Bloomberg Audio Studios, Podcasts, radio news.
This is Bloomberg Business Week with Carol Messer and Tim Steneveek on Bloomberg Radio.
Tech.
Good children, Well if others have.
Well, our next guest over the last fifteen years or so has helped countless students who needed a bit of assistance outside of the classroom with math, economics, physics, history and more. Maybe it was for a bit of help with ap calculus in high school, as it was for our producer Elizabeth Cedrin, or for me help with understanding accounting principles in business school.
I love it. Everybody has tapped into it.
We just had a colleague talk to us about what he is doing and tapping into it. And it's not someone who's like ten or twelve or fifteen.
No, he's like downloading the algebra one.
Right to resprect me cool thinking about it all right. Selcon a familiar name and voice to so many people out there.
Who've seen his YouTube videos.
He is the founder and see of the nonprofit con Academy as well as the con Lab School. He's got a new book out, Brave New Words. How AI will revolutionize education? Why that's a good thing? So it is so good to have you here with us, and you know, it's funny you just we've been talking a lot about artificial intelligence as you as you can imagine. First of all, how are you tell us about your world and how things have been?
Well, thanks for having me. I'm great. Things are busy, but busy in a good way. It's exciting, you know. Obviously the book just got released, But more of what the book's about is we're living in a science fiction book these days. Con Academy, a lot of folks know. I s always great to hear that even people in your own team are benefiting from it or have benefited
from it. But it all started with me tutoring my cousin, and everything that we've been doing over the last twenty years since then has been trying to try to scale up that type of personalization and what we're now seeing
with generative AI. Even though there's a lot of issues, a lot of risks, a lot of things to work through, and I try to address that in the book, there's a lot of optimism, or at least I have a lot of hope that it can can do some really positive things, getting us that much closer to giving world class education to everyone.
Well, it's so interesting, Saul, because when we talk about AI in the classroom, I think a year ago we were talking more about this, this idea that people were so professors, teachers were so concerned about chet GPT, for example, being used to write term papers or write essays, and you know, having the software to detect whether or not a GPT wrote an essay the question about having phones
in classrooms. Like, the idea that technology has helped the classroom is there, but it's also been that technology has hurt the education experience. That's a big part of the conversation right now. How do those two things conflict with one another? How do you see them?
The reality is technology is always a it's neutral. It can be positive or negative. You can think of a very simple technology like a knife. It can be used to hurt people, It could be used to cook, it could be used to build things. And we're seeing the same thing with cell phones and with AI. Yes, and the cheating is a real issue and people are still scrambling and you know, I have a whole chapter in
the book about cheating and the state of cheating. Even before AI was actually pretty bad, AI just really put a spotlight on it. Cell Phones, social media obviously addict students, especially young people in a certain way, and most of the literature it's not good for them. But even before AI, there's really positive things you can do with technology. You
could learn on con academy. You can write a paper, you can code, you could edit video, and most of us, I think, as long as done in moderation, would be happy if our students spend some time a few hours a week doing that type of thing. And so the key is how do you get how do you use technology to amplify positive intent and then how do you
also use it to mitigate the negative intent? We're building and I write about in the book where we're building tools that not only support students better in their writing. Won't do it for them, It'll tutor them, it'll coach them, but it can make that process transparent for teachers. And by doing that, it actually undermines cheating, whether it's AI cheating or other.
How does it though, Like I think one of the things the conversations we've been having TOIL is about kind of this idea of does it take some of our thinking away from us?
Like you know, sometimes you got to struggle.
Through a concept or a thought or a mathematical problem. Getting an assist can be helpful, but sometimes you got to sit with it for a little bit. And I do wonder about technology taking away some of that work that.
We use our brain for. And I'm just a little nervous about that.
It's a valid fear. And you know what I believe is one in some cases you do want the shortcut when you're working, et cetera. But in most cases, especially for since we want them to develop those skills. And that's why everything we've been working on con Migo, which is our AI assistant on con Academy, it doesn't answer the question. It will push you socratically. So the other day I was trying to get it to explain supernova's at a deeper level. I know what they are, but
exactly why does it explode? Why doesn't it just collapse? And its response wasn't it just give me a Wikipedia answer or like the type that you get on chat GPD. It's response was, well, can you explain to me what you already know? And then it kept pushing my thinking, Well, okay, if that leads to this, well, what do you think what happened there, and so it was a much more
socratic conversation that drives critical thinking. And so there's an opportunity to actually scale that type of learning that's been very hard in the past.
What what is your AI tool like? Where what data are you pulling on for it?
So we are using the underlying frontier models, whether it's you know now GBT four, Turbau and on me, et cetera. But then on top of that, there's a lot of work to make it really special purpose for education. We're doing a lot of what's called in the trade now prompt well, prompt engineering, prompt chaining, multiple calls. It's anchored on con academy content that's been vetted by professionals for many years. It can slurp that stuff in as well.
So we're putting a lot of the pieces together. We're not just you know, relying on the raw model, which I don't think is a responsible way to do it for an education use case.
It's still it is a brave new world that we're living in. Your book's called Brave New Words. Yesterday saw I don't know if you saw this, but a lot of people were posting on x about errors that Google's own AI search tool was making. My brother sent me one about eating rocks, and it actually had cited an
Onion article that was joking about eating rocks. But you know, if you looked at eating rocks on the internet, it said like it's recommended to eat rocks every day, and then you look at the source and it's the oniony, So like, this is not you know, this is by no means like an easy thing to figure out at this point.
So no, not not at all. And obviously I can't I can't defend everything out there, but regardless, obviously the providers are going to work harder to make sure that things like that don't happen. But it's part of digital literacy now for users to recognize that every answer you're going to get from an AI, you're gonna have to validate that, You're gonna have to put that in context, and not all AIS are considered equal. And this isn't
a new phenomenon. Well before generative AI, you could do a web search on Google and very click on something that's complete misinformation, or see a video on YouTube that's complete misinformation. So we've always had to try to give people that scale of how do they judge, how can they validate? I think that's true in an AI world, and I read a lot about it in the book.
There are ways, and we are building these ways that the AI can not only not do those things or can help you validate what's real.
Listen, you are someone who was so at the frontier in terms of figuring out a really smart way to use technology to teach kids, and they still do it. I mean you just think about what you did way back when, right when it was kind of I hate to say, like almost simpler times.
She doesn't hate YouTube, right, it's very early.
It's a while, but it has helped so many and continues to do so.
So, so you like.
Not glass half full with AI and this, Like you think it's going to be okay, You think it's going to be leaps and bounds in terms of teaching folks like I think about tutors. But anyway, like you are all in and you feel comfortable about this.
Well, what I've told everyone. You know a lot of people are reading the book. I'm getting interviewed together like, hey, you tend to be more optimistic than most people. I was like, no, look, I'm afraid of all the same stuff that a lot of people are saying. Bad people are going to use it to amplify bad intent. We're going to see deep fakes, we're going to see fraud, We're going to see state actors try to manipulate people
in bad ways. That's going to happen. But that doesn't mean that the good people should just freeze and do nothing. We should use these tools to amplify positive intent. And so what I'm telling everyone, whether AI is a net positive or net negative, isn't like a flip of the coin. We're not just going to like sit and watch. Good people should go out there and make the positive use
cases happen. We shouldn't just talk about it. And they exist, and we are already seeing that in education, you know, just earlier this week, we made a big announcement with Microsoft. We're making these teacher tools that are going to save teachers time doing things like lesson planning and grading papers and writing progress supports, saving teachers hours a week. We're making that free, even though it costs a lot of money for the computation, making that free to all us teachers.
So there's a lot of good that's already happening, and so I'm optimistic if good people put their energy and the resources behind it, we're going to get a net benefit from the technology.
As Carol mentioned, you harnessed tech early in your career two thousand and eight for YouTube. Is AI, in your opinion, more powerful than that?
Oh yeah, I think you know, even if we just even if they froze AI today, there's ten years of amazing things we could create with it. But the reality is the pace of innovations only accelerating. So yes, what we're going to be able to do in two, three, four or five years is going to be mind blowing, unbelievable.
I want to get right back to Salcon, a familiar name to so many people. He's the founder and CEO of the nonprofit con Academy as well as the con Lap School. He's got a new book out. It's called Brave New Words, How AI will revolutionize education and why that's a good thing. Sal We're going to get back
to the book in just a second. But as we were on our planning call this morning, we were talking about con Academy and we're all sharing our experiences with it, and it feels like with you because those early videos were really you and you you only teaching us so much, and we're asking ourselves, is it a nonprofit still? And you know, went up and looked, and it still is
a nonprofit. And I'm wondering talk to us a little about the structure of the organization and why a nonprofit is the right model for you.
Yeah, and it's always good to hear those stories. And I still make a lot of videos. I made a couple this morning.
That's amazing.
But you know, there was there's three Harvard Business Schools case studies on con Academy, and the question that it normally boils down to is should kind Academy be a for profit or nonprofit? We clearly have a social mission, but we have some of the behaviors or the properties of a high growth for profit. We're very tech focused. We're out here in Silicon Valley, we scale, et cetera, et cetera. You know, for me, it was honestly an emotional decision. When I incorporated con Academy back in two
thousand and eight. There were some vcs who were interested, and I'm not against for profit. My previous life, I was a hedge fund analyst, which was very for profit, but it felt like education and it still feels that way is a space where maybe the markets don't always lead to outcomes that are optimal or are outcomes that
are aligned with our values. And there were folks even back in two thousand and eight who were using the content, who were telling me how it was changing their life, and I was like, well, I'm not sure if they would be able to access it if they didn't have it. And as a hedge fun analyst, I also saw how you could have a founder who really is mission driven of a for profit, but you fast forward fifty years, one hundred years, the capital structure changes, the incentives change,
and I didn't want that to happen. So I did a kind of delusional thing. I was in this walk in closet when I filled out the paperwork with the RS, I said, what if Kin Academy could be like the next Oxford or the next Smithsonian. Even at that time, we had actually served more students than Harvard had in its history, even when I was just one guy in a walk in closet, And I said, what if we
could reach billions? And so it was a little bit of a delusional optimism, but you know, it feels less delusional now really interesting.
I do want to go back to the book, if we may, your book Brave New Words, how AI will revolutionize education and why that's a good thing. So where do you think AI will have the biggest.
Impact on education? Is it the approach? Is it a certain area? I'm just curious.
You know, the technology is so powerful. I think every hour that you spend brainstorming, you realize it'll have more and more implications. I mean in any industry, especially education. You know when open Ai first reached out to us in summer of twenty twenty two, and this was six months before Chat GPT came out and we were able to get early access to GPT four. Many folks know that even the first chat GPT wasn't built on GPT four, was built on a previous model. We immediately gravitated to, hey,
how can this be a tutor for any student? Really do the types of things that I was doing with my cousins twenty years ago and these frontier models now like GPT four, Gemini, et cetera, they really can emulate what a tutor could do via chat. And many folks have probably seen the demo that for Omni GPT four Omni that came out last week. It's me and my son.
They brought us in to see if we can do some fun stuff with it, and you can talk to it in a natural way, and it could see your writing and it can see your face, and it really felt like a live tutoring session. That's all coming. But then there's the other side of the equation. There's what you could do with teachers, helping them write lesson plans, helping be a co teacher for that teacher, be a
teaching assistant, helping them grade papers. That's already happening. We already getting feedback from school districts using Conmego that their teachers are saving five to ten hours a week. That's a huge benefit for teachers who are spread thin or in many cases overworked. So it's a benefit for them. But I see us going five years into the future where it's even able to facilitate interactions between people. A
teacher can say, hey, let's do a breakout. Hey, I says, okay, I'll set up the breakout rooms, and actually I will moderate the conversation in each breakout and I will bring it back to you. An AI and this we have to just be careful that the data privacy and it doesn't get creepy. But I can say, hey, we haven't called Mary in a little while, why don't we call
on her. She looks disengaged or he looks disengaged. Let's use that to personalize the lessons more so it really will feel like a teaching assistant.
Heysel, you mentioned con migo and conmego is featured prominently in your book. Can you explain what it is and how you guys built it on GPT for.
Yeah, well, this is back in almost almost two years ago, eighteen months ago. We saw there's a lot of potential for the raw technology, but we also said, hey, this could be used for cheating. If you're talking about kids, how do you know that they're not trying to do shady things with the AI? How do you protect their data privacy? How do you make sure that it doesn't make errors in math and hallucinate? And so we said there's a lot of work here to take this raw
AI model and make it appropriate for school. So our AI tool set for teachers and students we call con Migo, and it does all of That's what it's doing. It won't cheat, it will work with you socratically. It'll actually help undermine other forms of cheating, even AI based cheating. It makes everything to the teacher. It'll actively notify parents and teachers if students try to do something shady with the AI. We're introducing interfaces. Not everything should be a
chat interface. We're doing. The interface is where a student can write an essay, co create a document. Once again, all of that is transparent to the teacher. The AI can highlight parts of the document and give students feedback on it. We're created activities where it can act as a bit of a career and guidance counselor for a teacher, for a student. Once again, not to replace the human beings in their life, but really to amplify that impact.
Whenever student needs it, it's there, and then when they get with the teacher or their guidance counselor, they can go deeper.
Hey, listen, one thing we talk about a lot salads, you would imagine is college admissions. And you have a chapter towards the end of the book devoted to that. The first line in it says, the classic components considered in college admissions are grades, standardized tests, extracurriculars, essays, and letters of recommendation. AI will change how most, if not all, of these factors are value developed and evaluated. Those are your words, What impact will that have on that process?
It is really across the board. As we know, these things are all about writing. And what we do also know is, you know, there was a Varsity Blue scandal several years ago where we saw these corrupt coaches, admissions coaches that you could hire to almost fabricate a college admissions. But what we know, you know, and especially places like Silicon Valley where I live, New York, a lot of
affluent people, upper middle class people. Well before AI they were paying four hundred dollars they are it's still happening, paying four or five hundred dollars an hour to have these professionals advise their kids on how to maximize their essays, et cetera. Now, the ethical ones are just act as coaches, but it's still a huge advantage, and the less ethical
ones are essentially writing these essays. And you know, I think admissions officers, and I write about this in the book, have they've kind of known about this, but they've kind of intentionally not wanted to address this, the elephant in the room about the inequity and to the fact that the AIS can now do some of this work for all students. It really just puts a spotlight on something that's always there and it makes you wonder whether you can get richer points of data for a student than
a recommendation than an essay. But that's where the AI could be interesting too. Future admissions SAT type things could have free response. You could have to talk things through, you could write it out and then the AI could make sense of that.
So cool.
There's also a book about a chapter about employment in an AI world.
There's so much in this book.
Sal thank you for finding time for us to talk about sol Con. Of course, his new book, Brave New Words,