AMA with Sal Khan on AI + Education

But for now, I want to kick things off with a question that Aaron had asked in the Q&A and got a lot of upvotes, and that was: what role do you envision generative AI having in education beyond just AI-enabled software and apps, as it pertains to the whole of the future educational model?

Oh yeah, that's a big question. Um, there’s so, so I’m actually writing a book about this. It’ll be released in May, called "Brave New Words." Hopefully, you all like the title. I didn’t think of it; my book agent came up with the title.

But I actually think you’re going to see it affect nearly every dimension of our lives in some way, shape, or form. There are the things that many of y’all have already experienced through Kigo, and I can give you a sense of where we think this is going to go.

Obviously, we talk a lot about Kigo being a tutor, Kigo being a teaching assistant. Expect to see more supports for parents to think about how AI can help them out. Um, expect — we’re working on ways that Kigo actually can transcend Khan Academy. So imagine, you know already you can watch a video on Khan Academy and ask Kigo, like, "Hey, why do I need to learn this?" or "Can you quiz me on this?" or "Can you summarize this?"

Imagine if you had Kigo with you wherever you went on the internet. Imagine a world — and you know, this is all stuff we’re working on — where a teacher or a parent could assign things through Kigo. They can say, "Hey, you know, right now there’s activity where you could talk to historical characters," but what if a teacher said, "I want all my students to talk to Marie Curie and try to see if they can identify this mystery element?"

And then Kigo works with the students on it and then is able to report back to the teachers on how the students did in that simulation. So I definitely see all of that. I see — hopefully helping teachers’ lives with lesson planning, grading.

Um, I think it’s, especially for things like grading, it’s always going to be important to have a teacher in the loop, but the AI could help a lot. Progress report writing — expect to see more functionality for parents where not only can parents use Kigo to monitor what students are doing, but also, you know, get help, get advice as a parent.

I’m a parent like, "How come my kids aren’t going to bed on time?" or "How come they don’t seem to listen to me?" or, you know, whatever it might be.

Um, so those are, I would say, the near end. And of course, from a technological point of view, I think we’re only not too far away. You're going to have conversations with the AI. You’ll even be able to have video conferences with the AI.

Things like this I think — you fast forward even more. We have another sister non-profit that we started called Schoolhouse.com. We’re able to give free tutoring because the tutors are volunteers. We have a whole mechanism of vetting them, but we’re already using AI to give those tutors feedback on how they could make those tutoring sessions better.

So I could imagine in the education space — and we have to be careful here, we don’t want it to feel like Big Brother is watching every educator — that's not the spirit of this at all. But if they can help an educator, maybe even potentially while they’re teaching, say, “Hey, by the way, a bunch of your students just had questions on this topic. I answered it for them, but you might want to double-click on that."

Or, "Hey, you want me to break them up into breakout groups, and then I can help moderate?" The AI could help moderate each of the breakout groups. So I think you’re going to see a lot more AI facilitation between human beings in the real world, which I think is very, very exciting.

Um, and it’s not going to be limited to text interfaces. I already talked about speech, vision. There’s a lot of places it can go. I, you know, we can daydream in 10 years in the future what it might do with things like virtual reality, etc.

But, yeah, I think the world of assessment is going to change dramatically because of AI. Right now, standardized assessment is fairly narrow in what it can do. It’s historically either you type in a number or you have multiple-choice because it’s hard to do anything else at scale.

But imagine a world in five years where you could go through a simulation. You could have a conversation, go through a Socratic dialogue with an AI. And at that point, maybe the AI can be pretty good at benchmarking in a standardized way. We have to be very careful about things like bias, etc. But I think there are ways that we can get there.

I can keep going, but I’ll stop there so we can get more questions in.

That’s awesome! That’s a great way to kick things off. You know, we’d love to hear a little bit more about you. So what I would love for you to let us know is: how are you using Kigo so far? So come over to that poll.

You know, obviously, there’s so much you can do with it. There’s the tutoring side, the creative writing side, even college applications. Go ahead and take that poll on the screen right now! Let us know how you’re using the tool today because we’re always curious to make it better and even more effective for all your use cases.

So folks using it for tutoring, of course, teaching college applications. A lot of teachers on the line. So that's awesome!

Um, okay, great! Well, thank you for sharing that. I’ll share those results with you so you can look at that, and then let’s go to a question from, um, from Jorge. So Jorge says: Do you think the general level of ability for the workforce en masse will gradually need to improve because of the advent of AI? There are lots of interesting questions about AI, the workforce, the labor market. So curious to get your thoughts here.

You know, I sure hope so. You know, there's been some folks in the AI debate who have said, "Oh, generative AI can already do this. It can write a paper. Can I guess no one has to learn that stuff anymore?" And I’m like, “No, it’s the exact opposite."

Because I think we do recognize that this will be able to help us write software or code or create art, but — and it can elevate someone who might have been a staff-level writer to being an editor because they can now have staff writers of AIs. Or someone who was going to be an entry-level coder could now be a software architect.

But the reality is, you can’t be a great editor-in-chief if you can’t write well and if you can’t manage and give feedback to your writers, which in this case, some of the writers might be artificial intelligence.

So I think the imperative for folks to be able to keep their skill competencies up with the AI is actually really, really important. You know, as there — there’s these frames going on like you’re not going to get replaced with an AI, but you might get replaced with someone who is leveraging an AI, who’s using it as an extension of themselves.

And I think the best way to leverage AI is if you’re able to go up the skill chain as opposed to deskilling yourself.

So I think, you know, I talked a lot about this even before AI really came on the scene, but I think it’s even more imperative. We historically had an industrial-age labor pyramid where at the bottom you have a lot of labor, in the middle you have, I guess you could say, the information processing class, white-collar jobs, etc. And at the top of the pyramid you had researchers, professors, entrepreneurs, doctors, lawyers, you know, teachers — all like the highly skilled folks at the top.

We know what’s happening: AI and robotics, to some degree, are going to collapse those bottom two layers. At the same time, we’re going to have a massively more productive society, and so we have a social question: do we want all of that increased productivity to only accrue to, you know, I don’t know, 10% of the population?

That’s not a stable situation. That’s not a happy situation. That’s a dystopian situation. And you might have some type of redistribution, but I don’t think most people want to just get a check. They want to be able to participate and have a sense of purpose and a sense of meaning.

And I think the only way that you’re going to do that in this AI inflection point is by moving as many people to the top of that labor pyramid as possible. And it might sound optimistic or even delusional, but a couple of things: one, the tools are getting better and better and better, and AI is going to be able to help us do that.

The other thing is I always point out: if you go back 500 years in time, 20-30% was about how many people could read, even in fairly literate parts of the world. And topics like algebra or geometry were considered very fancy, esoteric topics.

Um, and you fast forward today, we expect pretty much everyone to learn how to read by the time they’re, say, seven or eight years old, and we expect — you know, unfortunately, it’s not happening as much as we would like, but we expect everyone to understand something like algebra at some point.

And so, the fact that we’ve already moved society this much forward — I think, if I were to say how many people are under-capable of understanding quantum physics or writing the next great novel or starting the next major company right now, folks might say, “Oh, 5-10% max.”

Well, what if through education we could make that 50% or 80% of people capable of it? I actually think we have to try. With that said, I think there’s always going to be — I think the jobs that are very human-centric are going to be very, very safe in this AI world.

You know, I’ve said it multiple times: I think teaching is actually going to be one of the safest jobs, and I’d like to imagine it’s going to become even more valuable as we go into this world because the AI is going to be able to take on a lot of the minutiae of it and a lot of the coordination.

But the value of a student being able to feel like that caring, thoughtful adult who also has content mastery, is helping them, coaching them, forming a connection with them, is huge. I was just reading a study about the value of even having the same teacher for two years in a row, right?

Um, you know, so that to me says it all — that relationship aspect. And I think there are going to be other fields where that human aspect of it can only be amplified if you can do other things with AI.

That’s awesome. Yeah, I saw that Adam Grant had called that out in his new book — the value of a teacher over multiple years. Definitely lots of interesting things we’re still learning about the human learning experience.

Okay, speaking of that, Emily’s got a really juicy question. I’m just blown away by these questions! These questions contain entire essays within them. Emily wants to know: how do you think AI should be talked about within the general public? It feels like only those seeking the information out for themselves are getting a real feel for the possibilities while news outlets focus on doomsday scenarios or fluff pieces. What do you think is the right way for us as a society to be talking about this new technology?

Both — you know, I think at a high level I have a couple of doomsday fears. Um, they’re mostly things that are out of my direct control. Uh, but I do try — if I meet politicians or people who have — I try to tell them about it.

I am worried about what a totalitarian government might be able to do with generative AI. Right now, they can already listen to all the phone lines, but it’s too much information for them to make sense of. Now they’re going to be able to use generative AI to listen on the phone lines, and then if anyone says anything that they don’t agree with, you know, they can do bad things. That I’m very worried about.

I’m worried about deep fakes, both for manipulation. You could imagine all already — this could happen; I hope it doesn’t — but someone could create videos of people waiting in line outside of their bank to get their deposits. And, you know, as they say, what is it, the — you know, a lie can go around the world twice while the truth is still putting its shoes on.

Um, and so you could create a run on the bank with something as simple as that. But then there are other forms of deep fakes that can manipulate politics, etc. I’m very worried about that. Obviously for crime, phishing attacks with the AI, where it pretends to be some family member who needs money — super so.

Those are, that’s the space that I’m really, really worried about. I’m for the most part very optimistic about what’s happening in education. I know, honestly, the biggest risk is cheating in education. I point out — I have a chapter about it in my book. Cheating was already there before generative AI. You know, there are sites that will write a student’s essay for $5 a page and all of that type of thing.

And I actually think generative AI is going to mitigate that. Uh, this is something we’re working on with Kigo. I hope we’ll have this in place in the next year or so. We’re already creating essay tools, not to write your essay for you, but you could put your essay in, and it’ll highlight parts of it and give you feedback. It’s like having a writing coach where it can give you feedback on how you tighten your argument, etc., etc.

In the next year, we hope to have it so that a teacher — and maybe even a parent — can make an assignment, a writing assignment, through Kigo. Kigo is able to work to partner with the student. The student does all the work, but Kigo coaches the student, and then when the students ready to submit their essay, Kigo can tell the teacher, "Oh, well, first of all, here’s the essay. But second of all, we’ve worked on it for about four or five hours. We had trouble coming up with a thesis statement, but we finally got there."

Um, you know, and by the way, based on the rubric, I would give it this score, and if a student went to ChatGPT and just copied and pasted that essay in, or if they went to, um, if their sister wrote their essay for them, or if they went to one of these essay mills on the internet and got it written by someone in, you know, Nigeria — which is apparently where a lot of these essays come from; there’s a lot of good writers in Nigeria — um, then Kigo will say, "I don’t know where this essay came from. It just kind of showed up. It looks shady. And by the way, it’s not consistent with the student’s writing that I’ve been observing in their classroom."

So not only can AI support students better, but I actually think it can mitigate some of the cheating that has been there for a very, very long time. So I think there’s going to be — and if we can save teachers time with lesson planning, and eventually writing things like progress reports and IEPs, and etc., and these are things we’re working on right now, expect to see more in the coming weeks or months.

Um, it’s going to be, it’s going to be real positive. What I would tell just anyone — I tell this to everyone in my family, I’ve told this to young people out there looking for jobs — it's first of all, no one should stress. It’s still very early days. I mean, just the fact that y’all are here or that we’re even talking about AI means that you’re an early adopter. You’re in the, you know, top 5% of people, even though it sometimes doesn’t feel that way.

Uh, but encourage people to just start using it. I’m encouraging everyone at Khan to just use it in your day-to-day. And maybe it’ll slow you down a little bit, you know, but as you, as you get used to it, you’ll start finding ways that it really enhances what you do and discover new applications.

So, you know, that’s — I’ve encouraged young people to say like, "Hey, create a prompt for generative AI that kind of can represent you, almost can be an interactive resume that’ll impress employers." Like, and you’ll, you’ll — you know, so things like an interactive cover letter — imagine instead of a cover letter just being a statement; people can ask it questions and it represents you.

So there’s — I encourage people to just start playing around with it, seeing if they can find it useful. If you’re planning a vacation, use generative AI for that. On Kigo, we have this thing called, for over 18 users, we have Chat with GPT-4.

Um, that’s general; you can do anything you want there, and then we have all of the more special-purpose things that really try to fine-tune some of the various use cases.

Yeah, and just to show sort — what A is talking about here, it’s right there: Chat with GPT-4, and definitely be creative. I think A is absolutely right: if you’re getting your hands dirty, you’re learning. So that’s the best way. And give us feedback!

You know, one of the reasons why we even put that there is, one, we wanted to make that kind of one-stop shopping. You shouldn’t have to go anywhere else; Kigo hopefully will have everything you need. But either if y’all give us feedback or if we see, "Wow, a lot of people are using this to say, I’ll use simple planification"— that’s maybe not core to education — but they’re using it for this use case for running a progress report, etc., etc., then that’s a big signal for us of like, "Okay, just raw GPT-4 is okay for doing like that, but let’s really engineer that and make it really nice, kind of like what we’ve done with lesson plans and things like that."

This is a great segue to our next question, which must be from an English teacher actually, because this question asker is worried about ChatGPT, Kigo, all these LLMs actually eroding original composition skills — the ability to start writing from scratch without that AI scaffold. Is that something we should be worried about?

Yes and no. Um, I agree that there is something powerful in a lot of creative fields of being able to not face a blank sheet, whether it’s writing, whether it’s coding, whether it’s painting, or composing music. That, you know — or composing music, that blank sheet can be very intimidating.

Um, but it can also be paralyzing for a lot of folks. I think there’s a, you know, what we are envisioning when teachers assign, say, a writing assignment through Kigo, the teacher can decide how much support the students get.

Um, and we would never make it so it writes it for the student, but I think that could be almost like a, you know, training wheels so that students can feel the joy of writing and the productivity of writing and the expression of writing without that cold start problem, that cold start fear.

But at some point, you can start taking the training wheels away so that they really do feel like they, you know, they can — they don’t need something there to talk about what your thesis statement is going to be, or let’s outline this thing together, etc., etc.

Yeah, it seems like if you think back to elementary school, ourselves, every kid learns with a scaffold — you know, write your introductory sentence, write your, you know, next sentence, your evidence. So in some ways, it’s just sort of speeding up that process. Exactly, giving a much — right, you’re exactly right. I mean, I’m looking at what my nine-year-old is doing. That’s exactly it; it’s very scaffolded.

And if we can get that all the time with the AI, even better. Here’s a really good question: this dives into that CW cheating again. Aon wants to know, is there really any point in teachers trying to catch cheating?

You know, there’s this program, originality.ai, trying to market itself to teachers to detect AI-generated text. Should educators be spending energy on this?

I think right now, um, I think there are two problems: anyone who tells you that they can detect AI-generated text is kind of not being genuine with you.

Um, it can detect certain aspects of AI-generated text. The problem is, and we’ve been hearing a lot — I’m sure many people here — there’s been a lot of false positives, and that’s a pretty horrible scenario to be in. You’re accusing a student of cheating when they didn’t cheat.

Um, because this isn’t plagiarism. So I think that’s not the way to address the cheating issue. My opinion is that the way to address the cheating issue is what I just described: have Kigo work with you on the essay, and then transparently report back to the teacher what it did.

And by the way, here’s the whole transcript. And if a student goes to ChatGPT or their sister and gets it written, Kigo just says, "I don’t know where this essay came from. It’s not consistent with the writing that is a much tighter way, um, and fairer way of measuring potential cheating than having some kind of algorithm that has a lot of false positives, and you’re going on to end up falsely accusing students."

Great. Um, here’s a juicy question for you. Um, someone is saying, as a father yourself, what advice are you giving your kids about AI today that maybe other parents could learn from in terms of how to guide their own children on this journey?

Yeah, and it’s once again, it’s still very, very early days. You know, it’s not like social media where a lot of kids are hooked on it. And I have much stronger opinions about social media: keep your kids off of it as long as possible or heavily regulate it.

Um, the AI, you know, we aren’t yet seeing kids addicted to AI and hopefully we never see them addicted — or if they are, it’s in a good way. You know, if you’re addicted to Khan Academy, within bounds, it can be a good thing.

Obviously, you still go run outside and have friends and all of that, but if you’re learning, that could be a good thing. Um, I think — so I don’t think this is a time where, let’s say, a student’s using Kigo, and they’re able to have interesting conversations with Kigo, etc., etc.

As far as we can tell so far, these are very healthy things. Um, and you know, my kids — there’s variation; some of my kids love doing their regular Khan Academy, and some of them not as much, and I have to force it. But then Kigo becomes a little bit of a treat. For example, my nine-year-old, it’s like a treat: "Okay, if you do your Khan Academy math for a bit, then you can write a story with Kigo," which he loves to do.

And I know it’s good for him because actually, when he’s doing that with Kigo, he’s reading more too. Um, and he’s — so, so I haven’t — in safeguarded sandboxes like we are trying to build, I haven’t seen the downside yet of students just being allowed to do what they want to do as long as there’s some guardrails.

I’ll also say, um, you know, one of the things that we — I think AI — I know I’m sounding very optimistic. I’m not about everything, the totalitarian governments, crime organizations, etc. But one of the things we are thinking about is Kigo can start to work beyond Khan Academy; essentially, it can be your co-pilot wherever you are on the internet.

As a parent, it could be really powerful from a safety point of view because if it’s always there on any page you go to, you know, you’re on Instagram, you’re on, um, you’re on YouTube, wherever. If one — the AI can make sense of what’s on the page and one decide whether it’s inappropriate or not.

Uh, as you all know, we already have an AI that’s deciding that on Kigo, and it’s flagging to parents and teachers if something’s off. But, you know, so if a student is on YouTube and all of a sudden watches a video that’s inappropriate, it can flag parents, it can even not allow the student to watch the video.

If the student is doing something that seems good — they’re on, you know, they’re reading the news on the New York Times or something — there still might be some stories that aren’t appropriate for them. Maybe it can remove some or soften some of the language or maybe change it to be closer to their reading level or, even better, quiz them on it so to make sure they understood the underlying issues or, you know, any background knowledge they need to understand it.

Um, so I’m optimistic that actually, um — and I — there’s a chapter about this in my book where I actually think thoughtfully we can have AI almost like a bit of a guardian angel for young people while they are on the internet.

But, you know, we’re going to keep close tabs, and definitely let us know if you think some negative trends are emerging amongst young people.

Yeah, great call out! And speaking of young people, we’ve got a college student here who’s asking for a little career advice. Um, he or she wants to know, um, if they want to sort of hitch their career to the AI star, what’s the best way to start doing that? What kinds of skills, what kinds of experiences should they be developing today?

Well, I’ll just double down on the advice I gave about 10 minutes ago. Like, if you want to hitch your wagon to the AI star, start using it. And everything you do — at minimum, if you’re a college student, I’m assuming you’re applying to internships.

You know, money-back guarantee for the advice I’m about to give you, but, um, if you wrote an AI prompt that, you know, people could go to and interact with it, but it represents you, and in your cover letter or resume, you gave a link to that, I think you’re going to get more interviews.

Because so many employers are like thinking through what it — they want to be around people who see — who seem to understand the world that we’re going into or at least are leaning into it and have an open mind and are trying it out.

And even if you don’t do that, if you just start playing around with it and are able to write about that in your resume or your cover letters and say, "Yeah, you know, I’ve been prompting AI to do the following tasks. I pretty much use it to figure out what groceries I should buy," I think a lot of employers are going to take note.

You know, there’s this whole field of prompt engineering that literally didn’t exist a year ago. Um, and it’s — it pays well. Um, and, uh, but it is a field that you have to be ready to keep adapting very, very fast. It’s not like you can learn some skills and just apply those same skills for the next 50 years.

Um, but if you’re the type of person who likes that type of a challenge, yeah, there’s there’s so much here. I’ll just put in a little plug: check out some of our jobs at Khan Academy because we definitely could use some of those skills. We’re hiring. These jobs, that’s right!

Um, great question from Aon about Kigo. Aon wants to know, okay, she’s using it with her own teaching needs today. When can she start using it with her students? How can she cross that Rubicon to get access to her kids?

So right now, uh, you know, we call it Kigo for classrooms. It’s something we’re doing with out a district-by-district level right now, and we’re ramping in fast. Uh, so many districts want to do this. So that’s one option: tell your superintendent, "Hey, I would love to be able to do this," or at least try it out or give us a little — you know, so that’s one.

Um, I think in terms of you being able to do it on your own, uh, you know, Jeremy and I and a few should chat about — I would love to make that possible. Um, but those are kind of the two levers right now. But I don’t know, Jeremy, anything to add to that?

Yeah, it’s definitely one of these challenges of we’re trying to fight against this cost that we have to pay to Open AI into all of our, you know, AI resources to make it as accessible as possible. And so we’re going to be announcing some big news in just a couple of weeks, so stay tuned for definitely a lot of big things coming up at the end of this year and into the start of the New Year.

Yes, as you can imagine, our mission is free, world-class education for anyone anywhere, and everyone here knows that Kigo is not currently free because of what Jeremy just mentioned. But a little foreshadowing for the big news is we are — it’s not going to be free just yet, but we are working feverishly to reduce the cost so that we can make this as accessible as possible. Aon says we make a great test subject, so Aon, we’ll have to follow up over after this.

Yeah, I mean, look, if you can really make the case, we might — anyway, talk to Jeremy. That's right, that's right!

Okay, here’s — this isn’t a question; it’s a little anecdote. Anonymous attendee says, "I ran something I had Kigo generate for me through an AI detector, like we were talking about before, and it came back as probably human. Something I wrote back came as AI-generated. I told Kigo about this, and it found it amusing." So there you go; Kigo has a sense of humor.

Well, I mean, that’s right there. Um, you know, it’s consistent with what I just said, which is just like these AI detectors are snake oil as far as I’m concerned.

Um, and they’re false positives, false negatives. Um, the reason why I probably didn’t detect Kigo’s is we’ve worked, as many of you all know, we’ve worked hard to give Kigo a human-like tone, um, which most bots do not have.

Uh, but in some ways, I guess it’s flattering that the AI detectors thought it was human.

Yeah, and then maybe one last question to end on. This is sort of like the capstone as we talked a little bit about the future of AI and education, all that. But like, what are your hopes and dreams with this thing? Like when you get excited about it — it’s not just the algorithm — like what do you want to see out there in the world through this technology?

Oh, um, you know, a lot of folks here know Khan Academy started with me tutoring my cousins as a pastime. I had a day job, and I was tutoring them. Word spread in my family free tutoring is going on, I’m tutoring 10, 15 cousins. I started writing software for them; that was the first Khan Academy.

It had nothing to do with videos, where I could give them practice problems and keep track of what they were doing. The teacher dashboards were actually from the very early days I was using it as my cousins’ tutor. Um, and then a friend suggested I make YouTube videos, and over the last, let’s call it, 10, 15, 20 years, everything that Khan Academy has really been trying to do is scale that kind of personalization that you could get if you could really have one-to-one.

And every teacher on the planet, every ed school teaches differentiation; every teacher aspires to it, but it’s very hard if you have 30 students in a classroom. People talk about — pre-pandemic, you had three grade levels in the average American classroom. Now there are six grade levels of spread in the average American classroom.

It’s very hard for — you could be Superman or superwoman, but to be able to address the needs of all of those different levels. Um, I think at minimum, I hope over the next couple of years — you know, Kigo can already do a lot as many folks already know, but it can take on more and more of that real personalized support when there’s no one else there to support them but connected to the teacher.

I think that’s key. A lot of the money that went into human tutoring after the pandemic — and you know, we as a country spent billions of dollars on — I think it was $86 billion. In hindsight, people say, "Oh, it actually wasn’t that effective because it was not connected to the teacher, it was not connected to their classroom, and it was for the most part logistically difficult for a lot of kids to access it."

And actually, some kids were embarrassed to access it. This, I think, has the opportunity to address all of that.

Uh, it’s available whenever folks want it. It can talk to the parents; it can talk to the teacher. There’s less embarrassment for students using it. You can connect it to the work that the students are actually doing, and I think you’re going to see it do — sorry, my dog is barking a lot.

You’re going to see it do more of what a good tutor would do, where it’s, you know, call up your parents and say, "Hey, this person hasn’t been doing their work lately. We should problem-solve this."

Um, or talk to the student in that way, or give the teacher more support, like, "Hey, here’s an idea: what if we did this with this group of students who seem a little disengaged?" So that’s very exciting.

If we can really accelerate learning and engagement, it’s interesting already. Khan Academy, if a student is engaged, the learning is good. We have 50-plus efficacy studies. So it’s really all about engagement.

Um, so, I’m very optimistic there. Teachers, they have a million things to do. Um, and if we can save some of that lesson planning, grading, rubric creation, etc., etc., time, awesome! Teachers need it for themselves, and that’s also more energy that they might be able to, you know, help students with — with all the ideals we’ve talked about around differentiation, etc.

So I’m excited about that. Um, I’m excited about broader interpretation of assessment. As we know, standardized assessments have been narrow in a lot of places. The curriculum became narrow because that’s the only thing you can measure.

I think that’s going to change as we can start to assess broader things but still do it in potentially standardized ways. I think you’re going to have a world where the space for creativity, like we’ve always talked about project-based learning, but it’s always been difficult.

Um, especially if you’re working with younger students, like what can they really do that’s going to be meaningful in the world? But now — you know, I’ve said I thought the screenwriters’ guild is taking it the wrong way. Instead of saying you can’t use AI to do some screenplays, they should be saying, "Hey, I could use AI to not just write the screenplay; I could produce the whole movie."

Like, you know, right now, I write a great screenplay, and I get a small sliver, and you know, they spend a hundred million dollars on it, and all these other people get all the money.

No, if you have a great idea for a movie, make the whole movie! And, you know, you’re already seeing with Kigo, students are able to produce things that are better than they would have been able to otherwise.

Once again, all of that’s transparent to the parents and teachers, so it’s not cheating, etc., but I think there are going to be a lot of opportunities there where the creative output is going to accelerate. You’re going to see a lot more things happen that weren’t — I’ve already — I’m always giving, you know, Uncle S is always giving talks to young people and saying, "You know, you have more potential to have more outlets for the world than any of us could dream of."

And this was before generative AI — this was just with YouTube and social media. You can publish yourself! Generative AI is going to take it to another level! Like, you can compose orchestral symphonies, you can, you know — anyway, there are going to be some really cool things.

So I would say that’s going to be positive. Um, I think on the internet — and this isn’t just for young people — if you think about it, part of the problem with the internet and social media is you have a bunch of AIs that are working for someone else and are manipulating your brain in order to meet the needs of someone else, in order to increase their revenue, their ad clicks, etc.

I’m excited because I think, you know, Kigo and generative AI is a co-pilot that can be an AI working on your side. So when you’re on social media, say, "Hey Jeremy, aren’t we supposed to be working on this paper?"

I don’t know if you should be looking at your ex-girlfriend’s marriage, Jeremy; that’s not — he doesn’t do that. But, you know, at their wedding photos. Um, you know, maybe we should get back to the topic — or something like that — or, you know, "Hey, you know, this isn’t good, or my sense is that this is actually fake news."

Um, "This isn’t — you should get a more balanced perspective." Like, imagine if we had something with us all the time that was helping us bring us back to a mentally healthier place and help bring society back to a less polarized place. That would be a huge boon!

Um, and I think there are ways to counteract some of the — you know, just as a bad actor could use these phishing attacks, you could have AI that answers the phishing attacks and tries to turn the tables on them. Just as a totalitarian government might try to monitor and police people, you could also use generative AI a little mischievously to make them distrust their inner circle.

Um, so, you know, we’ll see. That’s not above my pay grade, but I think it’s going to be, you know, AI technology is neutral. It’s an extension of human intent. And as long as the good folks use it at least as much as the bad folks or more, then we’re gonna have a better positive outcome from it.

Well, that’s a super hopeful note to end on, especially as we head into the holiday season. Um, really good things coming our way, and thanks to everyone for joining tonight.

Um, I love the questions; I love the energy. And S, thank you for sharing your expertise and wisdom here. So just keep letting us know; keep feeding us what you’re hungry for in the new year, and we’ll keep building for you.

Thank you all so much! Thank you, S! Everyone, have a beautiful evening! Thanks, everyone! Onward, onwards!