1 00:00:03,300 --> 00:00:05,400 Speaker: Welcome to make EdTech 100. 2 00:00:06,180 --> 00:00:10,530 I am LindyHoc Educator, K 12 Ed Tech Advisor, and your host. 3 00:00:11,340 --> 00:00:15,989 This is a podcast where we keep it real about what actually works in classrooms. 4 00:00:16,440 --> 00:00:21,450 No hype, no overwhelm, just practical strategies, honest stories and 5 00:00:21,450 --> 00:00:25,050 tools that make a real difference for teachers and students. 6 00:00:25,110 --> 00:00:28,200 So come along with me on a journey to make EdTech 100. 7 00:00:32,981 --> 00:00:37,601 Speaker 2: Today's episode is a part two, a Companion two. 8 00:00:37,631 --> 00:00:41,651 Episode nine is AI Safe for Students. 9 00:00:41,651 --> 00:00:45,671 So if you haven't heard that one, go back and listen first. 10 00:00:45,671 --> 00:00:47,171 If you don't, it's not the end of the world. 11 00:00:47,171 --> 00:00:49,571 You'll still get a lot out of this episode. 12 00:00:49,571 --> 00:00:56,411 But there's some important context that I laid there in that first part one episode. 13 00:00:56,461 --> 00:00:58,981 Today is AI Safe for Students. 14 00:00:59,011 --> 00:01:02,581 Part two is our subject matter. 15 00:01:03,061 --> 00:01:07,351 After that first episode dropped, some research also dropped. 16 00:01:07,401 --> 00:01:12,951 So I wanna continue this discussion of is AI safe for students and 17 00:01:12,951 --> 00:01:15,471 discuss that research summary. 18 00:01:15,831 --> 00:01:21,051 Also, I know I have some parent listeners, part one focus fully on the teacher 19 00:01:21,051 --> 00:01:27,771 perspective, and I gave my three things that I look for in an AI tool to use 20 00:01:27,801 --> 00:01:32,241 in learning or classroom environments, guardrails, teacher dashboards for 21 00:01:32,241 --> 00:01:37,791 visibility, and using it at key instructional moments for normally very 22 00:01:37,791 --> 00:01:40,371 short, five to 10 minute periods of time. 23 00:01:40,431 --> 00:01:41,571 Just as a summary there. 24 00:01:42,756 --> 00:01:47,556 Since I have some parent listeners, I wanted to kind of add onto that, and 25 00:01:47,556 --> 00:01:53,556 I created some dos and some don'ts when it comes to using AI tools 26 00:01:53,556 --> 00:01:58,566 with both students in a classroom environment as a teacher or instructor 27 00:01:58,566 --> 00:02:03,666 or educator, but also kids if you're parent, trying to help your kids with 28 00:02:03,696 --> 00:02:06,546 their schoolwork at home, for example. 29 00:02:06,546 --> 00:02:07,536 Let's get into it. 30 00:02:07,536 --> 00:02:13,776 Stanford has an AI hub for education that has over 800 research studies 31 00:02:13,776 --> 00:02:18,756 on all things AI and education, and just a couple, two, three 32 00:02:18,756 --> 00:02:21,366 weeks ago, they released a summary. 33 00:02:22,146 --> 00:02:27,096 This research repository, and they called it understanding the evidence 34 00:02:27,096 --> 00:02:30,546 base on AI and K 12 education. 35 00:02:30,996 --> 00:02:35,736 It has some important takeaways around the topic of is AI safe for students? 36 00:02:36,366 --> 00:02:40,356 First of all, the majority of the studies that were indicated as 37 00:02:40,356 --> 00:02:45,396 being high quality causal studies, which were not very many by the way. 38 00:02:45,396 --> 00:02:49,476 I'll talk about that more in a bit, focused on students as 39 00:02:49,476 --> 00:02:52,026 the users of ai, not teachers. 40 00:02:52,026 --> 00:02:53,526 That's really interesting. 41 00:02:53,745 --> 00:03:00,135 Moral of the story is that the number of studies deemed to be high quality causal 42 00:03:00,375 --> 00:03:05,896 were low, and of those that were deemed to be high quality causal, the majority 43 00:03:05,896 --> 00:03:10,005 of them were about students using ai. 44 00:03:10,741 --> 00:03:13,111 As a tool rather than teacher use. 45 00:03:14,111 --> 00:03:18,940 One important takeaway from this research summary was, I quote, the 46 00:03:18,940 --> 00:03:24,071 research base is growing quickly, but rigorous evidence is still thin. 47 00:03:24,280 --> 00:03:29,650 A k, a, we have some research, but a lot more is needed. 48 00:03:30,650 --> 00:03:33,741 That doesn't mean that we shouldn't take. 49 00:03:34,505 --> 00:03:39,755 What the research is saying and look at it and use it to guide what we're 50 00:03:39,755 --> 00:03:44,585 doing, especially when it comes to emerging technologies where of 51 00:03:44,585 --> 00:03:46,566 course it's a brand new technology. 52 00:03:46,566 --> 00:03:52,835 The research is going to be a constantly increasing and changing field. 53 00:03:53,625 --> 00:03:59,175 Of the high causal studies, one major takeaway from that research summary from 54 00:03:59,175 --> 00:04:05,235 Stanford was that learning improvements were noted when it comes to using AI 55 00:04:05,235 --> 00:04:12,566 as a learning tool, essentially with student and learners, but with one 56 00:04:12,655 --> 00:04:16,706 really big, but is the caveat there. 57 00:04:16,706 --> 00:04:24,806 So learning improvements were noted, but tool design matters. 58 00:04:25,306 --> 00:04:27,736 Here's some quotes from that study. 59 00:04:28,366 --> 00:04:32,146 Or I should say, research summary across the causal studies. 60 00:04:32,146 --> 00:04:37,546 Reviewed evidence suggests that AI tools designed with pedagogical guardrails, 61 00:04:37,816 --> 00:04:43,486 such as tutoring systems that give hints or guide reasoning, show more 62 00:04:43,486 --> 00:04:49,456 promising outcomes than general purpose chatbots that provide answers directly. 63 00:04:50,116 --> 00:04:54,976 Here are some other quotes from the studies directly, so 64 00:04:54,976 --> 00:04:56,416 that one was from the summary. 65 00:04:56,656 --> 00:04:58,816 These are from the studies that were deemed as high 66 00:04:58,816 --> 00:05:01,306 causal as part of that summary. 67 00:05:01,546 --> 00:05:06,106 Not all AI tools function the same way in learning environments. 68 00:05:06,916 --> 00:05:10,726 Tools that support reasoning may help support learning. 69 00:05:11,446 --> 00:05:15,886 While tools that simply generate answers may reduce the cognitive effort that 70 00:05:15,886 --> 00:05:18,826 supports durable skill development. 71 00:05:19,426 --> 00:05:23,926 In other words, using AI with students is not black or white. 72 00:05:24,436 --> 00:05:29,206 It depends entirely on the tool that you're using. 73 00:05:29,506 --> 00:05:34,306 Tools that guide and not give the answer are those that are 74 00:05:34,306 --> 00:05:37,186 linked to learning improvements. 75 00:05:37,756 --> 00:05:41,986 The research cites this as a Socratic style tool 76 00:05:42,466 --> 00:05:45,106 a Socratic style tool guides not give. 77 00:05:45,106 --> 00:05:49,996 So for example, if you were to give a Socratic style AI tool, a math 78 00:05:49,996 --> 00:05:53,836 problem, it would say something like, let's work the through this together. 79 00:05:53,986 --> 00:05:55,936 Let's break down the steps. 80 00:05:56,146 --> 00:06:01,426 Let's take it one by one, rather than just giving the answer outright, which is 81 00:06:01,426 --> 00:06:05,986 what you're gonna get if you're gonna go into say, like a chat GBT esque type tool. 82 00:06:06,986 --> 00:06:12,056 The research summary also says that learning science provides one way 83 00:06:12,296 --> 00:06:17,576 to interpret these findings, and it discusses various different educational 84 00:06:17,576 --> 00:06:25,466 theories on how an AI tool design would provide opportunity and also risk. 85 00:06:26,351 --> 00:06:30,101 If you dig down deep teachers that are listening, if you dig down deep 86 00:06:30,101 --> 00:06:34,151 to your teacher prep program, you likely remember your educational 87 00:06:34,151 --> 00:06:37,061 theory or educational psychology class. 88 00:06:37,061 --> 00:06:41,111 Every person that has a teaching degree has to take this class. 89 00:06:41,351 --> 00:06:44,201 And you learned about all the different theories out there. 90 00:06:44,251 --> 00:06:50,371 You may remember, Vygotsky Vygotsky talks about the zone of 91 00:06:50,371 --> 00:06:53,551 proximal development or the ZPD. 92 00:06:53,881 --> 00:07:00,061 This is the idea that you have to meet students within their zone 93 00:07:00,061 --> 00:07:05,731 of proximal development or their ZPD if you are outside their ZPD. 94 00:07:06,091 --> 00:07:11,641 The learner's not prepared essentially for the information that you're giving them. 95 00:07:12,211 --> 00:07:16,141 This is a great way to think about using a guard, railed large 96 00:07:16,141 --> 00:07:17,911 language model with students. 97 00:07:18,181 --> 00:07:24,421 They are not ready for the, give me the answer that these tools do us, 98 00:07:24,421 --> 00:07:31,291 that are adults with fully developed brains can also struggle with this. 99 00:07:31,291 --> 00:07:32,881 This can be we're overseeing that. 100 00:07:32,881 --> 00:07:38,071 This can be challenging for adults with fully developed brains as well. 101 00:07:38,461 --> 00:07:42,151 So it makes sense that it would be challenging for. 102 00:07:42,676 --> 00:07:48,016 Kids that do not have fully developed brain. 103 00:07:48,016 --> 00:07:49,246 Why don't I say that one more time? 104 00:07:49,306 --> 00:07:49,846 You get the point. 105 00:07:49,846 --> 00:07:50,446 You get the point. 106 00:07:51,446 --> 00:07:55,256 So if you remember from the Part one episode I gave you those three 107 00:07:55,256 --> 00:07:58,586 requirements when it comes to using AI as a learning tool, the 108 00:07:58,586 --> 00:08:03,776 guardrails, the teacher dashboard, and the key instruction moments. 109 00:08:04,046 --> 00:08:09,686 So I told you that the first requirement for the guardrails, the 110 00:08:09,686 --> 00:08:12,536 guardrails have different functions. 111 00:08:12,536 --> 00:08:17,696 One is to make sure that there's no inappropriate or developmentally 112 00:08:17,696 --> 00:08:20,576 inappropriate outputs by the ai. 113 00:08:21,056 --> 00:08:27,536 But another purpose of these guardrails is to set up the large language model 114 00:08:27,566 --> 00:08:32,216 that's running them in the background to be in that Socratic style, to 115 00:08:32,216 --> 00:08:35,396 guide, not give the answer essentially. 116 00:08:35,396 --> 00:08:41,456 That is kind of just a little bit of a look at that Stanford research 117 00:08:41,456 --> 00:08:48,626 summary, which I think gives us a lot to think about and really matches what my 118 00:08:48,626 --> 00:08:52,646 experience has been and kind of what I shared in that last episode all about. 119 00:08:52,766 --> 00:08:53,786 It's all about the tool. 120 00:08:53,786 --> 00:08:54,686 It's not black or white. 121 00:08:54,956 --> 00:08:56,336 It's not give or don't give. 122 00:08:56,546 --> 00:09:01,556 It's give, but with a giant, but, and the tools are. 123 00:09:01,901 --> 00:09:04,841 Very, very important, but also how you use them. 124 00:09:04,841 --> 00:09:08,351 So the tools, having the guardrails, the teacher dashboard, but also kind of how 125 00:09:08,351 --> 00:09:12,011 you use them in an instructional setting is what I shared in that last episode. 126 00:09:12,161 --> 00:09:16,721 Based on this idea that tool design matters, I created some 127 00:09:16,781 --> 00:09:22,391 dos and don'ts when using AI for both teachers and or parents. 128 00:09:22,536 --> 00:09:26,891 And I know many of you are teachers and parents, so here we go. 129 00:09:27,731 --> 00:09:32,501 I don't like to start with a don't, but it just really made 130 00:09:32,501 --> 00:09:34,121 sense to start with a don't here. 131 00:09:34,151 --> 00:09:36,011 This one's for teachers and parents. 132 00:09:36,281 --> 00:09:41,621 I do not want you to drop your students into an AI model, AI tool 133 00:09:42,011 --> 00:09:49,271 without any foundational AI literacy lessons and or conversations. 134 00:09:49,391 --> 00:09:53,591 Lessons can really just be conversations, and in this case, when it comes to 135 00:09:53,591 --> 00:09:54,701 AI literacy, they really should. 136 00:09:55,121 --> 00:09:57,431 Be a lot of conversations around it. 137 00:09:57,431 --> 00:10:04,241 So before the tool, the AI model, the AI tool ever goes in students' hands. 138 00:10:04,601 --> 00:10:10,331 They need to know that this is an ai, this is not a person, it is not your friend. 139 00:10:10,901 --> 00:10:12,821 I'm gonna talk a little bit more about that in a minute. 140 00:10:12,821 --> 00:10:16,771 That's an important one for all ages of kids and adults. 141 00:10:17,761 --> 00:10:18,871 Pause, I'll come back to that. 142 00:10:19,091 --> 00:10:21,611 , They need to know, here's what it can do. 143 00:10:21,821 --> 00:10:23,411 Here's what it can't do. 144 00:10:23,861 --> 00:10:25,391 Here's how it can help you. 145 00:10:25,451 --> 00:10:27,071 Here's how it can't help you. 146 00:10:27,191 --> 00:10:30,851 Here's why you should question it and push back. 147 00:10:31,301 --> 00:10:33,851 I've been teaching and talking about that a lot lately. 148 00:10:34,211 --> 00:10:37,121 We need to learn to push back on ai. 149 00:10:37,241 --> 00:10:38,261 It's not magical. 150 00:10:38,261 --> 00:10:40,721 It's not always right far from it. 151 00:10:40,751 --> 00:10:42,191 We know it hallucinates, right? 152 00:10:42,191 --> 00:10:46,331 So knowing when to push back on it and in appropriate ways 153 00:10:46,331 --> 00:10:47,681 is a really important part. 154 00:10:48,221 --> 00:10:49,481 Of that conversation. 155 00:10:49,561 --> 00:10:54,181 Now, you might be thinking, I don't feel that I am equipped or 156 00:10:54,211 --> 00:10:56,701 ready, or, I'm not a tech expert. 157 00:10:56,701 --> 00:11:00,661 I'm not an AI expert, so I can't have these conversations. 158 00:11:00,661 --> 00:11:03,931 But that is not the case at all. 159 00:11:04,231 --> 00:11:09,661 So much of AI literacy is having the conversations and just 160 00:11:09,661 --> 00:11:11,821 talking about all those things. 161 00:11:11,821 --> 00:11:14,851 I just gave you what it can do, what it can't do, what it is, what it is not. 162 00:11:15,841 --> 00:11:18,691 You just need to model the right questions. 163 00:11:19,261 --> 00:11:24,781 Sometimes it's presenting and taking advantage of like a teachable moment 164 00:11:24,781 --> 00:11:26,491 to have the conversations too. 165 00:11:26,821 --> 00:11:28,021 So if you have. 166 00:11:28,411 --> 00:11:33,631 Surrey on your phone if you have a, I can't say it, A-L-E-A-L-E-X-A 167 00:11:33,631 --> 00:11:35,851 or Google in your house. 168 00:11:35,851 --> 00:11:38,401 Hey Google machine in your house, AI device. 169 00:11:38,701 --> 00:11:43,711 Use those as a way to kind of spark these conversations. 170 00:11:43,711 --> 00:11:45,511 Hey, you know, I see you're talking to Surrey. 171 00:11:45,511 --> 00:11:48,571 Did you have, you thought about what Surrey really is, right? 172 00:11:48,841 --> 00:11:53,791 You understand, right that Sury is just an ai, not your friend, not a human, right? 173 00:11:53,941 --> 00:11:55,531 Like just an example there. 174 00:11:56,531 --> 00:12:03,221 I was working with a teacher with elementary age students and we were 175 00:12:03,221 --> 00:12:08,441 using a guard railed AI tool and we had had all these conversations with 176 00:12:08,441 --> 00:12:10,481 kids, with the kids in the classroom. 177 00:12:10,631 --> 00:12:11,621 Mostly the teacher. 178 00:12:12,161 --> 00:12:14,676 I was just kind of there supporting and uh. 179 00:12:15,446 --> 00:12:21,746 They started using the ai and of course one student in the 180 00:12:21,746 --> 00:12:24,566 room said, it's my friend. 181 00:12:24,566 --> 00:12:29,036 And then you know how something like that spreads like wildfire in a classroom. 182 00:12:29,036 --> 00:12:30,476 Then all of a sudden, oh, that's my friend. 183 00:12:31,196 --> 00:12:38,936 So we were able to use that as a teachable moment and we paused the tool. 184 00:12:38,966 --> 00:12:40,836 'cause that's another thing that these guardrail tools 185 00:12:40,836 --> 00:12:43,056 allow you to do is hit pause, 186 00:12:43,156 --> 00:12:50,356 and brought everybody together and we created a kind of t chart of what our 187 00:12:50,356 --> 00:12:54,616 characteristics of a friend and what are not characteristics of a friend. 188 00:12:54,886 --> 00:12:55,276 Okay? 189 00:12:55,305 --> 00:13:00,616 And we let the kids build the list and by the end they were 190 00:13:00,616 --> 00:13:02,566 going, oh, you are right. 191 00:13:02,626 --> 00:13:05,055 I don't, I don't think this AI is my friend. 192 00:13:05,156 --> 00:13:07,406 That was an example of a perfect, teachable moment 193 00:13:07,406 --> 00:13:08,576 of instead of that teacher. 194 00:13:09,176 --> 00:13:12,116 Just saying, Hey, nope, AI is not your friend. 195 00:13:12,566 --> 00:13:16,136 That that was the natural inclination, that anthropomorphism. 196 00:13:16,316 --> 00:13:20,816 So stopping, having that teachable moment and letting students think 197 00:13:20,816 --> 00:13:25,616 through it and come to that healthy, responsible, ethical conclusion. 198 00:13:25,616 --> 00:13:31,236 Their self, moral of the story, curiosity, healthy skepticism, 199 00:13:31,236 --> 00:13:35,796 conversations, discussions, those are the foundation of AI literacy. 200 00:13:36,096 --> 00:13:39,486 Not being an AI expert, not being a tech expert. 201 00:13:39,516 --> 00:13:46,356 You already have the skillset you need to know to question it and just 202 00:13:46,356 --> 00:13:48,066 start having those conversations. 203 00:13:49,066 --> 00:13:49,456 Okay. 204 00:13:49,516 --> 00:13:55,516 I said I would come back to this idea of companionship. 205 00:13:55,816 --> 00:14:02,476 This is becoming companionship with AI, in particular from humans to ai and humans 206 00:14:02,476 --> 00:14:09,406 developing really unhealthy relationships with AI and thinking that they are a 207 00:14:09,406 --> 00:14:11,446 friend, sometimes more than a friend. 208 00:14:11,656 --> 00:14:17,356 We're seeing this in all ages, kids to adults. 209 00:14:17,596 --> 00:14:21,046 You've likely seen something about this in the media. 210 00:14:21,496 --> 00:14:24,556 It is a major concern, but here's the problem. 211 00:14:24,946 --> 00:14:29,206 The knee jerk reaction is to say, as a parent especially, 212 00:14:29,206 --> 00:14:31,666 and a teacher, is to say, oh. 213 00:14:32,341 --> 00:14:34,141 Nope, we are not touching that. 214 00:14:34,231 --> 00:14:35,881 We're not touching that technology. 215 00:14:35,881 --> 00:14:36,961 Nope, nope, nope, nope, nope. 216 00:14:36,961 --> 00:14:40,741 We're just not even gonna let it in our lives, and that's gonna shut down 217 00:14:41,071 --> 00:14:45,781 the option of that happening or the chance of that happening, I should say. 218 00:14:46,781 --> 00:14:52,031 That's not how I want you as a teacher and or a parent to react to this. 219 00:14:53,001 --> 00:14:54,411 . I know that seems crazy. 220 00:14:54,411 --> 00:14:58,101 You're like, Lindy, you just told us this is a major concern and it's a problem. 221 00:14:58,551 --> 00:15:02,541 Why wouldn't we just block it no more? 222 00:15:03,541 --> 00:15:10,156 The problem is artificial intelligence is embedded in everything in our 223 00:15:10,156 --> 00:15:18,046 lives already, and it's not just technology, so not allowing the. 224 00:15:18,436 --> 00:15:23,236 A-L-E-X-A devices in your house or not allowing students into a large language 225 00:15:23,236 --> 00:15:29,956 model on a device, a laptop, a phone, doesn't mean that they're not going to 226 00:15:29,956 --> 00:15:33,226 be interacting with this technology. 227 00:15:33,316 --> 00:15:39,351 I always use the example I of walking down the street, you're likely having 228 00:15:39,676 --> 00:15:43,336 your face scanned by a video camera. 229 00:15:43,756 --> 00:15:47,716 And that video has AI built into it. 230 00:15:47,956 --> 00:15:53,216 That , has facial detection, that's ai and that is just walking down the street. 231 00:15:55,196 --> 00:16:00,326 So this is no longer like, oh, we can just limit a device. 232 00:16:00,356 --> 00:16:05,876 We can limit screen time and we can keep our kids safe from this technology. 233 00:16:06,416 --> 00:16:08,636 And this is just the beginning of it. 234 00:16:09,026 --> 00:16:14,846 So instead, you need to take the approach of setting up. 235 00:16:15,806 --> 00:16:22,766 Responsible, healthy interactions with specialized tools. 236 00:16:22,826 --> 00:16:24,716 And I'm gonna talk more about that here in a second. 237 00:16:25,716 --> 00:16:31,116 And having the conversations, and honestly, even if you don't wanna touch 238 00:16:31,236 --> 00:16:36,186 the technology, okay, that's okay, but you still have to have the conversation. 239 00:16:36,186 --> 00:16:40,926 You can't act like it doesn't exist because your kid, at some point in their 240 00:16:40,926 --> 00:16:47,436 life is going to come across an ai and you want them to understand that that 241 00:16:47,496 --> 00:16:52,806 AI is not human and you don't want to develop companionship with that ai. 242 00:16:53,226 --> 00:16:53,526 Okay. 243 00:16:53,676 --> 00:16:54,786 Enough babbling about that. 244 00:16:54,786 --> 00:16:58,566 But hopefully that babbling led to you understanding that 245 00:16:58,896 --> 00:17:01,566 companionship is 100% a major concern. 246 00:17:02,076 --> 00:17:05,466 Another reason for guardrails is to help with that, by the way. 247 00:17:05,466 --> 00:17:07,326 I'm not sure I said that clearly. 248 00:17:07,596 --> 00:17:11,916 And the kneejerk reaction is just to not let your kids touch it. 249 00:17:12,006 --> 00:17:17,556 But that's actually the better way to go about it, is to teach 250 00:17:17,556 --> 00:17:19,986 them responsible healthy use. 251 00:17:20,016 --> 00:17:24,486 Whether that is just through having conversations or ideally a 252 00:17:24,486 --> 00:17:29,166 combination of having conversations, interacting with some very 253 00:17:29,166 --> 00:17:31,836 specialized tools and technologies. 254 00:17:32,076 --> 00:17:35,196 And the great thing about that is it sparks the conversations, 255 00:17:35,556 --> 00:17:37,116 just like that example I gave you. 256 00:17:38,106 --> 00:17:44,076 Alright, so that is a don't, don't put your students kids 257 00:17:44,526 --> 00:17:51,576 into AI models, AI tools without first having some conversations. 258 00:17:51,906 --> 00:17:56,716 Those AI literacy conversations are so important and I should say. 259 00:17:57,061 --> 00:18:01,381 If you are getting them into these tools, don't just let it in there. 260 00:18:01,471 --> 00:18:04,171 Use them to spark those conversations. 261 00:18:04,621 --> 00:18:06,181 Okay, now let's go to a do. 262 00:18:06,211 --> 00:18:07,651 This one is for teachers. 263 00:18:07,921 --> 00:18:12,331 I want you to use tools that have guardrails and teacher 264 00:18:12,331 --> 00:18:16,321 dashboards to give you full visibility into students' progress. 265 00:18:17,821 --> 00:18:23,941 These tools use the same underlying AI models like the GPT models from chat, GPT, 266 00:18:24,181 --> 00:18:29,221 the Claude models, the Gemini models, but they have this educational scaffolding 267 00:18:29,221 --> 00:18:30,841 kind of layered on top of them. 268 00:18:31,351 --> 00:18:36,811 I gave you six of these tools in the last episode and talked much more 269 00:18:36,811 --> 00:18:40,411 in depth about the guardrails and the dashboard and the instructional 270 00:18:40,411 --> 00:18:41,471 moments and when and all of that. 271 00:18:42,646 --> 00:18:46,546 There are a few caveats here with older kids. 272 00:18:46,546 --> 00:18:51,106 There might be specific use cases, likely will be specific use cases, 273 00:18:51,106 --> 00:18:56,896 I should say, where you could use a Gemini gym if you have access to Gemini. 274 00:18:57,226 --> 00:19:02,386 And if you aren't familiar with a gym, it's really similar, pretty much the same 275 00:19:02,416 --> 00:19:05,956 as a custom GPT if you're a chat GPT user. 276 00:19:06,136 --> 00:19:13,726 And what gyms and custom GPTs do is they allow you to basically in the 277 00:19:13,726 --> 00:19:20,596 background prompt the AI to have a specific goal or have specific 278 00:19:20,596 --> 00:19:22,936 knowledge base that it's pulling from. 279 00:19:22,936 --> 00:19:29,926 So you can add files to a gym and a custom GPT so that it's pulling from those files. 280 00:19:29,926 --> 00:19:33,796 So imagine like a chapter from a textbook or something like 281 00:19:33,796 --> 00:19:35,926 that, that students are learning. 282 00:19:36,166 --> 00:19:38,356 That's something that you can load in the backend and it's 283 00:19:38,356 --> 00:19:39,946 going to use that information. 284 00:19:40,336 --> 00:19:48,316 To give its outputs essentially, so gyms, I would be hesitant to 285 00:19:48,316 --> 00:19:51,346 use custom GPTs with students. 286 00:19:51,796 --> 00:19:55,696 Maybe if you have GPT for education, you would wanna make sure all the data 287 00:19:55,696 --> 00:19:57,496 privacy and check for check is all good. 288 00:19:57,766 --> 00:20:01,426 But if you're a Google workspace for education school, Gemini 289 00:20:01,426 --> 00:20:03,106 is part of your core service. 290 00:20:03,136 --> 00:20:08,876 It is under the umbrella of your , data privacy agreement with Google, et cetera. 291 00:20:08,876 --> 00:20:13,496 So it's much cleaner in terms of compliance, if you may. 292 00:20:14,981 --> 00:20:20,096 That would be the way that I would go a hundred times over, over a custom GPT. 293 00:20:20,246 --> 00:20:25,016 But even if you're using a gym, you've got all that compliance 294 00:20:25,016 --> 00:20:28,386 checked, they're still not gonna give you that visibility that you need. 295 00:20:29,411 --> 00:20:34,661 You're not gonna get summary of insights and where to take the instruction next. 296 00:20:34,661 --> 00:20:39,581 So again, this is only gonna be used with older students and 297 00:20:39,581 --> 00:20:45,221 really sparingly and only in really specific instructional use cases. 298 00:20:45,701 --> 00:20:48,281 Now parents, let's talk about you. 299 00:20:48,701 --> 00:20:53,411 There are some freemium guardrail tools that you can use with your kids. 300 00:20:53,621 --> 00:20:59,891 I would start with either school AI spaces or Brisk Boost, and that will 301 00:20:59,891 --> 00:21:02,621 allow you to do, and you'll just have one. 302 00:21:03,011 --> 00:21:05,981 Kid, maybe two, three, how many kids you have in there? 303 00:21:06,041 --> 00:21:12,431 Likely one or two in that space or that brisk boost session interacting with it. 304 00:21:12,761 --> 00:21:16,391 So that's a really great option, but since you are working at more of 305 00:21:16,391 --> 00:21:21,161 a one-to-one or a few to one ratio compared to teachers that have more 306 00:21:21,161 --> 00:21:26,821 of like a 30 to one ratio, you have a couple of other options that are a lot 307 00:21:26,821 --> 00:21:30,660 more challenging to do in classroom environments that you can do with 308 00:21:30,660 --> 00:21:33,150 only one or two or three kids at home. 309 00:21:34,021 --> 00:21:37,531 Gyms and custom gpt, I already mentioned those. 310 00:21:37,771 --> 00:21:40,860 That's gonna be a great strategy for you to set up. 311 00:21:40,860 --> 00:21:45,241 Let's say your kid is trying to work through their science homework. 312 00:21:45,271 --> 00:21:49,201 You could set up a gym or a custom GPT saying this is their 313 00:21:49,201 --> 00:21:50,701 age, this is what they do. 314 00:21:50,701 --> 00:21:53,761 You can even load specific information in it, like I said, 315 00:21:53,971 --> 00:21:55,261 and it's gonna pull from that. 316 00:21:55,261 --> 00:22:00,390 And you're kind of like prompting the AI to guide that learning experience. 317 00:22:01,261 --> 00:22:03,900 Claude and I think chat GBT too. 318 00:22:03,931 --> 00:22:04,291 Yeah. 319 00:22:04,441 --> 00:22:08,851 Both have what are called projects and what they allow you to do is 320 00:22:09,120 --> 00:22:15,991 give instructions and or files, and then you can have multiple chats 321 00:22:15,991 --> 00:22:20,370 within that project that have the context of the instructions and 322 00:22:20,370 --> 00:22:22,351 the files that it's pulling from. 323 00:22:22,711 --> 00:22:26,811 Perplexity has something called spaces that functions very similarly. 324 00:22:26,811 --> 00:22:31,371 It's basically the same thing as projects in Claude and Chat GPT. 325 00:22:32,371 --> 00:22:38,431 The other thing you can do, which is similar but not as repeatable as a gym or 326 00:22:38,431 --> 00:22:44,461 a custom GPT or a project or a space, is you can set up a general purpose chat bot. 327 00:22:44,461 --> 00:22:48,601 So chat, GPT, Jim and I clawed, likely perplexity, maybe especially if you're 328 00:22:48,601 --> 00:22:54,181 doing more research stuff, you can set one of those up to guide, not give. 329 00:22:54,481 --> 00:22:57,031 By priming the chat with information. 330 00:22:57,481 --> 00:23:04,141 So for example, I would open up a new chat in one of those tools, large language 331 00:23:04,141 --> 00:23:06,781 model tools, and give it a prompt. 332 00:23:06,811 --> 00:23:10,711 Something like this, you are helping a 14-year-old 333 00:23:10,711 --> 00:23:12,841 complete school learning tasks. 334 00:23:13,141 --> 00:23:15,841 I want you to guide and not give the answers. 335 00:23:16,081 --> 00:23:19,471 You can break down the steps for completion, help brainstorm 336 00:23:19,471 --> 00:23:23,161 ideas, give steps to help complete the problem, et cetera. 337 00:23:23,761 --> 00:23:27,121 Ensure outputs are developmentally appropriate to age 14. 338 00:23:27,871 --> 00:23:28,261 Okay? 339 00:23:28,351 --> 00:23:33,601 So now one thing to understand here is when you go into a chat in any of these 340 00:23:33,601 --> 00:23:37,861 large language model tools, chat, GBT, Claude Gemini, are the three biggest ones. 341 00:23:39,421 --> 00:23:44,911 Every time you hit enter on a new prompt, it rereads the entire chat. 342 00:23:45,736 --> 00:23:46,456 To a point. 343 00:23:46,546 --> 00:23:46,846 It does. 344 00:23:46,876 --> 00:23:50,146 There does come a point where you hit what's called a context window limitation, 345 00:23:50,356 --> 00:23:52,156 where it can only go back so far. 346 00:23:52,396 --> 00:23:57,346 But if you haven't overloaded the chat with a lot of stuff, basically it's 347 00:23:57,346 --> 00:23:59,236 gonna reread that chat every time. 348 00:23:59,446 --> 00:24:04,576 So every time your kid is in there asking it things, Hey, help me with this. 349 00:24:04,576 --> 00:24:05,596 Help me with this problem. 350 00:24:05,596 --> 00:24:06,556 I'm working through this. 351 00:24:06,606 --> 00:24:10,146 It's gonna reread this prompt that I just gave you, and it's gonna know. 352 00:24:10,146 --> 00:24:14,676 I gotta make sure that I'm giving outputs that are appropriate to a 14-year-old. 353 00:24:14,706 --> 00:24:19,031 I've gotta make sure that I guide, not give, . This is essentially what 354 00:24:19,031 --> 00:24:22,586 a custom GBT and a gym and a project. 355 00:24:22,961 --> 00:24:26,021 Are doing, but just within one chat. 356 00:24:26,291 --> 00:24:29,891 And if the idea of creating a gym or a custom GPT or a project, 357 00:24:29,891 --> 00:24:31,031 you're like, oh, I'm not there yet. 358 00:24:31,271 --> 00:24:34,511 If you're newer to these tools, this is a really great way to start. 359 00:24:34,751 --> 00:24:37,211 Just open up a chat, give it a prompt like this. 360 00:24:37,571 --> 00:24:41,471 But this is similar to kind of what the prompt could be on the back end of a 361 00:24:41,471 --> 00:24:44,111 custom GPT or a gym, if that makes sense. 362 00:24:45,341 --> 00:24:48,791 Lemme give you one other example, and I wanna do this one for a writing 363 00:24:48,791 --> 00:24:53,141 task, since that is the biggest pain point when it comes to the copy 364 00:24:53,141 --> 00:24:56,651 paste idea of large language models. 365 00:24:56,651 --> 00:24:57,551 So here's another one. 366 00:24:57,971 --> 00:25:00,401 You are a homework helper for a fifth grader. 367 00:25:00,641 --> 00:25:02,981 Do not give answers directly. 368 00:25:03,191 --> 00:25:07,631 Instead, ask questions that help them think through their own response. 369 00:25:08,021 --> 00:25:11,561 If they ask you to write something for them, redirect them to try 370 00:25:11,561 --> 00:25:13,751 a first draft themselves first. 371 00:25:14,111 --> 00:25:16,961 The assignment is a blah, blah, blah, blah, blah. 372 00:25:17,351 --> 00:25:22,261 A book report on this title, a essay on this topic, . There's another example of 373 00:25:22,261 --> 00:25:27,571 what you could do within a chat or within like a custom GPT or a gym or a project. 374 00:25:28,441 --> 00:25:31,111 Now the AI has the context. 375 00:25:31,831 --> 00:25:35,041 I would test this in different tools. 376 00:25:35,731 --> 00:25:39,811 I constantly go between those four. 377 00:25:39,816 --> 00:25:43,531 Chat should be Gemini, Claude Perplexity, well, chate, Gemini, Claude. 378 00:25:43,531 --> 00:25:46,501 I constantly go between the three of them and almost always have 379 00:25:46,501 --> 00:25:48,721 all three open in my windows. 380 00:25:48,871 --> 00:25:52,921 Sometimes multiple different tabs open with multiple different versions of those. 381 00:25:53,171 --> 00:25:54,101 Perplexity. 382 00:25:54,101 --> 00:25:58,031 I use it very specific times, usually related to research type stuff. 383 00:25:58,136 --> 00:26:00,371 I'll talk a little bit more about that in a bit. 384 00:26:01,421 --> 00:26:07,241 So test it in the different ones, the different AI tools, general purpose, 385 00:26:07,361 --> 00:26:10,991 large language models, essentially, and see if one does better than another. 386 00:26:11,286 --> 00:26:15,936 And try to act like your kid and see what happens and see if you kind of 387 00:26:15,936 --> 00:26:17,616 gravitate towards one or another. 388 00:26:17,706 --> 00:26:23,916 Also, test different models in these different general purpose, 389 00:26:23,916 --> 00:26:25,596 large language models, tools. 390 00:26:25,956 --> 00:26:28,146 So this is something that a lot of people don't understand, especially 391 00:26:28,146 --> 00:26:32,166 if you're only using the free versions of these tools, especially chatt PT, 392 00:26:32,166 --> 00:26:35,946 the free version of Chatt PT, at least the last time I was in there, I don't 393 00:26:35,946 --> 00:26:40,356 think this has changed, only gives you its default model, so you don't 394 00:26:40,356 --> 00:26:42,156 even have the option to switch models. 395 00:26:42,516 --> 00:26:49,686 I am pretty sure the free version of Claude lets you change between the 396 00:26:49,686 --> 00:26:51,516 three different models that they have. 397 00:26:51,606 --> 00:26:56,346 I'm fairly certain, okay, so if you're on the free version of trash, 398 00:26:56,346 --> 00:26:59,916 GBT and you're not willing to pay, by the way, I don't wanna go there, 399 00:27:00,006 --> 00:27:05,186 but you're gonna get so much out of these tools if you pay for them. 400 00:27:06,156 --> 00:27:07,596 Maybe I'll do a whole episode on that. 401 00:27:08,236 --> 00:27:13,726 So go into Claude if you're on the free version of chat, TBT, and in 402 00:27:13,726 --> 00:27:16,396 the chat window on the bottom right, you see a little dropdown, it'll 403 00:27:16,396 --> 00:27:18,406 always default to the Sonet model. 404 00:27:18,616 --> 00:27:21,346 If you click that, it also gives you two other models. 405 00:27:21,376 --> 00:27:22,876 Sonet is their default. 406 00:27:22,876 --> 00:27:26,956 That's kind of like the go-to for anything and everything. 407 00:27:27,016 --> 00:27:31,186 Um, I always call it kind of like my, my go-to default. 408 00:27:31,636 --> 00:27:35,656 If I'm not doing anything like particularly specific in my prompt. 409 00:27:36,646 --> 00:27:39,196 Haiku is a faster model. 410 00:27:39,196 --> 00:27:40,276 I never use it. 411 00:27:41,746 --> 00:27:46,566 Opus is their reasoning model and reasoning models. 412 00:27:46,566 --> 00:27:49,746 Essentially, they use what are called chain of thought prompting 413 00:27:49,746 --> 00:27:56,916 in the background to work through their outputs in more detail. 414 00:27:57,606 --> 00:28:01,866 It takes longer to get a response out of a reasoning model like Opus. 415 00:28:02,436 --> 00:28:05,136 Then it does a default model like sonnet. 416 00:28:05,536 --> 00:28:10,966 But you usually get much more detailed responses and it lowers 417 00:28:10,966 --> 00:28:14,446 the chance of hallucinations because it's using that chain of 418 00:28:14,446 --> 00:28:16,276 prompting to kind of check itself. 419 00:28:16,976 --> 00:28:18,896 I'm gonna make sure I'm giving you the right information here. 420 00:28:19,106 --> 00:28:22,466 So, Claude, yes, you have sonnets are default. 421 00:28:22,466 --> 00:28:25,886 It says for most efficient, for everyday tasks. 422 00:28:25,886 --> 00:28:26,891 It actually tells you that. 423 00:28:27,431 --> 00:28:33,281 When you click the dropdown, then you have Haiku says, fastest for quick answers me. 424 00:28:33,791 --> 00:28:37,751 And then Opus is most capable for ambitious work. 425 00:28:37,961 --> 00:28:40,901 So it doesn't state that it's a reasoning model, but from my 426 00:28:40,901 --> 00:28:43,661 understanding, it is a reasoning model. 427 00:28:43,781 --> 00:28:44,141 Okay. 428 00:28:44,351 --> 00:28:51,281 Then Jim and I has their, they call their models and it's in the same spot as 429 00:28:51,281 --> 00:28:55,031 Claude in the chat window where you put your prompt in the bottom right corner. 430 00:28:55,031 --> 00:28:58,911 If you click that, it'll likely default to fast. 431 00:28:59,391 --> 00:29:06,531 So it says fast answers, quickly thinking, solves complex problems. 432 00:29:06,531 --> 00:29:11,301 They also have a pro model, which is really good with advanced math. 433 00:29:11,361 --> 00:29:14,901 So if you have a high school math student, or you're a high school math teacher, 434 00:29:15,201 --> 00:29:17,391 definitely check that pro model out. 435 00:29:17,661 --> 00:29:19,341 And it's also their coding model. 436 00:29:20,361 --> 00:29:23,391 And then chat GPT at the moment. 437 00:29:23,856 --> 00:29:24,336 They quit. 438 00:29:24,336 --> 00:29:29,376 , Notice that Gemini quit numbering and really naming their models. 439 00:29:30,366 --> 00:29:33,246 Claude is still naming them and numbering them. 440 00:29:34,506 --> 00:29:39,876 Sonnet, Opus Haiku and they each, I think we're at 4.6 right now, 4.6 441 00:29:39,876 --> 00:29:41,856 for Opus and Sonnet, 4.5 for Haiku. 442 00:29:42,156 --> 00:29:47,526 Gemini is just calling it Fast Thinking Pro and then chat, CBT. 443 00:29:47,526 --> 00:29:51,816 We have instant that's for everyday chats and thinking. 444 00:29:52,446 --> 00:29:54,906 That's for complex questions. 445 00:29:55,446 --> 00:30:00,576 So that was a little bit of a tangent sidebar there, but I think an important 446 00:30:00,576 --> 00:30:05,346 one to understand what I mean by go test different models and different 447 00:30:05,346 --> 00:30:11,136 tools to see which one is going to be the best for whatever task you're 448 00:30:11,136 --> 00:30:13,956 trying to help your kid with at home. 449 00:30:15,126 --> 00:30:20,436 If you're trying to help them with advanced math, absolutely do not use the 450 00:30:20,441 --> 00:30:22,446 default models in any of these tools. 451 00:30:22,956 --> 00:30:25,926 It's not gonna likely be a whole lot of help, and it's probably 452 00:30:25,926 --> 00:30:27,666 gonna be full of hallucinations. 453 00:30:28,666 --> 00:30:36,046 Likely you never know, but I feel like the likelihood is pretty, pretty high. 454 00:30:37,046 --> 00:30:41,336 So I kind of summarize that the default models in any of those tools 455 00:30:41,396 --> 00:30:47,126 are going to give really quick, not super thorough answers, but 456 00:30:47,156 --> 00:30:48,836 that's sometimes what you want. 457 00:30:49,886 --> 00:30:53,846 The reasoning models often called the thinking models are gonna take 458 00:30:53,846 --> 00:30:57,086 longer to answer, but typically have more thorough answers with 459 00:30:57,096 --> 00:30:58,896 , less chance of hallucinations. 460 00:30:59,896 --> 00:31:03,646 Just in general, the Claude models tend to be more creative 461 00:31:03,976 --> 00:31:05,956 and better helping with writing. 462 00:31:06,196 --> 00:31:11,026 So anytime I am ideating and wanting AI to help me think through an 463 00:31:11,026 --> 00:31:16,096 idea or build on that idea or help me brainstorm, I'm almost always 464 00:31:16,096 --> 00:31:18,976 gonna go use one of the CLA models. 465 00:31:19,426 --> 00:31:23,446 Depending upon exactly what type of ideating I'm doing, I'd probably start 466 00:31:23,446 --> 00:31:27,736 with the sonnet models, and then maybe once I narrowed it down to one or two 467 00:31:27,736 --> 00:31:32,056 ideas, I might then go into the Opus model and dig a little bit deeper with it. 468 00:31:32,746 --> 00:31:39,976 Anytime I want AI to help me finesse my writing, maybe reword this a little bit. 469 00:31:40,096 --> 00:31:46,456 I always go into the Claude models 100% over chat, GBT, I would say Gemini's 470 00:31:46,456 --> 00:31:50,446 kind of in the middle there and depends a little bit on what type of writing. 471 00:31:51,616 --> 00:31:54,616 Gemini's pretty good at more of the technical writing. 472 00:31:55,616 --> 00:32:00,926 Now if you're at a school that has Google Workspace for education, 473 00:32:01,766 --> 00:32:02,636 you're gonna have Gemini. 474 00:32:02,636 --> 00:32:03,986 It's part of your core service. 475 00:32:04,046 --> 00:32:06,146 That means it's compliant, all that good stuff. 476 00:32:06,386 --> 00:32:13,406 It has some amazing multimodal opportunities that the other. 477 00:32:13,916 --> 00:32:17,696 Large language model tools, chat, pd, Claude Perplexity don't have. 478 00:32:18,296 --> 00:32:27,146 So for example, you can click a button, you can add the nano banana model, which 479 00:32:27,146 --> 00:32:32,846 is Gemini's image Generation model is really good at creating infographic type 480 00:32:32,906 --> 00:32:37,856 visuals so you can turn something they're trying to learn into an infographic. 481 00:32:38,276 --> 00:32:44,306 Notebook, lm, I'm gonna talk a little bit more about this in a bit, but 482 00:32:44,456 --> 00:32:49,766 Notebook LM is amazing and has all sorts of multimodal stuff in it. 483 00:32:50,036 --> 00:32:55,106 So more of the story if you're at a Google school, if your kid's at a Google school. 484 00:32:55,886 --> 00:33:02,336 Jim and I has a ton of power and integrates really well into like Google 485 00:33:02,336 --> 00:33:04,016 Docs and the whole Google ecosystem. 486 00:33:05,016 --> 00:33:05,886 Alright, so to wrap that up. 487 00:33:07,431 --> 00:33:10,431 Remember, don't do any of that without first having those 488 00:33:10,431 --> 00:33:12,231 AI literacy conversations. 489 00:33:13,231 --> 00:33:15,211 Now for a don't. 490 00:33:15,271 --> 00:33:16,771 This one's for teachers and parents. 491 00:33:16,981 --> 00:33:20,311 I don't want you to drop students into one of those general purpose 492 00:33:20,311 --> 00:33:22,711 chatbots with no structure. 493 00:33:23,071 --> 00:33:27,301 Don't put your students and or kids into a large language model with no 494 00:33:27,301 --> 00:33:32,971 prompting or priming or those guardrails set up without any of that set up. 495 00:33:32,971 --> 00:33:37,711 The AI is going to default to answer mode and answer mode is what makes it super 496 00:33:37,711 --> 00:33:39,871 easy for them to outsource their thinking, 497 00:33:41,011 --> 00:33:42,511 especially for young people. 498 00:33:43,511 --> 00:33:47,741 Alright, now a do, this is for teachers and parents. 499 00:33:47,921 --> 00:33:48,791 I mentioned it. 500 00:33:48,911 --> 00:33:50,711 Notebook, LM Notebook. 501 00:33:50,711 --> 00:33:51,941 LM is a big do. 502 00:33:51,941 --> 00:33:53,351 It is so, so, so good. 503 00:33:53,351 --> 00:33:55,521 So if you're not familiar, , it's made by Google. 504 00:33:55,671 --> 00:33:58,431 So it's part of the Google ecosystem and it's built around 505 00:33:58,431 --> 00:34:00,231 Gemini, the Gemini models. 506 00:34:00,711 --> 00:34:08,271 And essentially it allows you to give it sources of information. 507 00:34:08,271 --> 00:34:13,341 So the, that can be a website, it can be a document, it could be a research study, it 508 00:34:13,341 --> 00:34:15,351 could be a Google doc that you've created. 509 00:34:15,981 --> 00:34:19,341 And then it pulls information. 510 00:34:19,431 --> 00:34:25,311 And in the chat, when you're chatting with the ai, it pulls its responses from 511 00:34:25,311 --> 00:34:29,901 those sources and even cites those sources of where it came up with the response. 512 00:34:30,576 --> 00:34:35,586 This is called a RAG Model Retrieval Augmented Generation. 513 00:34:35,796 --> 00:34:36,396 There you go. 514 00:34:36,486 --> 00:34:38,046 There's a mouthful for you. 515 00:34:38,256 --> 00:34:43,896 This idea that you give the AI a knowledge base, and it's using all 516 00:34:43,896 --> 00:34:48,546 of its large language model knowledge and training data, essentially, 517 00:34:49,926 --> 00:34:55,536 but it's specifically pulling its answers from that knowledge base. 518 00:34:55,896 --> 00:34:56,256 Okay? 519 00:34:56,286 --> 00:34:57,546 So that's what Notebook LM does. 520 00:34:57,756 --> 00:35:06,816 So it's a really great way to teach students how to research and check and 521 00:35:06,816 --> 00:35:11,291 verify information from large language models, but making sure the responses 522 00:35:11,351 --> 00:35:15,516 are specifically coming from the sources of information that you give it. 523 00:35:16,516 --> 00:35:20,386 It also makes this multimodal learning a breeze. 524 00:35:20,386 --> 00:35:22,756 So if you're in regular Gemini, you can use nano banana. 525 00:35:23,341 --> 00:35:25,801 You can tell it to create an infographic from the information. 526 00:35:26,011 --> 00:35:31,081 You can do that in Notebook L limb as well, and it's gonna take all of that 527 00:35:31,081 --> 00:35:34,861 source documentation, that knowledge base, and create an infographic. 528 00:35:34,861 --> 00:35:39,031 And there's actually a button that specifically says infographic. 529 00:35:39,361 --> 00:35:44,041 You can also create an audio summary that creates kind of a podcast 530 00:35:44,041 --> 00:35:48,011 style explanation of the sources. 531 00:35:48,011 --> 00:35:51,731 This is fantastic because I always say kids always have earbuds in 532 00:35:51,731 --> 00:35:53,381 their ears, so take advantage of it. 533 00:35:53,381 --> 00:35:53,831 Right? 534 00:35:54,141 --> 00:35:57,201 It'll create like two people talking through whatever content it is. 535 00:35:57,391 --> 00:36:02,101 It'll create a video of whatever the source knowledge is or 536 00:36:02,101 --> 00:36:03,091 whatever prompt you get it. 537 00:36:03,091 --> 00:36:06,301 It creates flashcards and so many other things. 538 00:36:06,331 --> 00:36:11,701 It is so, so good for both teachers and students. 539 00:36:12,701 --> 00:36:13,871 Teachers of course. 540 00:36:14,831 --> 00:36:18,011 You need to be at a Google school. 541 00:36:18,281 --> 00:36:22,331 If you're going to use this with your students, it could still be something 542 00:36:22,331 --> 00:36:26,651 that you can use as a teacher tool and maybe you use it to create an 543 00:36:26,651 --> 00:36:29,771 infographic and then share it in your learning management system. 544 00:36:29,821 --> 00:36:33,721 But you don't wanna put your students in notebook LM unless you are at a 545 00:36:33,721 --> 00:36:35,611 Google school that has approved it. 546 00:36:35,611 --> 00:36:36,451 All of the good stuff. 547 00:36:37,451 --> 00:36:41,261 Alright, another do, gonna do two dos back to back. 548 00:36:41,261 --> 00:36:45,641 And for this one I want to talk about research with one big caveat. 549 00:36:45,881 --> 00:36:51,671 I have an entire session that I do on basically the new research 550 00:36:51,821 --> 00:36:57,281 skillset with AI and large language models in particular, research is 551 00:36:57,281 --> 00:37:00,851 changing a lot and there's a lot that. 552 00:37:01,271 --> 00:37:06,911 Humans need to know how to do research using these large language models. 553 00:37:07,481 --> 00:37:11,441 I'm gonna pack the essentials of that session. 554 00:37:11,441 --> 00:37:11,981 I do. 555 00:37:12,221 --> 00:37:19,151 That's easily 1, 2, 3 hours into just a few minutes here, so I'll do my best 556 00:37:19,871 --> 00:37:24,971 if you teach high school and maybe, maybe, maybe, maybe, maybe middle school. 557 00:37:24,971 --> 00:37:25,661 It really depends. 558 00:37:25,661 --> 00:37:32,051 With middle schoolers on their AI literacy level, there are really specific use 559 00:37:32,051 --> 00:37:35,021 cases where teaching them how to research. 560 00:37:35,021 --> 00:37:42,671 Using these AI tools and large language models is really, really important and 561 00:37:42,671 --> 00:37:47,861 also can be really appropriate, especially for those high school aged students. 562 00:37:48,411 --> 00:37:52,971 This is an example of when you might want to put them into a general 563 00:37:52,971 --> 00:37:58,671 purpose, large language model, not necessarily a guard railed model with 564 00:37:58,671 --> 00:38:00,411 the teacher dashboard, all of that stuff. 565 00:38:00,441 --> 00:38:03,981 But I wouldn't just do this willy-nilly. 566 00:38:04,791 --> 00:38:07,641 This would be after tons of AI literacy conversations. 567 00:38:07,641 --> 00:38:11,271 It would be after I've used those guardrail tools with them, I would 568 00:38:11,271 --> 00:38:16,461 start with notebook L and teach them how to follow citations. 569 00:38:16,731 --> 00:38:22,611 So kind of like scaffolding in, and then before I put them into it, I would've 570 00:38:22,611 --> 00:38:27,951 put these large language models up on the big screen and walk through the process 571 00:38:27,951 --> 00:38:33,021 and talked and showed them how you verify information, how you follow cited sources, 572 00:38:33,231 --> 00:38:37,991 which tools have cited, sources, which tools don't have sided sources, et cetera. 573 00:38:38,541 --> 00:38:39,951 That's my big caveat there. 574 00:38:41,631 --> 00:38:48,171 I recommend either Perplexity or Gemini right now for research after notebook 575 00:38:48,171 --> 00:38:51,351 L. So Notebook L is number one, but the problem with notebook L is you have 576 00:38:51,351 --> 00:38:56,481 to have the right source information loaded into it, so it takes potentially 577 00:38:56,481 --> 00:38:59,061 a little bit of adult help to do that. 578 00:38:59,361 --> 00:39:04,451 If you wanna teach kids how to research and find appropriate source information 579 00:39:04,451 --> 00:39:08,171 and how to verify that's correct, then load into notebook themselves, 580 00:39:08,471 --> 00:39:09,821 that's where this comes into play. 581 00:39:10,181 --> 00:39:16,091 So Perplexity and Gemini as of today are the two of those four big tools. 582 00:39:16,091 --> 00:39:18,281 Parity, Gemini Chat, GBT, Claude. 583 00:39:18,746 --> 00:39:25,946 That automatically default to provide cited links that can be followed to verify 584 00:39:25,976 --> 00:39:27,866 and evaluate sources, in other words. 585 00:39:28,076 --> 00:39:32,726 So hopefully everyone has at least gone into chat t and done one prompt. 586 00:39:32,726 --> 00:39:36,176 If you haven't, please go do it right now and you're gonna see that it's gonna 587 00:39:36,176 --> 00:39:38,066 give you a response with the default. 588 00:39:38,066 --> 00:39:41,276 And don't change any default models, don't change any default settings. 589 00:39:41,576 --> 00:39:44,966 But what it doesn't do, it gives you the response, but it doesn't tell 590 00:39:44,966 --> 00:39:47,036 you where that response came from. 591 00:39:47,036 --> 00:39:52,676 It doesn't give you cited websites that you can use to verify that 592 00:39:52,676 --> 00:39:55,556 that information is correct and it's not hallucinating. 593 00:39:55,556 --> 00:39:59,786 That's what Perplexity and Gemini do as of default. 594 00:40:00,176 --> 00:40:02,546 And Gemini, this is a newer ish thing. 595 00:40:02,546 --> 00:40:03,596 It hasn't always done this. 596 00:40:03,596 --> 00:40:07,166 So if you haven't been in Gemini for a while, it does do this by default now. 597 00:40:07,496 --> 00:40:09,286 So when you put a prompt into Gemini,. 598 00:40:09,336 --> 00:40:12,876 When you get that prompt back, you're gonna see, and sometimes 599 00:40:12,876 --> 00:40:13,836 it'll look like a little link. 600 00:40:13,836 --> 00:40:17,676 Sometimes it'll look like a little number, like a citation and a research paper. 601 00:40:17,946 --> 00:40:21,786 When you click on that, it'll give you a website that you can click 602 00:40:21,786 --> 00:40:26,436 and open to say, Hey, look, this is the website that is showing the 603 00:40:26,436 --> 00:40:28,326 information that I'm also giving you. 604 00:40:29,076 --> 00:40:35,046 Part of that is going and looking and saying, okay, this is a credible website, 605 00:40:35,046 --> 00:40:37,566 or This is not a credible website. 606 00:40:37,926 --> 00:40:40,746 Also, both perplexity and Gemini. 607 00:40:40,836 --> 00:40:44,946 At the end usually, or off to the side, there'll be a sources button 608 00:40:44,946 --> 00:40:48,006 and you can click that and it'll show all of the sources that it 609 00:40:48,006 --> 00:40:52,356 cited throughout the entire response that it gave you, essentially. 610 00:40:53,886 --> 00:40:57,516 So what I do is I put it up on the big screen, I model it. 611 00:40:57,516 --> 00:41:00,636 We have lots of discussions and conversations. 612 00:41:01,296 --> 00:41:01,956 Around. 613 00:41:02,016 --> 00:41:05,586 I talk about the technology, all of the things we compare outputs 614 00:41:05,586 --> 00:41:06,966 across the different models. 615 00:41:07,176 --> 00:41:13,156 I show them and I put up chat, GPT and Claude, and I show them, at least as 616 00:41:13,156 --> 00:41:17,836 of right now, that those tools do not automatically provide sided information. 617 00:41:18,196 --> 00:41:25,456 There is a web setting that you can turn on in both of those tools that if you turn 618 00:41:25,456 --> 00:41:31,096 it on, then sometimes, not always, always, most of the time it'll give you setted 619 00:41:31,096 --> 00:41:35,086 links, but you have to know to turn that on and you have to know where to find it, 620 00:41:35,136 --> 00:41:38,665 so we talk through all of that and yeah, and then just show them, 621 00:41:38,665 --> 00:41:41,545 Hey, look, see how perplexity and Gemini give you these citations. 622 00:41:41,815 --> 00:41:45,356 I give them, I kind of have four different strategies that we talk through of how 623 00:41:45,356 --> 00:41:48,175 to validate information, et cetera. 624 00:41:48,205 --> 00:41:51,086 One of them is ask different models. 625 00:41:52,075 --> 00:41:54,025 Go give the same prompt to different models. 626 00:41:54,025 --> 00:42:00,325 These are essentially AI literacy lessons that students must know and they 627 00:42:00,325 --> 00:42:06,115 must know prior to getting into an AI model that does not have guardrails. 628 00:42:06,445 --> 00:42:12,325 So Notebook LM is a great scaffold to learning how to research responsibly 629 00:42:12,325 --> 00:42:14,005 using large language models. 630 00:42:14,335 --> 00:42:15,235 Start there. 631 00:42:15,805 --> 00:42:19,015 Then we get into Gemini and perplexity. 632 00:42:19,795 --> 00:42:21,655 Lots and lots and lots of scaffolding. 633 00:42:21,925 --> 00:42:22,525 AKA. 634 00:42:22,675 --> 00:42:25,735 We're not just throwing them in there, even high school kids. 635 00:42:26,735 --> 00:42:27,605 Alright, next. 636 00:42:27,605 --> 00:42:32,345 This is a don't for both teachers and parents. 637 00:42:33,345 --> 00:42:37,215 I don't want you using a general purpose, large language model 638 00:42:37,785 --> 00:42:43,200 for research tasks without that source verification built into it. 639 00:42:44,850 --> 00:42:48,300 So if you're able to know how to go into Chatt PT, and this is 640 00:42:48,300 --> 00:42:50,940 specifically for research tasks. 641 00:42:52,260 --> 00:42:55,920 Specifically, you're trying to find information, verify that 642 00:42:55,920 --> 00:42:58,230 information, compile that information. 643 00:42:58,650 --> 00:43:02,130 That's very different than just going and asking it, what should 644 00:43:02,130 --> 00:43:03,000 I have for dinner tonight? 645 00:43:03,180 --> 00:43:07,740 That's not the greatest example, but we'll go with it, so specifically for research 646 00:43:07,740 --> 00:43:11,700 tasks, you need that source verification. 647 00:43:11,730 --> 00:43:15,870 Ideally, all large language models in the future are going to have this kind of 648 00:43:15,870 --> 00:43:17,970 like source verification built into it. 649 00:43:18,390 --> 00:43:20,460 I hope we're not there yet. 650 00:43:20,465 --> 00:43:22,350 There's gonna be times where you're gonna use it. 651 00:43:22,350 --> 00:43:26,100 There's gonna be times when you don't, but research is when you do want to use it. 652 00:43:27,570 --> 00:43:36,330 Now, a do for both teachers and parents do teach your kids to prompt AI in 653 00:43:36,330 --> 00:43:37,800 ways that actually help them learn. 654 00:43:38,800 --> 00:43:42,671 These are five different strategies, and I've got these in a blog post, 655 00:43:42,671 --> 00:43:47,770 which I'll put in the show notes that I teach to personalize a task. 656 00:43:47,950 --> 00:43:53,200 So one, if you need something simplified, something like explain this like a third 657 00:43:53,200 --> 00:43:55,270 grader, explain this like a kindergartner. 658 00:43:55,270 --> 00:43:58,660 I've even gotten down to explain it like a preschooler before. 659 00:43:58,660 --> 00:44:03,970 And it'll take really complex tasks and try to explain them as simply as possible. 660 00:44:03,970 --> 00:44:05,800 And then you can build back up from there. 661 00:44:06,100 --> 00:44:09,580 But if you're not understanding the complex tasks to start with, 662 00:44:09,881 --> 00:44:15,020 it's really hard to do anything from there until you understand it. 663 00:44:15,020 --> 00:44:15,920 And then you can go back up. 664 00:44:17,090 --> 00:44:19,340 You can have it personalized reading level. 665 00:44:19,550 --> 00:44:24,140 So change it to a 700 Lexile score, change it to a 500 Lexile score. 666 00:44:25,550 --> 00:44:29,180 You can use it to translate if you're a multilingual learner or your 667 00:44:29,180 --> 00:44:32,930 kids are multilingual learners, you can have it translate to Spanish, 668 00:44:32,930 --> 00:44:34,880 you can have it give outputs in. 669 00:44:35,360 --> 00:44:36,500 Two languages. 670 00:44:36,500 --> 00:44:38,120 So English and Spanish. 671 00:44:39,120 --> 00:44:43,680 If your kid or student is not interested at all in whatever it is that they're 672 00:44:43,680 --> 00:44:47,160 learning, and maybe you're a parent and it's 7:00 PM and you're trying to 673 00:44:47,160 --> 00:44:51,120 get through this so everybody can go to bed at night and there's no motivation 674 00:44:51,120 --> 00:44:55,380 there, you can have the AI related to something that they aren't interested in. 675 00:44:55,440 --> 00:44:59,190 So how does whatever concept you're learning relate to soccer 676 00:44:59,190 --> 00:45:01,260 or Fortnite or cats or curling? 677 00:45:01,560 --> 00:45:02,880 I've been doing that a lot lately. 678 00:45:03,150 --> 00:45:04,680 I'm like, tell me how this relates to curling. 679 00:45:05,460 --> 00:45:09,240 Got into it at the Olympics, whatever it is that your kid is interested into. 680 00:45:09,240 --> 00:45:12,510 Or you, maybe this, this applies to you as well. 681 00:45:13,170 --> 00:45:16,410 Now this is one that I hadn't been doing, but I am now doing 682 00:45:16,410 --> 00:45:19,950 more and more for task initiation. 683 00:45:20,460 --> 00:45:22,230 So have the AI take. 684 00:45:22,920 --> 00:45:26,280 Whatever the learning task they're doing or whatever the homework is for that 685 00:45:26,280 --> 00:45:31,440 night and say, break this task into steps to help me complete it step by step. 686 00:45:32,430 --> 00:45:36,120 You can even then have it break it into more steps, or maybe it breaks it too far. 687 00:45:36,270 --> 00:45:37,890 You can say, oh, maybe not that far. 688 00:45:37,890 --> 00:45:38,730 Bring it back. 689 00:45:39,030 --> 00:45:42,270 Wherever your kid is at in terms of their executive functioning skills, 690 00:45:42,450 --> 00:45:46,830 have the AI help you break it down step by step by step by step. 691 00:45:47,830 --> 00:45:53,650 I'm digging into this more and more of how AI can support executive functioning. 692 00:45:54,040 --> 00:45:58,660 I hear from almost every educator I talk to in every school that I'm in, 693 00:45:58,930 --> 00:46:03,370 that one of their biggest challenges with students is executive function or 694 00:46:03,550 --> 00:46:06,070 the lack of executive function usually. 695 00:46:06,370 --> 00:46:12,520 So when AI tools break tasks into steps, guide students through the process, 696 00:46:12,820 --> 00:46:16,300 they're building those executive functioning skills, they're helping kids 697 00:46:16,300 --> 00:46:22,900 build those skills, task initiation, cognitive flexibility, metacognition. 698 00:46:24,700 --> 00:46:27,670 These are skills that many students, especially neurodiverse 699 00:46:27,670 --> 00:46:33,280 learners, generally struggle with, and a well designed. 700 00:46:33,610 --> 00:46:35,200 Remember, tool Design Matters. 701 00:46:35,200 --> 00:46:40,360 A well-designed AI tool can help kids learn how to approach problems and 702 00:46:40,360 --> 00:46:43,630 accomplish big tasks with lots of steps. 703 00:46:44,630 --> 00:46:49,790 This is really hugely helpful for a teacher that has 30 kids in a classroom. 704 00:46:50,030 --> 00:46:52,310 You can't sit down with every kid and walk them through each 705 00:46:52,310 --> 00:46:55,190 step every second of every day. 706 00:46:56,210 --> 00:46:59,760 And some need more steps broken down than others, but this is also 707 00:46:59,760 --> 00:47:04,350 a huge help for parents trying to get homework done at night. 708 00:47:04,530 --> 00:47:05,550 Maybe you have multiple kids. 709 00:47:05,550 --> 00:47:07,710 You're trying to get all their homework done at night, and you 710 00:47:07,710 --> 00:47:10,560 can't sit one-on-one with each kid. 711 00:47:12,330 --> 00:47:16,920 So executive functioning, helping with task initiation. 712 00:47:17,670 --> 00:47:18,330 Huge. 713 00:47:18,660 --> 00:47:20,730 But notice, so I gave you those. 714 00:47:20,730 --> 00:47:22,140 What were those five examples? 715 00:47:22,410 --> 00:47:25,560 Simplification, changing the reading level, translation, personal 716 00:47:25,560 --> 00:47:28,110 connection and task initiation. 717 00:47:28,410 --> 00:47:32,670 Every one of those prompts puts the student in the driver's seat, 718 00:47:32,670 --> 00:47:34,140 the learner in the driver's seat. 719 00:47:34,470 --> 00:47:39,630 They're not asking for the answer they're asking to be met where 720 00:47:39,630 --> 00:47:46,060 they are in their zone of proximal development, their ZPD, ? That is using 721 00:47:46,060 --> 00:47:47,530 this technology in a powerful way. 722 00:47:48,580 --> 00:47:51,850 That is AI literacy and action. 723 00:47:51,850 --> 00:47:56,260 And it's a skill that they're gonna take through the rest of their life, 724 00:47:56,260 --> 00:48:01,420 far beyond the homework assignment that they're doing at that moment. 725 00:48:02,420 --> 00:48:06,500 On that note, please don't assume that students know how 726 00:48:06,500 --> 00:48:08,735 to prompt effectively most. 727 00:48:09,515 --> 00:48:10,685 Absolutely do not. 728 00:48:10,775 --> 00:48:16,125 In my experience, this goes back to this idea of digital natives, which 729 00:48:16,125 --> 00:48:18,015 I really wish we would get rid of. 730 00:48:18,045 --> 00:48:23,055 I always say that yes, we may have quote digital natives in terms of 731 00:48:23,055 --> 00:48:24,645 they're not scared of the technology. 732 00:48:24,855 --> 00:48:28,545 They pick it up, they start pressing buttons versus adults, 733 00:48:28,545 --> 00:48:32,455 typically older adults are quite a bit more hesitant, right? 734 00:48:32,455 --> 00:48:34,975 And they're not just button clickers. 735 00:48:35,275 --> 00:48:38,245 Kids typically are button clickers 'cause they're used to it. 736 00:48:38,245 --> 00:48:39,295 It's what they grew up with. 737 00:48:40,295 --> 00:48:45,095 But that does not mean that they know how to use these digital 738 00:48:45,185 --> 00:48:52,655 tools and devices for learning and ethical, responsible, productive use. 739 00:48:53,225 --> 00:48:59,735 They know how to use it to play Fortnite or Angry Birds or Candy Crush. 740 00:48:59,795 --> 00:49:00,125 I don't know. 741 00:49:00,125 --> 00:49:01,685 I don't play digital games. 742 00:49:01,685 --> 00:49:03,035 I don't even know what's in right now. 743 00:49:03,605 --> 00:49:07,025 I don't think Candy Crush and Angry Birds are in anymore, 744 00:49:08,025 --> 00:49:09,465 regardless, whatever it is. 745 00:49:09,495 --> 00:49:10,575 Roblox there. 746 00:49:10,575 --> 00:49:11,865 That was a much better example. 747 00:49:13,095 --> 00:49:17,625 They know how to use this technology to play Roblox in Fortnite. 748 00:49:17,805 --> 00:49:22,785 That is very, very different than leveraging technology as a 749 00:49:22,785 --> 00:49:27,945 learning tool to help you organize, to help you accomplish tasks. 750 00:49:28,245 --> 00:49:29,145 Very different. 751 00:49:29,505 --> 00:49:35,595 And the latter has to be taught, has to be modeled specifically. 752 00:49:35,985 --> 00:49:40,545 They pick up on the former pretty quickly 'cause they are quote, digital natives, 753 00:49:40,815 --> 00:49:44,055 but they don't pick up on the ladder. 754 00:49:44,505 --> 00:49:48,195 The same is now going for using AI tools. 755 00:49:50,025 --> 00:49:52,635 They don't know how to use them effectively. 756 00:49:52,845 --> 00:49:57,825 The number of kits that I've sat and watched just take the very first. 757 00:49:58,155 --> 00:50:01,305 Response that comes from a large language model and copy and paste it, 758 00:50:02,085 --> 00:50:04,035 we all know they're doing it, but because they don't know any other, 759 00:50:04,035 --> 00:50:05,595 they've been taught to do it differently. 760 00:50:06,595 --> 00:50:11,275 They also don't know the right tools to use or how to prompt correctly to use 761 00:50:11,275 --> 00:50:16,315 it in ways that support your learning and your thinking and don't outsource 762 00:50:16,555 --> 00:50:17,785 your learning and your thinking. 763 00:50:17,965 --> 00:50:23,635 So proper prompting of AI takes teachers, building that into a lesson. 764 00:50:23,695 --> 00:50:28,285 Parents, you have to model it at home and both teachers and parents 765 00:50:28,315 --> 00:50:32,935 having the conversations no matter who or how it has to be taught. 766 00:50:33,935 --> 00:50:36,880 Alright, that is the end. 767 00:50:37,805 --> 00:50:38,975 Our dos and don'ts. 768 00:50:38,975 --> 00:50:40,745 Let me give you a quick little recap here 769 00:50:41,745 --> 00:50:44,385 because I blabbered, I know I gave you a lot of information 770 00:50:44,385 --> 00:50:45,435 and I blabbed a lot for that. 771 00:50:45,435 --> 00:50:48,585 There is, did you notice there's a lot to talk about and cover here 772 00:50:49,005 --> 00:50:52,815 and the answer of is AI safe for students is not black and white? 773 00:50:53,265 --> 00:50:57,225 I hope with all of my babbling that that is the conclusion that you have 774 00:50:57,225 --> 00:51:02,385 hopefully come to after the last episode, part one and this episode. 775 00:51:03,385 --> 00:51:04,405 Okay, so here we go. 776 00:51:05,185 --> 00:51:07,285 Here is an overview of my dos and don'ts. 777 00:51:07,285 --> 00:51:09,025 Don't for teachers and parents. 778 00:51:09,055 --> 00:51:15,115 I do not want you to put your students into AI tools without any AI literacy 779 00:51:15,145 --> 00:51:17,185 lessons and conversations First. 780 00:51:18,185 --> 00:51:20,615 Next do for teachers. 781 00:51:21,065 --> 00:51:25,805 I want you to use tools that have guardrails and teacher dashboards to give 782 00:51:25,805 --> 00:51:28,415 you full visibility into student progress. 783 00:51:28,865 --> 00:51:34,085 There are a few caveats and special use cases and circumstances with 784 00:51:34,085 --> 00:51:39,755 older kids where you may put them into like a Gemini gym, 785 00:51:40,755 --> 00:51:47,175 but that is not the norm and not where I want you to start either 786 00:51:48,175 --> 00:51:53,155 parents, you can use these guardrail tools with your students. 787 00:51:53,155 --> 00:51:56,485 Actually, I did a webinar on this for varsity tutors and I showed them how to 788 00:51:56,485 --> 00:51:59,275 use school AI spaces, notebook, limb. 789 00:52:00,205 --> 00:52:01,885 And one or two others. 790 00:52:01,885 --> 00:52:03,505 But those were the two big ones that I showed them. 791 00:52:03,595 --> 00:52:06,325 So you can use these tools too as long as they're free meal. 792 00:52:06,685 --> 00:52:10,795 So I mentioned school AI spaces and brisk boosts were two good ones to start with. 793 00:52:12,385 --> 00:52:16,345 You also have a little bit more flexibility than teachers 'cause 794 00:52:16,345 --> 00:52:21,195 you're working with much smaller number of kids, so you can use some gyms, 795 00:52:21,195 --> 00:52:24,045 some custom GPTs projects, spaces, 796 00:52:24,825 --> 00:52:32,445 you can quote unquote prime those gyms and custom GPTs and projects and spaces with 797 00:52:32,445 --> 00:52:36,075 instructions that say, guide not give. 798 00:52:36,105 --> 00:52:37,425 This is the age of the student. 799 00:52:37,425 --> 00:52:38,535 This is what they're learning. 800 00:52:39,105 --> 00:52:44,705 You can also, if you need a baby, step into creating a gym or a custom GPT. 801 00:52:45,035 --> 00:52:49,565 You can also just do this within a chat in a large language model and give it a 802 00:52:49,565 --> 00:52:52,355 prompt and say, this is what we're doing. 803 00:52:52,355 --> 00:52:53,435 This is the age of the student. 804 00:52:53,435 --> 00:52:56,915 'cause you wanna make sure the outputs , are age and developmentally appropriate. 805 00:52:57,215 --> 00:52:58,295 This is what I want you to do. 806 00:52:58,295 --> 00:53:00,005 This is what I don't want you to do. 807 00:53:01,005 --> 00:53:05,565 If you're gonna do that, test different tools, test different AI models 808 00:53:05,565 --> 00:53:08,955 and find the one that works best. 809 00:53:09,345 --> 00:53:14,325 Also, I'm not sure I specifically said this, but the one that works best 810 00:53:14,325 --> 00:53:17,655 isn't always going to be the same. 811 00:53:17,955 --> 00:53:19,905 It depends on the task that you're doing. 812 00:53:20,085 --> 00:53:21,435 I gave the example of math. 813 00:53:21,705 --> 00:53:25,845 If I'm helping with math, especially more advanced math, I'm gonna 814 00:53:25,845 --> 00:53:29,445 use a very different model than if I'm helping with writing. 815 00:53:30,445 --> 00:53:32,305 So a don't for teachers and parents. 816 00:53:32,305 --> 00:53:35,545 Don't just drop students into a general purpose, large language 817 00:53:35,545 --> 00:53:38,545 model or chat bot with no structure. 818 00:53:39,655 --> 00:53:42,925 Maybe, maybe, maybe like upper high school kids. 819 00:53:43,525 --> 00:53:48,685 That would depend on a lot of stuff, a lot of different caveats there. 820 00:53:49,165 --> 00:53:54,175 But ideally, at minimum, go into the chat and give it some background and kind of 821 00:53:54,175 --> 00:53:59,635 prime that chat with information about what you're wanting the student to do. 822 00:54:00,115 --> 00:54:02,965 Also, teacher, I don't know if I specifically said this, but 823 00:54:02,965 --> 00:54:06,025 teach your kids how to do this, especially the older kids, how to 824 00:54:06,025 --> 00:54:08,875 prime their own chats to tell it. 825 00:54:09,235 --> 00:54:10,435 Do this, don't do this. 826 00:54:10,435 --> 00:54:10,915 Guide me. 827 00:54:10,915 --> 00:54:18,700 Don't give the answer a big do notebook, lm. For both teachers and parents, 828 00:54:19,120 --> 00:54:23,950 depending upon for both, really, both what age student you're working with. 829 00:54:24,220 --> 00:54:28,570 This might be more of a U tool where you go into it and maybe generate an audio 830 00:54:28,570 --> 00:54:33,700 summary or an infographic, or it might be a tool that you can put students in 831 00:54:33,880 --> 00:54:38,710 and have them create infographics and videos and flashcards and audio summaries. 832 00:54:39,710 --> 00:54:44,600 Also, notebook element is really great as a first scaffold into 833 00:54:44,600 --> 00:54:49,610 teaching students how to research correctly using large language models. 834 00:54:50,610 --> 00:54:57,270 That leads to my next do, which is for parents and teachers about research. 835 00:54:58,270 --> 00:55:03,340 It takes a lot of scaffolding, a lot of AI literacy lessons and conversations. 836 00:55:03,955 --> 00:55:08,395 Putting them in notebook, lm, putting it up on your projector or your screen 837 00:55:08,395 --> 00:55:12,115 in the front of the room, showing them perplexity, showing them Gemini, showing 838 00:55:12,115 --> 00:55:18,205 them notebook L, and teaching them how to verify information and evaluate 839 00:55:18,205 --> 00:55:20,935 information for research purposes. 840 00:55:20,935 --> 00:55:21,625 Especially. 841 00:55:21,870 --> 00:55:28,230 I don't, for teachers and parents, don't just put them in any general purpose, 842 00:55:28,230 --> 00:55:35,070 large language model for research tasks without source verification built into it. 843 00:55:36,070 --> 00:55:38,200 A do for teachers and parents. 844 00:55:38,530 --> 00:55:45,460 Do teach your kids how to prompt AI period, but also specifically how to 845 00:55:45,460 --> 00:55:50,830 prompt AI in ways that actually help them learn to simplify, to translate, 846 00:55:51,220 --> 00:55:53,560 to help them with task initiation, 847 00:55:54,190 --> 00:55:54,940 that's it. 848 00:55:55,390 --> 00:55:55,870 That's it. 849 00:55:56,200 --> 00:55:56,920 That was a lot. 850 00:55:57,190 --> 00:56:00,940 We covered a lot of ground today and I could talk about so much more, but 851 00:56:00,940 --> 00:56:02,770 here's what I want to leave you with. 852 00:56:03,370 --> 00:56:05,650 You don't have to be an AI expert to do this. 853 00:56:05,980 --> 00:56:09,970 You don't have to be a tech expert. 854 00:56:10,070 --> 00:56:11,630 You don't have to have the perfect lesson. 855 00:56:13,190 --> 00:56:15,020 You don't have to feel fully comfortable. 856 00:56:15,020 --> 00:56:18,200 You're likely never gonna feel fully comfortable because this 857 00:56:18,200 --> 00:56:20,540 technology is moving so fast. 858 00:56:22,190 --> 00:56:28,340 What matters most, more than the tool or anything else is the conversation. 859 00:56:28,760 --> 00:56:30,410 Have the discussions. 860 00:56:30,680 --> 00:56:31,940 Talk to your students. 861 00:56:32,000 --> 00:56:33,140 Talk to your kids. 862 00:56:34,430 --> 00:56:37,670 Ask them what they're using, ask them if they think it's always right, 863 00:56:37,750 --> 00:56:41,950 those conversations are AI literacy, and that is more important than 864 00:56:42,010 --> 00:56:47,290 anything right now to help avoid the bad things about this technology. 865 00:56:47,290 --> 00:56:51,500 Like companionship, conversations are available to every 866 00:56:51,500 --> 00:56:53,180 teacher, to every parent. 867 00:56:53,330 --> 00:56:55,730 No login required, no tools required. 868 00:56:56,660 --> 00:57:01,910 Even as the tools keep changing, those conversations are the foundation to making 869 00:57:01,910 --> 00:57:07,010 sure that we're using this technology in healthy, responsible, ethical ways. 870 00:57:07,370 --> 00:57:12,170 The research will keep coming, but building curious, skeptical, 871 00:57:12,530 --> 00:57:17,330 thoughtful humans, that is what we need to be focused on right now. 872 00:57:20,840 --> 00:57:22,910 Speaker 3: Thanks for joining Make EdTech 100. 873 00:57:23,630 --> 00:57:28,310 I know educator time is valuable and I'm honored you choose to spend yours with me. 874 00:57:29,360 --> 00:57:34,010 For more EdTech strategies you can use tomorrow and ways to bring me to your 875 00:57:34,010 --> 00:57:36,510 school or event, head to LindyHoc.com. 876 00:57:37,700 --> 00:57:41,600 If this episode resonated, hit subscribe so you don't miss the next one. 877 00:57:42,380 --> 00:57:42,720 I'm LindyHoc. 878 00:57:43,400 --> 00:57:45,710 Go forth and make EdTech 100.