing. On another hand, you can kind of see like, well, that's not great. If at five is where we pique our curiosity and we ask 20 to 30,000 questions a day, then what? Then over time, it just goes down and is that what we want?
- [50:44]
A
No.
- [50:45]
D
No, that's not what we want. So I think the most important thing is think about why that curiosity is dipping. And I think when they start going to school, they start to have more structured learning, and they're also starting to. And I'm not sure if this is like, maybe not a generic thing, but kids as an adult, I can at least see that myself. Sometimes I'm afraid to ask because maybe I should know this. Well, if I don't know why a scent is so strong when it's in the hand sanitizer versus a scent that is, you know, from a candle, I don't know. Is that a stupid question? You know, like, you run through these ideas of judging, right? And kids don't judge. Kids don't have any filter. So over time, because of life or because of us thinking what we should know, we stop asking. But we actually don't know. Not just the answer, but we're not even asking all of the related questions around it. Veritasium actually has a really good video which my kids has watched on why is rainbow curved? After watching the video, kids didn't really understand all of it because it was kind of built for a little bit more adult. So they start asking questions to ace, why is it curved? What happens if the reflection, the refractions. And then. And I'm sitting there going, I don't know any of this stuff, right? And I am thinking, oh, my goodness, thank goodness they are asking. So I think that's kind of the context. Like, you know, first of all, parents have influence. Second, kids are Actually really curious. And all we have to do is keep that curiosity. And I think your question's really good in that it brings out something really important, which is, is curiosity driven learning is actually the most motivated learning. I have seen articles and research that shows like 2x10x, like, you name it, every adult will probably agree. If the kids are curious and they want to learn, you can't stop them, right? You cannot stop them. And they will find their way into the dinosaurs and the whales and the, the rainbows and colors and whatever they want to know. So the reason we focus on curiosity is because we want kids to actually love to learn and we want to keep that curiosity. And we believe that from a system perspective, we can build a system that can keep that curiosity going, get them to ask better questions. However, I think there is a area that we started the conversation on which is positive thinking and AI doing all of that positive reinforcement. We've actually built it into ACE to not provide as much of the positive reinforcement, to be more critical to the question they're asking and then getting them to think, but also react in a more neutral way. And I think that's something that's really important to think about. They're curious, but if you just keep enforcing with curiosity. Yes, you're right. Yes, you're right. You got this. You got this. You know how this works, and you know you understand how it works. Yep, yep, yep. It's not the right way for kids to learn because they need exploration, they need to pause and think deeper. You need to kind of poke at the kind of question they're asking and say, well, why do you think that and what makes you think that? And that's the kind of question that ACE is actually asking our kids, and that makes curiosity stronger. I have so much to say. And it's like, I think about this every day. And I think for kids, or maybe for parents, the question is like, what if every single learning session that your kids have starts with the stuff that they are actually interested in? So don't start with math, don't start with science, don't start with social sciences, don't start with history, but it can lead to all of it. Start with whales, start with rainbow. And it can lead to science and can lead to eyeballs and reflection, refraction can lead to math. Can start with the bridges, it can lead to construction. And so what if curriculum is built differently? And I think there's a lot of early childhood education systems that are starting to do that. But even for secondary and for high School, we can start thinking curiosity based learning, and that could potentially change learning forever.
- [55:35]
A
I totally agree. And there really is an actual body of research about. Exactly. I mean, I know you just said that, but just for the listeners, curiosity based learning is an actual real type of learning, like project based learning, like problem based learning. And you're absolutely right. The research is very strong that that is such a motivating factor.
- [55:54]
D
Inquiry based learning, Socratic learning, and I think asking better questions. So imagine if schools are not measuring on the kind of answers kids are giving, but measuring on the quality of the questions that they're asking.
- [56:12]
A
That's a very powerful vision. And what's so interesting about it is it's a type of relationship or a type of communication, I should say that we've just never had access to before. I mean, I think we think of all the different things that AI can do, and I think maybe we take for granted that it can sort of spur curiosity. But I know that I walk around as an adult saying, huh, I wonder why this and that. And then just going right into the AI instead of, you know, it used to be you at Google before that you'd have to go get a book from somewhere, but now, no matter how specific it is, no matter how esoteric it is, you can get an answer.
- [56:47]
D
The answer's right there.
- [56:48]
A
The answer's right there.
- [56:50]
D
And maybe it's not good for us either. Right. Wouldn't it be really nice if we get to explore a little bit and then find the answer? We get to think a bit. And there are actually research about if you problem solve and you go through challenges and actually have grit and keep working on it and working on it and then solve it. There's a feeling that humans get that goes to the core of the love of learning when you actually learn. Learning is so addictive.
- [57:24]
A
Oh yeah.
- [57:25]
D
And forget the engagement metrics that people do on gamification and all these addiction metrics and streaks and you know, all the stuff that like technology comp. Forget it. Love of learning is so much more addictive. But you have to go through the steps of getting challenged, solving a problem and going through and actually think about it and going, wait, this doesn't make sense. This doesn't make sense. I don't get it. I don't get it. I don't get it. And then explore, explore, explore. And then when you finally figure it out yourself and by asking all the right questions, then when you understand it and really, truly understand it, that sense of learning is so addictive and you'll keep wanting to learn more. And that's the stuff that we really want to help our kids, you know, get to. And if they don't get a chance, because the AI out there that are just giving them answers are available to them, easy to access, and parents are not thinking about it this way, we're failing our kids. We're not doing them the right things. And as parents, I don't think that's the right thing to do. So I'm, you know, just really bullish about how to think about this. And I want parents and educators and I want the technology companies to also think this way. Otherwise we're not doing, you know, right by our kids.
- [58:48]
A
Absolutely. Two things strike me as I hear you talk about this, it's very inspiring. You know, one is you mentioned earlier on in the conversation, you know, how kids are so curious and they sort of dive deep into these subjects like dinosaurs or Pokemon. And it feels like what often happens with young people is as they go to school and they start to feel like they, they don't have mastery over the body of knowledge that school is about. Right. They don't know history yet, they don't know mathematics yet. They find bodies of knowledge that they can master and they're curious about and they go very deep. And that's why, you know, every kid of a certain age can tell you, you know, whether Charmander is fire based or water based because it's like it's not real. I mean, it's not real world knowledge, but it is real knowledge and it's, and it's deep and, and it's complex and system based and it feels like, you know, kids are sort of seek that systems of knowledge to understand and master and really dive into.
- [59:44]
D
They want it so bad. They want it so bad. My daughter Lana is 10 years old. Her first question to Ace was what happens to dust over course of time? Does it just disappear? Or if you stack up dust all the dust in the world, how far does it go? I think her recent question was like acidity of liquids, you know, what makes this thing acidic and base. What does that mean? And we are no longer able to answer that question in the right. In a way that actually like quench her thirst of the knowledge. So Ace is there to talk and then we're there to add. And as a family we're getting more and more curious. It's actually really amazing to see when you watch your kids and the kind of questions they ask.
- [60:33]
A
I totally agree. And the questions that you're mentioning that Your daughter is asking, I think are such perfect examples of taking that curiosity, that childlike, fun curiosity. What would happen if you piled up all the dust in the world? That's a really interesting question. Why are rainbows curved? Really interesting question. But it's the kind of thing we just forget to think about it. I think we dive into, you know, superhero universes or Paw Patrol. Right. And understand and start thinking of the real world as sort of opaque and complex and judgmental. Right. People judge you if you don't know about how smell works in hand sanitizer or you think they might. So you go to something that you can understand. So I feel, it feels like that that safe space of AI is really a powerful place for a variety of reasons. It's transparent to parents, but it's still a safe space that's non judgmental and you can ask questions and just go as deep as you want. And. And then, yeah, the other thing I'm kind of. The other thing I was going to say is exactly what you, you already sort of hinted at that idea that kids get into dinosaurs and the parents will say, here's the four dinosaurs that I know the names of, unless they're a paleontologist. And then the kid just keeps going and they learn 50 dinosaurs and 100 dinosaurs and they very quickly outpace the parent. And it's so interesting to have AI there to sort of allow that T shape that just, just go as deep as you want in any subject. And I'm curious what that's looked like in the context of curiosities. What do people do when they go deep in an app like curiosities and they're asking Ace, you know, a hundred questions about some particular topic.
- [62:02]
D
So one of the things that we actually designing specifically, and it actually took a long time to figure it out, is today's AI is actually designing for, I would say this thin layer of surface knowledge. And to actually design it for deep knowledge requires a different way of building a system. So what does like this thin layer surface look like? You ask a question, it says, oh, what are the kind of the adjacent questions or knowledge around it? You explore this one and then all of a sudden you start asking about stars and you're at the moon and then you're at the sun and then space and then yes, it's exploring, but it's exploring very shallowly. Shallowy, that's a word. The most important thing when kids are curious, they actually have something they wanted to know and ask and, you know, wonder about go deep and so if we can figure out what exactly they're asking. So first thing ACE does is actually help with work with the kids to kind of understand what are you asking conversationally. And because it's a voice AI, we're trying to conversationally understand. Ah, you really wanted to know the inner working of the stomach lining of a whale. Okay. And why? Oh, because you ate something the other day and you are interested in whales and you are wondering, how does this process in my stomach versus the whale stomach. Wait, they eat all this crap and little fishes, big fishes, how does it work? And so then you're like, ah, now I know why you're asking this question. So, and it's a very human thing, right? You ask me a question, I'm like, like, do you mean this or that? So first you have to understand why are they asking the question? And then you can go deep. And then one of the things that we're looking at research wise is thinking about what if you take a topic, once you understand why and really kind of use AI to build a PhD level understanding of that topic, then find a way to explain it to a 10 year old. So I don't understand how whales stomach lining work and how they're able to eat all this and not get a stomach ache. But a kid asks, how do they not get a stomachache? So take that and you go do all the research behind the scenes and this is something we're developing in the process. All the research in the universe go build a PhD level understanding of that particular topic. And every kid is unique, right? So that unique learning experience and learning journey, but not at the surface level, not at the shallow, not at the level in which you just like talk about basic stuff, but to really understand it, to really scientifically, mathematically, historically understand something and then not just tell them in a big answer, but work with them to explore it, figure it out, and then like really draw it out of them, I think, and I can't say we're there yet, we're working toward that, but if we can get there, man, what a world that looks like I want to be in it, because I have so many little things that is so unique to me and my weirdness, the way I question things and the way I think about things, the way I compare things, and that's the learning journey I cannot find for me that I want to be on. And I think if we even take a step back and look at AI, and today, I think adults are really worried about job markets and all the things that we're worrying about what the future of our children is going to be. What are they going to do? I think it's really important if our future generation is on the journey in which every little bits of curiosity can be expanded, expanded, expanded so that, that each person has their own identity, own learning, long journeys. I think we can do more because we can be more creative and, you know, AI can power the stuff that we don't want to do. And then humans can actually learn and grow and become in some sense another species. Right. So that's the positive stuff, you know, that I think about the optimism that I bring to the table. But we have to start with the kind of conversation that they're having today. We can't let those conversations just end with an answer and not keep that curiosity going because then we are not creating the creative minds of tomorrow. And that would be a shame.
- [66:46]
A
Absolutely. So many of the things you're saying resonate with me both from a learning perspective and just a personal perspective. I think it's a very powerful set of principles you, you're articulating here.
- [66:58]
B
Right.
- [66:58]
A
You know, start with trust and safety, make sure that the adults know what's happening and that they can, they have control over what's happening. Then, you know, create an AI model that is designed to build better questions and to ask questions and to ask why and to sort of dig in and encourage curiosity and not to judge and not to be sycophantic. And you start adding these things up and each one of them takes it, you know, further and further from the general purpose models, you know, where you ask any kind of question and it gives you this sort of like four paragraph answer that tries to sort of be the definitive answer. And then at the end it's like, well, do you want to go deeper into this or to something else? And you're like, wow, that is not the kind of conversation I wanted. Or well, what do you want to talk about next? And you're like, it's a funny attitude. And I think you're, it is an optimistic stance.
- [67:45]
D
You remind me of something. When I use the AI that's out there. For the most part, I can't remember the stuff that I asked. So if I need to go search bar, I'm going to go look for it. It's not in my memory. How do we get kids to actually remember what they learned? By exploring, by having a conversation, by learning through verbalization, by communicating, by speaking. And that's the human part of it. And when you said it, because then it's like, oh, yeah, we go do those things, but. But we forget. We don't even know what we are searching for. I think my husband and I were playing Tetris and we had a little, like, setting thing that we needed to set. And then we asked and figured it out and then we got it done. Couple months later, we're like, oh, my gosh, we're in the same spot. The settings run again. We did not learn anything. We went back looking for our history. We couldn't find the history. And we're stuck in this weird Tetris like, mode that we can't remember. And we're just talking as a family, like, what happened? What happened to our memory? Why can't we remember? And the funny thing is, you know, as I think about this, we were building this every day for kids. I can imagine. And we know as parents, this is making us more curious and more, you know, able to remember things. But you can imagine this could be really good for elders as well.
- [69:12]
A
Absolutely.
- [69:13]
D
Down the road. Right. It's not our market currently, but every human, we can't just sit there mindlessly typing and getting the answers and not remembering anything. Going back, searching for things and then having apps just serving like, you know, things up to us. We have to kind of, you know, it's a fiduciary duty to human race and humankind to think, to think harder, to be more curious, to ask questions, to remember what we've learned. And that's what I think being human is about. And technology and tools should enable us to be more human, not the other way around.
- [69:56]
A
I totally agree. And I feel like you're really on the way to that. I find myself having these long conversations with AI models and going really deep and having. Having deep discussions and sort of making sense of something. And then at the end of it, it's like, can you please synthesize this for me and do a synopsis and email it to me? Because I'm like, Even at the end of the conversation, you're like, this has gone so far and it's just gone. Which is. It's actually very strange. Right. I mean, you know, I'm laughing and
- [70:23]
D
I shouldn't be laughing because it's not funny, right. In some sense it's happening and it's like, maybe it's over for me, right? But I don't want it to be that for my children.
- [70:39]
A
Well, what it is, is it's not optimized for the human brain. Right. I mean, it's reactive. It'll answer the questions you ask. But it doesn't know how little working memory we have. Right. It doesn't know that we have to retrieve and verbalize, as you say, and do all these things to solidify our memory. Like it's not trained to know how we think and how we learn. It's trained to learn, learn. It doesn't sort of feed back to us in a way that actually makes it memorable. You can ask it for, for all sorts of, you know, mnemonic devices or various things, but it's not going to do it on its own because it's not, that's sort of not how it's trained. And I feel like what you're doing with curiosity is sort of trying to really dig into the curiosity of kids and build on it and teach them to ask better questions and to really make sense of the world is, is a, is a really big step in the right direction.
- [71:26]
D
Yeah. Yeah. I am so proud of my kids. I'm so proud of, you know, the companies that, and this company that I've built and if the two come together to make them even more curious, more able to learn and love learning, that's what we want. And I wanted to also offer this to other families and making it, you know, accessible to everyone so that they have a chance to see what our kids can potentially do. They're so smart, they're so curious and our only, really, our only job is to keep that going. Right. And not to keep the candle light going and not to push it out 100%.
- [72:11]
A
Light the fire, keep it burning. Kids are born with that fire for learning but we douse it unfortunately in the world douses it for them. So I, I love that metaphor, er, of Curiosity's AI. If you have not been convinced to download it from this conversation, I don't know what more we could possibly say. It is a really, really interesting, very human centered, very curiosity based, of course, approach to learning with AI. And it's a breath of fresh air to me because I feel like you have a very optimistic, very deeply held human perspective of what this could be and how it can build critical thinking and, and be a real boon for our kids rather than something that we should all be afraid of, which is increasingly the worry. We appreciate you being here with us on EdTech Insiders.
- [72:57]
D
Thanks Alex.
- [72:59]
A
We have a great guest today on The Week in EdTech. From EdTech Insiders. We are talking to Rob Abel. He is with EdTech Futures, which is his organization for investing in various things. He's mostly retired but he was the chief executive officer of One EdTech for many years and knows the EdTech space and the infrastructure inside out. And I wanted to bring him on today because he had some really interesting takes on the recent Instructure canvas breach. Rob, welcome to EdTech Insiders.
- [73:30]
C
Thanks, Alex. Thanks for having the program and thanks for having me on. Appreciate it.
- [73:34]
A
Oh, absolutely. Well, before we get into the canvas itself, give us a little bit of your background with what One edtech is and, and how it grew to be such a major standards body within the EdTech space.
- [73:47]
C
Yeah, it's a pretty interesting story. I'll keep it very brief, but in 2006, after already being an experienced entrepreneur and ed tech leader and so forth, that I took the CEO job at this tiny little nonprofit, six people that was working on standards, you know, interoperability, technical data, interoperability standards. And we grew it. In 18 years, we grew it from 50 member organizations to about 1200.
- [74:19]
D
Wow.
- [74:19]
C
So it was more than a 20x and it was, it was not an easy thing to do because the industry was really not, you know, there were a lot of little standards organizations out there, including what was called at the time IMS Global Learning Consortium, and we later renamed it to One EdTech. It was really a sea change. There are the standards that people know about. Many people know more about the standards themselves. There's one called lti, for instance, Learning Tools Interoperability, which really has been a game changer in the sector. And so it's become a major force where the organization obviously is still going without me. And it's, it's about as large as the World Wide Web Consortium in terms of the number of members, the number, number of revenue, the amount of revenues and so forth, which if you had thought about it when we started it and when we, you know, started turning it around in 2006, you never would have thought that could have ever have happened, to be honest. So it was a pretty amazing run there. And I think it's playing an important role right now, including in the data breach, potentially in the data breach that we're talking about here. And I'm going to be a little bit critical times here of One edtech because I think it has an important role to play that it should be playing. And I want to make sure people understand that. That's my main point. I think. I'm, I'm not a pundit, you know, or a blogger or anything like that. I'm, I'm a leader. And I think that we need to take this breach Seriously. And get the most out of it in terms of moving our sector ahead 100%.
- [75:42]
A
And I would be surprised if almost everybody in this podcast, their eyes perked up when they said, right, LTI, IMS, Global One, EdTech. I think I bet 90% of the listeners to this are part of the, part of the consortium in one way or another, because it really has been a game changer. Almost every EdTech product, I think every EdTech product I've ever worked on uses the standards and the lti, you know, standards to pass information back and forth. And obviously interoperability has become a huge topic. It's even more so in the last few years, but continually over the last couple of decades. So let's get into this Instructure breach and when EdTech's relationship to Instructure, tell us about what happened. Let's start there.
- [76:25]
C
Yeah, so there's a fellow by the name of Phil hill, Phil on EdTech. And Phil Hill is an analyst of the sector and he really has followed the breach very closely. I saw it as basically as an outsider to edtech because it made major news headlines really kind of everywhere in the non edtech press. And I'm like, oh my goodness, this is a big problem. And so basically what happened is there was a professional hacking group, group that through a series of attempts that started way back in September but then also culminated in late May, basically got access to all of the user information that's stored in Canvas. And Canvas has 60% of the LMS market. It's a huge number of users worldwide. I mean, we're basically talking about millions of records. It was one of the largest data breaches in the history of any, any sector. And Phil's excellent work really, I thought, was following what Phil was doing and he was checking into it. And in checking into it, he really indicated some things that were pretty startling to me that have ramifications on our sector and leadership when it comes to security and privacy, and also on some of the work of one EdTech. Because LTI was brought into the discussion, not because LTI was the source of the breach, but Phil felt that Canvas was not doing a great job at all. Matter of fact, a rather poor job of how they were handling the whole situation. I mean, essentially they ended up paying some sort of ransom so that the data wouldn't be released. Nobody knows how much it was. But you know, in theory, if you're going to pay the ransom, you should pay the ransom up front before they get the data, right? I mean, so Canvas instructure obviously thought that they could outwit these hackers, who in my opinion were not particularly sophisticated, but they were one step ahead in structure as it went. And the bottom line, and Phil's main criticism was that the way Canvas handled it, they just basically weren't transparent enough. They still haven't been transparent enough, although they have done a technical disclosure of what happened. But they also were not prepared to really help their customers, educational institutions deal with the fact that, that they had major downtime and this was right around finals time for some of the higher education institutions and so forth. So Phil was very critical of that. And my sense is that I was amazed by the fact that there wasn't more coverage in the edtech space, to be honest, because this is one of the largest data breaches of all time. It was against the leading, the market share leader in a very important category, the lms. And I think we really need to dig into this. My whole call to leadership here through a article I wrote on LinkedIn called a data breach that may live in infamy. If people want to read it, they can Google that was really to get into, well, what's going on here? What is One EdTech doing about. This involves LTI One EdTech has a program called Trusted Apps and Canvas is a trusted app. So we need to up our game. This is exactly the time when we need to up our game and get into this and really be leaders, you know, going forward. And I think Phil did a great job, but now it's time for the rest of us to do what we need to do.
- [79:43]
A
Yeah. Phil Hill is a legendary ed tech reporter and especially around the LMS space, he's been sort of the leading voice on the LMS space and how it works and how it's evolved for many years. And yeah, he had a lot to say about the brief breach and it was, it was interesting. It seemed like the way this breach sort of played out, they ended up paying the ransom, but there were many steps along the way. They sort of kept going back and forth and Instructure kept ignoring them or sort of going back and forth and they kept escalating. And it was really like it was almost this conversation between Instructure, Canvas and this hacker group that sort of kept escalating until it became obviously clear that they truly had access to this, these unbelievably large amount of data. And of course, this is students, it's K12 students and higher ed students, it's professors. I mean, it's, there's a Lot of information in an lms.
- [80:32]
C
Oh yeah, absolutely. And you know data privacy is very important and you can look at from the higher ed perspective. But Canvas is very big in the K12 sector as well.
- [80:41]
A
Yes, of course.
- [80:42]
C
And that, that makes it even worse. But one of the major criticisms here is that in this like sort of four step process, this for series of four breaches after the second or third basically Canvas, you know, there was a little bit of a disruption somewhere and basically Canvas said well we've got it all figured out, it's all clear, this is what they told the customers. And then literally like the next day or two days later, boom, you know, basically the systems were down for a significant amount of time. So the question is kind of really like what do we do, you know, going forward and what does it mean for the industry? And I think what we have to understand here is that what we're seeing, you know, people are asking why, why did this happen? I think what we're seeing here and the important message here for whether it's for Canvas or 1edtech or Canvas's competitors or whatever is what we're seeing is the result of the investments we're making in security and privacy are not sufficient. That's bottom line. I mean we're getting, we're getting outwitted, we're getting beat. We're going to, we're kind of loose goosey were very complacent about it. This is one of the reasons why I felt I had to speak out. I'm retired. I'm not beholden to anyone. I'm not afraid of what a supplier might do to me because they're, I'm not buying anything from them and they don't sponsor me to do anything and, and so forth. And I think the bottom line is one of the things we've learned over the years, especially in the world of standards is accountability is actually a good thing. It's actually good for the market. It's actually good for the suppliers to be serious and accountable about these things because like, because that's how they get ROI out of standards. That's how they get return on investment. They're not serious about standards and they just become a marketing badge of some sort. You know, you get a certification and boom, here I'm this, I'm a trusted app or whatever. I'm again I'm picking on one EdTech but there are lots of other examples in the industry but I know one edtech very well and so I'm just putting that out as an example, you get that little marketing badge. But the reality is that marketing badge has to mean something. Those certifications have to mean something. And organizations need to be ready to stand behind those certifications. And when the information on LTI came out, the main thing that Phil uncovered about lti, again, LTI was not the source of the breach. And the version of lti, the modern day version, which in say, modern day, this is really like five, six years now, is a version called LTI 1.3. And this is a very, very secure version of LTI. In fact, it's probably more secure to be using LTI for integrations than any API you're going to get used directly with an LMS provider. But the problem, what Phil uncovered is that Canvas was actually still supporting an old deprecated version called LTI 1.1, which IMS 1 EdTech deprecated literally in like 2018, 2019, even after a long process working with these suppliers. And Canvas is obviously one of them. And they were even on the IMS1 EdTech board of directors. And so I nearly fell off my chair because believe me, LTI 1.1 is very, very hackable. It's very insecure. I mean, you couldn't have a more insecure way. Basically, it's got shared secrets and keys using a deprecated version of, of OAuth. Okay, that's from, you know, that was deprecated long ago. So I felt that when we changed LTI 1.3, we were fortunate that we hadn't had any major mishaps with LTI 1.1 up until that point, you know, 2018 or whatever it was. And now to see that Canvas is actually promoting it, you know, Canvas has its own directory of apps. Just build out its integration ecosystem. And I went there and I checked because there was nothing in, you know, I couldn't tell. I could only tell that they were using LTI 1.1 or enabling it. I couldn't tell how much. And I was thinking, well, maybe this is just a couple of rogue apps. And then I went and found this site on the Canvas or the Instruction website. That's basically the product directory that they had. That was from there an acquisition they did from a company called Learn Platform that basically showed that the number of apps they had in their directory favored LTI 1.1 to LTI 1.33 to a ratio of 4 to 3. 4 to 3. So there were basically like 6 or 700 apps that they were claiming integrated with Canvas through LTI 1.1 versus you know, a left. So, so this is really disturbing to me. And, and now there becomes an issue not just with Canvas, but with one edtech because one edtech is a member driven organization where the members have rules, their standards and they have to implement those standards. They should not be saying, well, we're LTI 1.3 certified. And then a customer comes and buys it goes, well, that's great. I got the latest lti. Well, not really. They were actually still doing integrations with LTI 1.1. And at the same token, OneEdTech has this trusted apps seal that it provides a certification on data privacy. It's kind of a long story. It's actually really great work in terms of helping institutions get a leg up on app vetting. But the problem is, again, all that, you see, all that's being marketed is this little badge that they get that says this is a trusted app. And obviously Canvas is not a trusted app. And so I was at this point, so basically I'm saying, well, we got to do something about this, guys. This is our chance to shine. Because the whole One EdTech system around certification was designed very, very carefully to ensure that one edtech could call these, these suppliers, held them accountable and say, hey, you know, something's not right here, so we need to dig in. Maybe it's a problem with the standard, maybe the standard is not correct, maybe we need to fix it. Maybe it's a problem with your product that we didn't catch, in which case we have to improve, blah, blah, blah, something about the standard, something, something about our testing process. Right. And so forth. So that was, that's my call to action. And I, I really think that, I'm really trying to say, you know, that the whole reason we were able to grow IMs from, you know, 50 members to 1200 members and all that and really get interoperability working is because we were not complacent with the status quo. We said to ourselves, we have to do this. We said, we took the attitude, it was a very small team, but we took this attitude of we've got all these standards, we've got very little interoperability, we have to do better and we can do better.
- [87:04]
A
Better.
- [87:04]
C
We can do better. And that's, to me, that's the attitude that the edtech sector needs to take about privacy and security. I think we should be doing what we need to do to become leaders in privacy security. And I think the setup for the trusted apps program and so forth is there, but I don't see the current, unfortunately I don't see the current oneed tech management team following through on this again. I don't see any other certification providers following through on it either, to be honest. So again, I'm sort of picking on one edtech because I know them but they're really sort of doing the normal thing right now that most people do that offer certifications. And I'm not going to name the other ones because I don't want to throw them under the bus. But they are there. They are there and this is the time to shine and remediate and figure out what's going on and to improve accountability.
- [87:53]
A
It's a great call to action. And you know, so last question for you. You know, there's been a lot of cybersecurity issues rising in the edtech and the education sector sector. School districts have been under attack by major cybersecurity hackers for a number of years now. And it makes me wonder about fields like health and fields like finance where you have lots of very privileged personally identifiable information, information you wouldn't want leaked. And there's a whole world of cybersecurity around health information and around financial information. It's very, very strict. And I feel like education is a little bit behind and people maybe don't realize that a place like Canvas, I mean, I think it was actually pretty ins of this hacker group to realize that Instructure Canvas by being one of basically three major LMSs, one of which is Google. So they have, you know, world class security. But PowerSchool and Instructure run a huge amount of the K12 LMS market and they desire to learn. There's a few more in higher ed as well. Like they have records akin to what you'd get if you hacked a major hospital chain. I mean you're talking about millions of people's Pisces, millions of grades, information about so many different aspects of, of life for, for families and for schools and for teachers and for universities and for workplaces. Cannabis is also used in some workplaces. So it's like it's a target that is a sophisticated target, but it's not as sophisticated in its defenses as bank of America or you know, as Kaiser Permanente. And I'm wondering, you know, would you feel like that that's an apt comparison. Is that how edtech has to start thinking about it?
- [89:26]
C
So itself it is if we want to continue to do business with, you know, up and coming, smaller, more innovative companies that, you know, edtech is not a big market unfortunately, you know, relative to Some of these other markets in terms of what can happen and you know, the money that these organizations can make, but they have to make these investments. I mean, the good news is there are better and better products out there to help defend against these threats. Again, again, not in the edtech sector, specifically, not in the finance sector or the health sector. You know, CrowdStrike was mentioned as coming, and CrowdStrike has great products I don't want to market or these other products. But Palo Alto Networks, there's a whole emerging field and there's specifically work being done to prevent AI, you know, threats, because. Threats, because AI is going to increase the amount of threats. So I really think that we have to take this more seriously in EdTech and we have to understand that we can do better and we should do better. And we need to get together a sufficient set of highly motivated parties who really want to make a difference here. And, you know, I keep thinking it's like educause at one level is an organization that has the kind of actual expertise in cybersecurity. That's where the chief information security officers get together and so forth. And one, EdTech has the capability of actually holding suppliers accountable through certifications and so forth. But if you are going to call something a trusted app, boy, it better be trust. It better be trustable. That's all I can say. I'm telling you, when we started that program, that is the seriousness that we had in it. And I think if you take that attitude, you can run these issues to ground and you can actually build up a stronger security and privacy posture than exists in any of these other markets, whether it's healthcare, what have you? Because they don't. So, for instance, going forward, right, why can't we, you know, have a performance penalty for downtime for data loss? Why don't we voluntarily. Why doesn't somebody. And actually, all it would take is one supplier to come forward and say, you know what? We believe in our product so much that we're going to pay you if we have a problem like this. Because you know what? You can say, well, that's ridiculous. And I'm a businessman, right? It's not ridiculous. It actually will make that supplier better. It'll make the whole industry better. Because people are raising the bar and they're going to figure out how to do this. They're going to figure out how to do what they need to do to get ahead of these threats and to make the education sector more secure. I think the knowledge is there. It's just a question of getting the motivated parties together and having actual motivation to want to make changes.
- [92:12]
D
Yeah.
- [92:13]
C
And that's basically what my writings were about. And I went deep into some of those areas and if people want to, you know, read it further, be happy to. The articles cat over 1100 reads already on LinkedIn, which is a lot, you know, for, to me, I mean for within this space. Right. You know, within the SedTech space. So people are reading it and they're thinking about it. But to me, I'm, I'm an action orientation oriented guy and it's when you get down in the trenches and you see what actually happened and you see what's actually going on with these certifications and so forth, that's when the institutions can start to say to one EdTech or any other provider, well great, you have that certification. What does it really mean? Does it really mean I can trust that app? You know, does it really mean whatever that I'm getting LTI 1.3 or am I getting something else? That's the level of seriousness that I would like to see happen in this conversation. And after that I'll be a very happy camper because I'm retired and I'm not going to actually do anything about it. But that's the kind of leadership that we need right now.
- [93:11]
A
It's a call to action. I, I definitely recommend all the listeners check out Rob's article on LinkedIn and what he's been writing about this. I think it's really insightful and as well as Phil on EdTech, Phil Hill is, you know, legendary LMS. I think he's the best in the world about LMS report and period. So I always have to follow him. This has been really insightful and I think it's a great call to action for the whole industry to take this stuff much more seriously. If the instructor breach wasn't enough to get people to take it more seriously, I think thinking about it through this lens of more than half of the app that the, the apps in the Canvas ecosystem were using an outdated insecure version of lti. I mean that's, that should really be a wake up call and it should be something people are really paying attention to, including, I'm sure many of the listeners here are hopefully checking their canvas integrations right now and saying, are we one of those apps that is set up on LTI 1.1? I didn't realize it was so insecure. Thank you so much for being here with us on EdTech Insiders. Rob Abel was the CEO of One EdTech, formerly IMS Global for 18 years and he's now with EdTech Futures. Thanks so much for being here with us on it.
- [94:13]
C
My pleasure. Thanks Al.
- [94:15]
A
Thanks for listening to this episode of EdTech is Insiders. If you like the podcast, remember to rate it and share it with others in the EdTech community. For those who want even more EdTech Insider, subscribe to the free EdTech Insiders newsletter on substack. Tuck Advisors was founded by entrepreneurs who built and sold their own companies. Frustrated by other M and A firms they created, the one they wish they could have hired but couldn't find one who understands what matters to founders and whose North Star KPI is the percentage of deals closed. If you're thinking of selling your edtech company or buying one, contact Tuck Advisors now. Tuck Advisors was founded by entrepreneurs who built and sold their own companies. Frustrated by other M and A firms, they created, the one they wished they could have hired. Hired but couldn't find one who understands what matters to founders and whose North Star KPI is the percentage of deals closed. If you're thinking of selling your edtech company or buying one, contact Tuck Advisors now.