1 00:00:00,080 --> 00:00:03,920 Speaker 1: These sees Bloomberg Business Week with Carol Messer and Tim 2 00:00:03,960 --> 00:00:07,640 Speaker 1: Stenebec on Bloomberg Radio. We did catch up with sal 3 00:00:07,720 --> 00:00:10,559 Speaker 1: Cohn in late January. We got his thoughts on learning 4 00:00:10,560 --> 00:00:12,920 Speaker 1: in our post pandemic world, what things are sticking, what 5 00:00:13,039 --> 00:00:16,200 Speaker 1: things are not. We talked online education, something he and 6 00:00:16,200 --> 00:00:18,279 Speaker 1: his team have been doing for about fifteen years or so, 7 00:00:19,440 --> 00:00:21,639 Speaker 1: and now educators of all types, as you know, have 8 00:00:21,680 --> 00:00:24,560 Speaker 1: a new tool maybe to use. So back to talk 9 00:00:24,600 --> 00:00:26,360 Speaker 1: with us once again about the role of AI and 10 00:00:26,520 --> 00:00:30,240 Speaker 1: education and their specific moves in that regard. So com 11 00:00:30,240 --> 00:00:33,080 Speaker 1: back with us. Founder and CEO Khan Academy with us 12 00:00:33,159 --> 00:00:35,479 Speaker 1: via zoom from Mountain View, California. Sow, good to have 13 00:00:35,520 --> 00:00:38,280 Speaker 1: you back with us. How are you. I'm doing great, 14 00:00:38,280 --> 00:00:39,839 Speaker 1: Thanks for having me. Well, it's great to have you 15 00:00:39,920 --> 00:00:42,879 Speaker 1: here once again. Late January. We talked about a lot 16 00:00:42,920 --> 00:00:45,560 Speaker 1: of things, including AI, which it feels like all of 17 00:00:45,600 --> 00:00:47,760 Speaker 1: a sudden everybody was talking about it once again. You're 18 00:00:47,840 --> 00:00:51,920 Speaker 1: launching your own AI platform. Can you tell us about it? Yeah, 19 00:00:52,000 --> 00:00:54,720 Speaker 1: And when we talked in January, I was under n 20 00:00:55,280 --> 00:00:59,040 Speaker 1: I was under NDA, so I couldn't tell you what. Okay. 21 00:00:59,440 --> 00:01:03,480 Speaker 1: Back in back in August, open AI reached out to us. 22 00:01:04,040 --> 00:01:06,440 Speaker 1: They were working on GPT four. As many folks know 23 00:01:06,560 --> 00:01:08,640 Speaker 1: Chat GPT when it came out in November, it was 24 00:01:08,640 --> 00:01:12,040 Speaker 1: based on GPT three or three point five. But when 25 00:01:12,040 --> 00:01:15,320 Speaker 1: we saw what was possible with GPT four, we said, hey, 26 00:01:15,360 --> 00:01:17,640 Speaker 1: this could actually allow us to address some of the 27 00:01:17,680 --> 00:01:21,520 Speaker 1: holy grails in education, making it truly interactive, giving every 28 00:01:21,520 --> 00:01:24,880 Speaker 1: student a socratic tutor, being able to do things that 29 00:01:25,040 --> 00:01:28,240 Speaker 1: might have only looked like science fiction a few months ago. 30 00:01:28,600 --> 00:01:32,480 Speaker 1: And Open AI wanted to work with us because they said, look, 31 00:01:32,880 --> 00:01:35,280 Speaker 1: GPT four is going to be exciting but also scary 32 00:01:35,680 --> 00:01:37,560 Speaker 1: in certain ways. We want to be able to launch 33 00:01:37,600 --> 00:01:41,120 Speaker 1: with some social positive use cases, especially in education, especially 34 00:01:41,200 --> 00:01:44,760 Speaker 1: from folks that with folks that people trust. So we've 35 00:01:44,800 --> 00:01:48,440 Speaker 1: been working pretty feverishly on it for many months under NDA. 36 00:01:48,480 --> 00:01:50,280 Speaker 1: So it was really hard in January when you were 37 00:01:50,320 --> 00:01:52,720 Speaker 1: asking me questions about Chat GPT for me to not 38 00:01:52,800 --> 00:01:56,240 Speaker 1: spill the beads. But a couple of weeks ago we 39 00:01:56,360 --> 00:02:00,440 Speaker 1: launched and what we're calling con Migo, which is really 40 00:02:01,440 --> 00:02:06,320 Speaker 1: our incarnation of AI on Kin Academy using GPT four. 41 00:02:06,760 --> 00:02:10,760 Speaker 1: It does essentially two major strands of things. One is 42 00:02:10,800 --> 00:02:13,880 Speaker 1: now when a student is doing something on kin Academy, 43 00:02:14,200 --> 00:02:17,000 Speaker 1: it acts as their tutor, and it really does as 44 00:02:17,000 --> 00:02:19,200 Speaker 1: opposed to just giving you the answer. If a student says, 45 00:02:19,639 --> 00:02:21,519 Speaker 1: you know, telling me the interests, I'm your tutor. I 46 00:02:21,520 --> 00:02:23,320 Speaker 1: can help you and if you know, what do you 47 00:02:23,360 --> 00:02:24,960 Speaker 1: think is the next step. So it really tries to 48 00:02:24,960 --> 00:02:27,200 Speaker 1: push them socratically. It's not just in math, it's in 49 00:02:27,280 --> 00:02:32,320 Speaker 1: every subject. We have history, science, reading, comprehension. There's now 50 00:02:32,400 --> 00:02:35,080 Speaker 1: a tutor and this is a beta now where we've 51 00:02:35,080 --> 00:02:37,400 Speaker 1: only launched it to a few thousand people. There's a 52 00:02:37,440 --> 00:02:40,080 Speaker 1: waiting lists that's forming for folks who want to join. 53 00:02:40,120 --> 00:02:42,000 Speaker 1: It's about twenty thousand people right now, so we're trying 54 00:02:42,040 --> 00:02:44,840 Speaker 1: to take people off as quickly as possible. But the 55 00:02:44,880 --> 00:02:47,200 Speaker 1: other thing we realized is the things that would have 56 00:02:47,200 --> 00:02:51,040 Speaker 1: been impossible without AI. We have activities now where students 57 00:02:51,040 --> 00:02:56,000 Speaker 1: can have conversations with historical characters or literary characters. There 58 00:02:56,080 --> 00:02:58,919 Speaker 1: was a student recently, a high school student. She was 59 00:02:58,960 --> 00:03:01,119 Speaker 1: trying to understand some of the symbolism and the Great 60 00:03:01,160 --> 00:03:03,080 Speaker 1: Gatsby and she says, wait, I have conmigo, I can 61 00:03:03,240 --> 00:03:05,560 Speaker 1: I can now talk to Jay Gatsby himself? And she 62 00:03:05,639 --> 00:03:08,400 Speaker 1: has why are you staring at the green light? And 63 00:03:08,400 --> 00:03:11,040 Speaker 1: he's like, well, it symbolizes the things that I'm trying 64 00:03:11,080 --> 00:03:13,840 Speaker 1: to attain but I just can't. And she actually apologized 65 00:03:13,840 --> 00:03:16,680 Speaker 1: to him for taking up this time. The AI simulation. 66 00:03:17,320 --> 00:03:19,840 Speaker 1: We're creating teacher tools for them to create lesson plans 67 00:03:19,880 --> 00:03:22,360 Speaker 1: for them to refresh their own knowledge. We have a 68 00:03:22,520 --> 00:03:24,920 Speaker 1: kind of a built in guidance counselor now as an 69 00:03:24,960 --> 00:03:28,359 Speaker 1: AI I see a quick quick take piece. Yeah, a 70 00:03:28,440 --> 00:03:31,079 Speaker 1: conversation between teachers and students. We're going to be talking 71 00:03:31,200 --> 00:03:34,320 Speaker 1: and con migo. Yeah. Yeah, But it's just so it's 72 00:03:34,440 --> 00:03:37,280 Speaker 1: it's fascinating because of the amount of use cases that 73 00:03:37,320 --> 00:03:41,840 Speaker 1: you mentioned. Is it available? Talk to me about the availability, 74 00:03:41,840 --> 00:03:46,080 Speaker 1: because I don't think I'm clear on how widespread access 75 00:03:46,200 --> 00:03:48,000 Speaker 1: is at this moment and where you're at in that 76 00:03:48,120 --> 00:03:52,600 Speaker 1: and where it's going. Yeah. So when we launched two 77 00:03:52,600 --> 00:03:54,880 Speaker 1: weeks ago, and we're still kind of in the process 78 00:03:54,920 --> 00:03:57,240 Speaker 1: of launch launching, there's a couple of things we wanted 79 00:03:57,240 --> 00:03:58,800 Speaker 1: to figure out. If one, we just wanted to make 80 00:03:58,840 --> 00:04:03,640 Speaker 1: sure it's working well for in the last weeks we've 81 00:04:03,680 --> 00:04:05,960 Speaker 1: started to get some really positive feedback on that one. 82 00:04:06,120 --> 00:04:07,440 Speaker 1: The other thing we're trying to figure out is how 83 00:04:07,440 --> 00:04:09,240 Speaker 1: do we cost this out and how do we resource 84 00:04:09,360 --> 00:04:13,280 Speaker 1: this the computation costs for leveraging these these next generation 85 00:04:13,360 --> 00:04:15,880 Speaker 1: models like GPT four, it is not free even if 86 00:04:15,880 --> 00:04:19,480 Speaker 1: you're a Chat GPT pro customer. And these are folks 87 00:04:19,480 --> 00:04:21,840 Speaker 1: paying twenty dollars a month to open AI. They only 88 00:04:21,880 --> 00:04:24,839 Speaker 1: get GPT four for I think at twenty five interactions 89 00:04:25,440 --> 00:04:27,480 Speaker 1: per day and you have to wait three hours, so 90 00:04:27,520 --> 00:04:29,920 Speaker 1: they've so in a lot of ways, Conmigo, what we're 91 00:04:29,920 --> 00:04:33,839 Speaker 1: doing is in many ways the most unfettered access to 92 00:04:33,880 --> 00:04:37,120 Speaker 1: GPT four. So we're trying to figure out ways that 93 00:04:37,880 --> 00:04:39,800 Speaker 1: we can we can resource it. So right now there's 94 00:04:39,800 --> 00:04:42,200 Speaker 1: a waiting list, when people to get off the waiting list. 95 00:04:42,279 --> 00:04:44,640 Speaker 1: Right now we're prioritizing people who are donating. It's a 96 00:04:44,680 --> 00:04:47,920 Speaker 1: twenty dollars donation. It helps us resource the computation costs 97 00:04:47,960 --> 00:04:49,479 Speaker 1: and also do the R and D. We are not 98 00:04:49,560 --> 00:04:52,719 Speaker 1: for profits, so that is a donation, a true donation. 99 00:04:53,160 --> 00:04:54,640 Speaker 1: But we're trying to figure out how we can bring 100 00:04:54,720 --> 00:04:57,680 Speaker 1: that that that price down over time so that we 101 00:04:57,720 --> 00:05:00,919 Speaker 1: can make it more and more accessible. Because rather I 102 00:05:01,000 --> 00:05:03,599 Speaker 1: think about you and kon Academy as this great leveler 103 00:05:03,720 --> 00:05:06,560 Speaker 1: when it comes to education. Right, your videos are out there, 104 00:05:06,560 --> 00:05:08,880 Speaker 1: anybody can access them. So how do we make sure 105 00:05:08,920 --> 00:05:13,080 Speaker 1: that once again, we don't have a technology potentially that 106 00:05:13,440 --> 00:05:17,200 Speaker 1: widens the gap rather than reduces it. Yeah, this is 107 00:05:17,240 --> 00:05:19,480 Speaker 1: front of mine for us, and I think there's two dimensions. 108 00:05:19,480 --> 00:05:21,440 Speaker 1: So that one is just the cost issue that we've 109 00:05:21,440 --> 00:05:24,840 Speaker 1: already talked about. Obviously, everything that we stand for is 110 00:05:25,160 --> 00:05:27,919 Speaker 1: our mission is free world class education for anyone anywhere. 111 00:05:28,160 --> 00:05:31,360 Speaker 1: We're going to try to make this as accessible as possible. 112 00:05:31,680 --> 00:05:35,520 Speaker 1: Keeping in mind there is a higher cost, but if 113 00:05:35,520 --> 00:05:37,800 Speaker 1: you think about if we can bring this down to 114 00:05:38,480 --> 00:05:40,760 Speaker 1: let's say a few dollars per student per month, this 115 00:05:40,839 --> 00:05:45,720 Speaker 1: is dramatically lower cost than conventional tutoring. And a lot 116 00:05:45,760 --> 00:05:47,359 Speaker 1: of the school districts have been spending a lot of 117 00:05:47,360 --> 00:05:50,360 Speaker 1: money on conventional tutoring lately with money that was released 118 00:05:50,360 --> 00:05:54,480 Speaker 1: after the pandemic, and they haven't seen results because it 119 00:05:54,520 --> 00:05:56,680 Speaker 1: hasn't been connected to what students are doing in class, 120 00:05:56,720 --> 00:05:59,880 Speaker 1: and it isn't happening during class, So the kids who 121 00:05:59,880 --> 00:06:03,039 Speaker 1: are engaging or are usually the kids who don't need it. 122 00:06:03,400 --> 00:06:05,520 Speaker 1: The kids who need it aren't engaging after school, etc. 123 00:06:06,080 --> 00:06:07,560 Speaker 1: So we think we're gonna be able to get this 124 00:06:07,680 --> 00:06:10,880 Speaker 1: to a much broader set of students. We're starting to 125 00:06:10,920 --> 00:06:14,000 Speaker 1: partner with some school districts, large urban school districts, so 126 00:06:14,040 --> 00:06:17,640 Speaker 1: we can bring it to their students for free, and 127 00:06:17,680 --> 00:06:21,280 Speaker 1: then hopefully we make it more and more accessible over time. 128 00:06:21,320 --> 00:06:23,520 Speaker 1: The other dimension to your question, the way we've been 129 00:06:23,560 --> 00:06:26,080 Speaker 1: thinking about it is a lot of school districts announce 130 00:06:26,120 --> 00:06:30,080 Speaker 1: that they're banning chat GPT, and there's some good reason 131 00:06:30,160 --> 00:06:32,839 Speaker 1: why they're doing that. Obviously, students can use it to cheat, 132 00:06:32,880 --> 00:06:35,120 Speaker 1: They can do shady things with it because it's a 133 00:06:35,200 --> 00:06:40,080 Speaker 1: very open tool on Conmigo, all of the conversations are 134 00:06:40,160 --> 00:06:43,760 Speaker 1: viewable by the teachers and the parents. There's a strong moderation. 135 00:06:44,600 --> 00:06:47,960 Speaker 1: All of the activities are designed to be very pedagogically positive. 136 00:06:48,040 --> 00:06:51,239 Speaker 1: There's hopefully no cheating. More. It will write a paper 137 00:06:51,279 --> 00:06:52,880 Speaker 1: with you, it won't write a paper for you. It 138 00:06:52,920 --> 00:06:55,440 Speaker 1: will tutor you, but it will not tell you the answer. 139 00:06:56,000 --> 00:06:58,440 Speaker 1: So some of those same school districts who banned it, 140 00:06:58,720 --> 00:07:00,680 Speaker 1: and I was afraid when the band because I'm like, 141 00:07:00,680 --> 00:07:02,680 Speaker 1: this is going to create another digital divide because the 142 00:07:03,160 --> 00:07:05,560 Speaker 1: well resource kids are still going to use it. Those 143 00:07:05,600 --> 00:07:07,920 Speaker 1: same school districts who banned it. Now we're saying, hey, 144 00:07:07,960 --> 00:07:09,520 Speaker 1: this is what we wanted. We don't want to stick 145 00:07:09,520 --> 00:07:11,440 Speaker 1: our head in the sand. We want to move forward 146 00:07:11,480 --> 00:07:12,800 Speaker 1: with the times. We just want to do it in 147 00:07:12,800 --> 00:07:16,360 Speaker 1: a safe way that's actually pedagogically positive. So we're pretty 148 00:07:16,360 --> 00:07:19,560 Speaker 1: excited that hopefully we can bring get the cost down. 149 00:07:19,600 --> 00:07:22,600 Speaker 1: Some of that is open AI and Azure and Microsoft 150 00:07:22,600 --> 00:07:25,760 Speaker 1: and all of them getting the efficiencies, and then on 151 00:07:25,800 --> 00:07:28,120 Speaker 1: our side, we're trying to bring the efficiencies in so 152 00:07:28,160 --> 00:07:30,320 Speaker 1: that we can make it as accessible as possible and 153 00:07:30,560 --> 00:07:33,080 Speaker 1: make sure that all kids have access to this. But 154 00:07:33,200 --> 00:07:36,840 Speaker 1: access is such a multi layered thing, Sal, I wonder 155 00:07:37,040 --> 00:07:39,520 Speaker 1: how you're thinking about partnerships when it comes to that, 156 00:07:39,560 --> 00:07:42,400 Speaker 1: because we're still trying to wire up the country. Well, right, 157 00:07:42,440 --> 00:07:45,280 Speaker 1: we've got the internet access thing. I'm on your website. 158 00:07:45,280 --> 00:07:48,720 Speaker 1: It looks like it's like an iPhone is being used 159 00:07:48,880 --> 00:07:51,760 Speaker 1: or samsuent or whatever it is. Like, what about people 160 00:07:51,760 --> 00:07:54,360 Speaker 1: who don't have access to a smartphone? Right, we have 161 00:07:54,400 --> 00:07:56,160 Speaker 1: about a minute and a half left, But I wonder 162 00:07:56,200 --> 00:07:57,800 Speaker 1: how you're thinking about that. And then we'll come back 163 00:07:57,800 --> 00:08:02,200 Speaker 1: and talk some more. Yeah, it's a tough question. Simple 164 00:08:02,240 --> 00:08:05,120 Speaker 1: answer is digital access in the US at least has 165 00:08:05,120 --> 00:08:08,080 Speaker 1: gotten dramatically better. Cell Phone access in the rest of 166 00:08:08,120 --> 00:08:10,760 Speaker 1: the world also has gotten better. We are trying to 167 00:08:10,800 --> 00:08:13,800 Speaker 1: skate to where the puck is going on that front. 168 00:08:14,720 --> 00:08:17,840 Speaker 1: In the US, we think it's all about getting it 169 00:08:17,880 --> 00:08:20,760 Speaker 1: to students in the school setting because at home, as 170 00:08:20,760 --> 00:08:23,960 Speaker 1: you know, there is more inconsistent access. And so that's 171 00:08:23,960 --> 00:08:27,080 Speaker 1: why we're working so closely with school districts, which for 172 00:08:27,120 --> 00:08:29,840 Speaker 1: the most part now do have one to one laptops 173 00:08:29,920 --> 00:08:32,960 Speaker 1: or at least close to that inside of their classrooms. 174 00:08:33,120 --> 00:08:35,040 Speaker 1: And then in other countries, in places like India, we're 175 00:08:35,040 --> 00:08:38,440 Speaker 1: partnering with state governments and that is multi multi layered. 176 00:08:38,440 --> 00:08:40,720 Speaker 1: We need to translate the content, we need to align 177 00:08:40,720 --> 00:08:42,200 Speaker 1: it to their standards, and we also have to figure 178 00:08:42,240 --> 00:08:44,400 Speaker 1: out how we get the vice access right. This is 179 00:08:44,400 --> 00:08:45,920 Speaker 1: why we always like talking to you because it's the 180 00:08:45,920 --> 00:08:48,680 Speaker 1: global perspective. It's not just about teachers and kids and 181 00:08:48,679 --> 00:08:50,840 Speaker 1: here in the US, but it's really about teachers and 182 00:08:50,920 --> 00:08:53,720 Speaker 1: kids everywhere in the world. So so one of the 183 00:08:53,720 --> 00:08:56,440 Speaker 1: things you're talking about, you know, your new AI platform 184 00:08:56,440 --> 00:08:59,640 Speaker 1: that you guys have created, and I came across, are 185 00:08:59,679 --> 00:09:03,439 Speaker 1: we and prepping for our conversation with you. Stanford put 186 00:09:03,440 --> 00:09:05,600 Speaker 1: out a report earlier this month warning that decisions made 187 00:09:05,600 --> 00:09:09,360 Speaker 1: by AI could lead to nuclear level catastrophe. Why is 188 00:09:09,360 --> 00:09:12,400 Speaker 1: it not too early to start playing around with the 189 00:09:12,480 --> 00:09:18,160 Speaker 1: next level AI when it comes to education? Well, I think, 190 00:09:18,200 --> 00:09:21,120 Speaker 1: like any technology, there could be a lot of positives 191 00:09:21,120 --> 00:09:23,880 Speaker 1: and there could also be a lot of possible negatives. 192 00:09:24,000 --> 00:09:26,680 Speaker 1: And I think as a society, it's all about mitigating 193 00:09:26,679 --> 00:09:29,920 Speaker 1: the negatives and then using it for the positives where possible. 194 00:09:31,760 --> 00:09:34,400 Speaker 1: The genie is to some degree out of the bottle 195 00:09:34,480 --> 00:09:36,400 Speaker 1: a little bit, and so I know, I tell this 196 00:09:36,520 --> 00:09:38,559 Speaker 1: the team at Kon Academy, and tell this to our donors. 197 00:09:38,760 --> 00:09:41,880 Speaker 1: I view it as our responsibility as a nonprofit in 198 00:09:42,080 --> 00:09:46,599 Speaker 1: education to show how you can use these technologies and 199 00:09:46,679 --> 00:09:49,600 Speaker 1: a very positive and a very safe and a very 200 00:09:49,640 --> 00:09:53,800 Speaker 1: trusting way, Because if we do that, then we're going 201 00:09:53,840 --> 00:09:55,360 Speaker 1: to be able to reach all these students that we've 202 00:09:55,400 --> 00:09:57,920 Speaker 1: been talking about, and we'll hopefully be able to accelerate 203 00:09:57,920 --> 00:10:02,960 Speaker 1: their learning pretty dramatically. Now, there are other concerns about 204 00:10:03,160 --> 00:10:05,559 Speaker 1: what happens if you have super intelligent AIS. You know, 205 00:10:05,559 --> 00:10:08,040 Speaker 1: who know who knows what GPT five or six or 206 00:10:08,080 --> 00:10:11,000 Speaker 1: seven is going to be able to do planning military 207 00:10:11,040 --> 00:10:14,360 Speaker 1: strategy or figuring out what targets are, etc. But even there, 208 00:10:15,280 --> 00:10:19,440 Speaker 1: I personally think that I would rather have the actors 209 00:10:19,440 --> 00:10:22,199 Speaker 1: in the US being on the cutting edge than US 210 00:10:22,280 --> 00:10:27,040 Speaker 1: pausing and slowing things down and that and letting folks 211 00:10:27,080 --> 00:10:30,360 Speaker 1: who are maybe not as altruistic continue to move forward. 212 00:10:31,559 --> 00:10:33,320 Speaker 1: You know, no matter what we do, I guarantee you 213 00:10:33,360 --> 00:10:35,080 Speaker 1: the Chinese or the Russians are not going to be 214 00:10:35,120 --> 00:10:38,040 Speaker 1: stopping their generative large language model research. So I actually 215 00:10:38,080 --> 00:10:41,720 Speaker 1: think it's a national security issue to keep moving, not 216 00:10:41,800 --> 00:10:44,960 Speaker 1: to stop. Yeah, that's that's such an incredible point. And 217 00:10:45,000 --> 00:10:47,320 Speaker 1: also that you were mentioning that this is just a 218 00:10:47,360 --> 00:10:50,400 Speaker 1: story we've heard before, Like every single innovation has this 219 00:10:50,480 --> 00:10:55,840 Speaker 1: big question, and you can't just like let the United 220 00:10:55,840 --> 00:10:58,440 Speaker 1: States be the only one to not have social media, 221 00:10:58,480 --> 00:11:01,920 Speaker 1: for example, because we're worried about the potential downfalls. Right, 222 00:11:03,720 --> 00:11:07,439 Speaker 1: that's right, And you know social media is not I don't. Well, 223 00:11:08,040 --> 00:11:10,680 Speaker 1: it can be a national security issues, as we've sometimes seen. 224 00:11:10,720 --> 00:11:13,080 Speaker 1: It can lead to polarization, it can lead to you know, 225 00:11:13,120 --> 00:11:17,240 Speaker 1: people believing things that aren't true, but it doesn't quite 226 00:11:17,320 --> 00:11:20,880 Speaker 1: have the same strategic consequence as something like artificial intelligence. 227 00:11:20,960 --> 00:11:24,120 Speaker 1: So an artificial intelligtion is going to intersect with social media, 228 00:11:24,120 --> 00:11:26,120 Speaker 1: It's going to be influencing how people think, et cetera. 229 00:11:26,240 --> 00:11:30,360 Speaker 1: So I feel much better if relatively positive actors in 230 00:11:31,040 --> 00:11:35,600 Speaker 1: the States and aligned or countries aren't thinking about how 231 00:11:35,640 --> 00:11:39,400 Speaker 1: to use this responsibly and not falling behind folks who 232 00:11:39,400 --> 00:11:43,360 Speaker 1: are who have less less issues with using it irresponsibly. 233 00:11:43,600 --> 00:11:47,839 Speaker 1: Are you thinking about or concerned about any legislation on AI. 234 00:11:47,880 --> 00:11:51,000 Speaker 1: I know we talked about rulings at the school district level, 235 00:11:51,040 --> 00:11:53,760 Speaker 1: but are you concerned about any federal legislation that might 236 00:11:53,800 --> 00:11:59,959 Speaker 1: impact usage. I'm always worried when people want to pass legislation, 237 00:12:00,000 --> 00:12:02,520 Speaker 1: should a regulation without having a clear picture of what 238 00:12:02,559 --> 00:12:05,320 Speaker 1: they're trying to protect against. I think right now there's 239 00:12:05,320 --> 00:12:08,480 Speaker 1: a lot of knee jerk fears based on dystopian science 240 00:12:08,520 --> 00:12:11,960 Speaker 1: fiction novels about AIS taking over, etc. You know, a 241 00:12:11,960 --> 00:12:15,280 Speaker 1: lot of the folks calling for regulation can't exactly pinpoint 242 00:12:15,320 --> 00:12:18,800 Speaker 1: to what tangibly they think is going to happen, and 243 00:12:19,240 --> 00:12:23,400 Speaker 1: they can't point to actual examples of that happening. You know, 244 00:12:23,440 --> 00:12:26,040 Speaker 1: the Internet is actually much more of a wild West 245 00:12:26,320 --> 00:12:29,880 Speaker 1: than large language models are right now, and I think 246 00:12:29,920 --> 00:12:32,679 Speaker 1: for the most part, it's probably a good thing. It's 247 00:12:32,720 --> 00:12:36,440 Speaker 1: been left unregulated. I'm not antiregulation. What I would say is, 248 00:12:36,679 --> 00:12:41,200 Speaker 1: let's see where there are issues that pop up, and 249 00:12:41,240 --> 00:12:44,160 Speaker 1: then determine is regulation the best mechanism to do that, 250 00:12:44,480 --> 00:12:46,040 Speaker 1: and then and then that should be a you know, 251 00:12:46,200 --> 00:12:50,040 Speaker 1: chev robust conversation about that. For teachers, what would something 252 00:12:50,080 --> 00:12:54,040 Speaker 1: like this, this kind of functionality and using generative AI 253 00:12:54,200 --> 00:12:56,880 Speaker 1: specifically like you guys are doing, What does it mean 254 00:12:56,920 --> 00:12:59,800 Speaker 1: for teachers specifically who We recently had an author on 255 00:13:00,160 --> 00:13:03,000 Speaker 1: an incredible book and just you know, she embedded herself 256 00:13:03,040 --> 00:13:06,080 Speaker 1: for a year with three different teachers and the stresses, 257 00:13:06,120 --> 00:13:08,960 Speaker 1: the strains not paid well. It's it's a really tough 258 00:13:09,040 --> 00:13:11,880 Speaker 1: profession right now, and to be fair, not really well 259 00:13:11,920 --> 00:13:16,080 Speaker 1: respected by many people out there. And that's unfortunate because 260 00:13:16,080 --> 00:13:19,720 Speaker 1: this is talking about educating the next generation. So what 261 00:13:20,000 --> 00:13:25,480 Speaker 1: might something like this do for teachers? Well, first and foremost, 262 00:13:25,840 --> 00:13:28,160 Speaker 1: you know, your average teacher has thirty kids in the classroom. 263 00:13:28,200 --> 00:13:31,120 Speaker 1: They wish they could replicate themselves, give more personalized attention 264 00:13:31,120 --> 00:13:33,680 Speaker 1: to every student. This is something we've always worked on 265 00:13:33,720 --> 00:13:36,720 Speaker 1: kind academy, but having this AI tutor in the room 266 00:13:36,760 --> 00:13:40,200 Speaker 1: with you. It is like replicating you, we've already seen this. 267 00:13:40,240 --> 00:13:44,079 Speaker 1: When we've been piloting it at some classrooms, the teacher's like, wow, 268 00:13:44,120 --> 00:13:45,800 Speaker 1: it just answered a question that I would have had 269 00:13:45,840 --> 00:13:49,440 Speaker 1: to try to answer, but I wouldn't been able to 270 00:13:49,440 --> 00:13:51,800 Speaker 1: connect with this other student. So it's already doing that. 271 00:13:51,880 --> 00:13:55,160 Speaker 1: But I think even more importantly, teachers spend about thirty 272 00:13:55,160 --> 00:14:01,520 Speaker 1: forty fifty percent of their time planning less than refreshing knowledge. 273 00:14:02,200 --> 00:14:05,400 Speaker 1: We have activities four teachers. We've made them a first 274 00:14:05,480 --> 00:14:08,959 Speaker 1: year user here where it helps them develop lesson plans 275 00:14:09,000 --> 00:14:12,120 Speaker 1: in seconds that would have otherwise have taken hours. It 276 00:14:12,160 --> 00:14:15,520 Speaker 1: helps them prepare their knowledge, It helps them think about 277 00:14:15,600 --> 00:14:18,080 Speaker 1: like almost like an instructional coach, what they how they 278 00:14:18,080 --> 00:14:21,000 Speaker 1: can intervene with their students. So we're kind of calling 279 00:14:21,080 --> 00:14:24,000 Speaker 1: Kanmigo the r AI a tutor for every student and 280 00:14:24,040 --> 00:14:26,320 Speaker 1: a teaching assistant or an army of teaching assistance for 281 00:14:26,360 --> 00:14:29,920 Speaker 1: every teacher. And so I think it'll hopefully lower their 282 00:14:30,360 --> 00:14:32,120 Speaker 1: burden and be able to free up more time for 283 00:14:32,200 --> 00:14:34,840 Speaker 1: the teachers to spend with the students. Can it help 284 00:14:34,920 --> 00:14:39,720 Speaker 1: with grading at all the simpleasures? Yes, I think it is. 285 00:14:40,280 --> 00:14:43,240 Speaker 1: We didn't launched two weeks ago. We have been piloting 286 00:14:43,240 --> 00:14:44,720 Speaker 1: some of this, but we want to make sure that 287 00:14:44,760 --> 00:14:48,240 Speaker 1: it's really robust before we launch it. But I think 288 00:14:48,280 --> 00:14:51,160 Speaker 1: within the next year you're going to see things like 289 00:14:51,640 --> 00:14:56,480 Speaker 1: a teacher saying, hey, Kanmigo, let's develop an activity for 290 00:14:56,520 --> 00:15:00,440 Speaker 1: our students, and Kanmigo says, all right, let's have them 291 00:15:00,480 --> 00:15:02,680 Speaker 1: all right, you know, whether the Civil War could have 292 00:15:02,680 --> 00:15:04,600 Speaker 1: been avoided and how, and then the teachers like that 293 00:15:04,640 --> 00:15:08,000 Speaker 1: sounds great, And then every student sees that assignment and 294 00:15:08,040 --> 00:15:10,800 Speaker 1: then they work with Conmigo to kind of talk about 295 00:15:12,240 --> 00:15:14,680 Speaker 1: whether the Civil War would have happened otherwise. And then 296 00:15:14,880 --> 00:15:17,840 Speaker 1: Kanmigo assesses it and tells the teacher, Hey, these students 297 00:15:17,840 --> 00:15:22,040 Speaker 1: had some really exemplary responses. These students struggled. You know, 298 00:15:22,360 --> 00:15:24,520 Speaker 1: I would give these students an a. I would have 299 00:15:24,560 --> 00:15:26,560 Speaker 1: these students continue to work on it. Right, that's not 300 00:15:26,640 --> 00:15:28,760 Speaker 1: science fiction anymore. I think we're months away from that, 301 00:15:28,840 --> 00:15:31,560 Speaker 1: not years, Especially when we talk about individualized learning, like 302 00:15:31,760 --> 00:15:34,280 Speaker 1: you think about this, it sounds conceptually like it could 303 00:15:34,280 --> 00:15:36,280 Speaker 1: really take us to that level. Hey, Saal just got 304 00:15:36,280 --> 00:15:39,040 Speaker 1: about twenty five thirty seconds. Well we look back five 305 00:15:39,120 --> 00:15:41,440 Speaker 1: years from now, ten years, and is there a superlative 306 00:15:41,480 --> 00:15:45,240 Speaker 1: to describe what generative AI may mean to education. And 307 00:15:45,280 --> 00:15:47,520 Speaker 1: I'm not trying to look for that crazy headline, but 308 00:15:47,560 --> 00:15:50,320 Speaker 1: I'm just curious how you're looking at it. Just quickly. 309 00:15:51,560 --> 00:15:54,440 Speaker 1: I think this is the biggest inflection point of our lifetimes, 310 00:15:54,520 --> 00:15:56,200 Speaker 1: and I think this is going to give a tutor 311 00:15:56,240 --> 00:15:58,560 Speaker 1: to every student on the planet and a teaching assistant 312 00:15:58,560 --> 00:16:01,440 Speaker 1: to every teacher. So it's a big deal. Well, listen, 313 00:16:01,480 --> 00:16:03,680 Speaker 1: thank you so much, and so appreciate you coming back 314 00:16:03,720 --> 00:16:06,160 Speaker 1: when you couldn't talk about it, now you could. We 315 00:16:06,200 --> 00:16:09,520 Speaker 1: really appreciate you sharing that with us in our Bloomberg 316 00:16:09,520 --> 00:16:13,200 Speaker 1: listeners and viewers. So appreciate it. All right. SoCon, be well, 317 00:16:13,240 --> 00:16:16,320 Speaker 1: Take care. SoCon, Founder, chief executive officer of Khan Academy.