1 00:00:00,000 --> 00:00:05,120 Shouldn't we start talking? How should we educate kids going forward? 2 00:00:05,120 --> 00:00:08,640 And when this education could be a solid point? 3 00:00:08,640 --> 00:00:37,200 And ladies and gentlemen, boys and girls, children, all ages, dogs, cats, robots, and everybody in between, welcome to HTTTA, how to talk to AI. I am your host, Wes the Synth Mind, Synth Mind Wes. 4 00:00:37,200 --> 00:00:50,640 And as always, in a globe governed by gigabytes and genetic algorithms, one guide galvanizes to decode the grander, generalize the complexities and guide us through the uncharted grids of artificial intelligence. 5 00:00:50,640 --> 00:00:55,760 Who is this, you may ask? It's GoToGo herself. Gee, how are you this week? 6 00:00:55,760 --> 00:01:03,200 Hi, I'm great. I recomposed myself. I'm great. This is absolutely normal introduction. I'm totally used to that. 7 00:01:03,200 --> 00:01:10,400 There you go. I am merely a conduit for generative AI to speak through with various adjectives that start with the letter G. 8 00:01:10,400 --> 00:01:23,840 Episode two of video, episode 12, we're working out some kinks, but we appreciate the response from all our audience thus far. Here we are. Another week in AI. There's always exciting stuff happening in AI. 9 00:01:23,840 --> 00:01:28,400 Some exciting launches in AI as well. Mid-Journey 5.2. 10 00:01:28,400 --> 00:01:32,000 Sure. I mean, have you gotten a chance to noodle around with 5.2? 11 00:01:32,000 --> 00:01:40,240 Yeah, everyone is super excited about zoom function. And I really want to still create this infinite zoom of image. 12 00:01:40,240 --> 00:01:45,680 Which movie was it? Limitless? You know, this episode where it just keeps going. 13 00:01:45,680 --> 00:01:53,360 So that's something I really want to recreate now with Mid-Journey. We had already tools for that before, but I think just fascinating. 14 00:01:53,360 --> 00:02:00,640 But another feature, which I tested just from the curiosity standpoint, was shortened prompts. 15 00:02:00,640 --> 00:02:14,720 You say slash, shorten, input your prompt and Mid-Journey is evaluating which parts of your prompt do nothing or you should eliminate and makes you, I think, five suggestions. 16 00:02:14,720 --> 00:02:26,720 That was very interesting. So the next thing that I want to test now, I haven't done yet to take Prompt Perfect, optimize prompt and just see how that weights together with Mid-Journey's judgment. 17 00:02:26,720 --> 00:02:31,520 Because usually that's exactly what Prompt Perfect was doing, optimizing for Mid-Journey. 18 00:02:31,520 --> 00:02:36,320 I still use Prompt Perfect as opposed to the describe function in Mid-Journey. 19 00:02:36,320 --> 00:02:46,320 For those that don't know, within Mid-Journey, I think what, five, five zero or five one and beyond, you can upload a image and say Mid-Journey describe it. 20 00:02:46,320 --> 00:02:55,520 And it will use kind of like the reverse of a diffusion model to describe through computer vision and a bunch of other things what is in that picture and create a prompt. 21 00:02:55,520 --> 00:03:07,120 However, I happen to always find Mid-Journey's version of that a little stale and our good friends over at Prompt Perfect's version of the same tool, which, mind you, they had debuted, I think, a month before, at least. 22 00:03:07,120 --> 00:03:11,920 If not longer, I always think there's a better job with least with more colorful prompt language. 23 00:03:11,920 --> 00:03:19,120 But now we have some measure to kind of go back and see if all that extra colorful flowery language even matters. 24 00:03:19,120 --> 00:03:25,920 And from Perfect, you have more of the controls, like shorten the prompt, increase quality and stuff like that. 25 00:03:25,920 --> 00:03:32,720 So it's not just that you upload image and it describes it, but you can adjust it to what you are actually looking for. 26 00:03:32,720 --> 00:03:38,320 Listeners, hopefully know, check out the show notes for a little discount code to our friends over at Prompt Perfect. 27 00:03:38,320 --> 00:03:44,720 Give it a shot. I have to imagine that the zoom feature is also pretty compelling to creators. 28 00:03:44,720 --> 00:03:47,120 You said you got a chance to play with that a little bit, right? 29 00:03:47,120 --> 00:03:55,520 Yeah, and I think it's interesting that this kind of mid-journey's way to respond to this generative fill in Photoshop, 30 00:03:55,520 --> 00:04:02,120 where you have to expand and prompt it again to fill and this is just like does it automatically. 31 00:04:02,120 --> 00:04:05,420 And the results are really, really compelling and really great. 32 00:04:05,420 --> 00:04:08,620 It's becoming less shocking, this type of features. 33 00:04:08,620 --> 00:04:13,920 I don't know how it is on your side, but when I saw it tested, I was like, yeah, cool. 34 00:04:13,920 --> 00:04:17,420 That's how I imagine it should be functioning. 35 00:04:17,420 --> 00:04:20,720 Am I just getting like completely numb? 36 00:04:20,720 --> 00:04:28,320 We are really just getting a little numb to it because it's just like, you know, take any for those that can't really conceptualize right now what we're talking about. 37 00:04:28,320 --> 00:04:31,720 We have a couple pictures of how the zoom out function works. 38 00:04:31,720 --> 00:04:37,220 Basically imagine an image and then if you were to step back at certain degrees, 39 00:04:37,220 --> 00:04:44,120 you know, 1.75 times, two times away from the main image subject, what would that look like? 40 00:04:44,120 --> 00:04:49,520 What things would be happening around that image if you were to zoom out from the initial one? 41 00:04:49,520 --> 00:05:00,120 So it can get very creative, adding people to a scene or a portrait image or a landscape, entirely different features as you zoom out. 42 00:05:00,120 --> 00:05:09,320 It's a very interesting use for generative AI because it not only can be used for completely generated image creations, 43 00:05:09,320 --> 00:05:15,520 but re-imaginings of images that were taken, existing photos or pieces of art. 44 00:05:15,520 --> 00:05:18,520 What would that look like if, you know, they were to zoom out a little more? 45 00:05:18,520 --> 00:05:24,620 Now you gave me an idea to extend my own images, but also extend my own paintings. 46 00:05:24,620 --> 00:05:26,320 Ah, there you go. 47 00:05:26,320 --> 00:05:28,120 That's an interesting, no? 48 00:05:28,120 --> 00:05:28,520 Yeah. 49 00:05:28,520 --> 00:05:35,320 So we talked a lot about the sentiment about prompts and generating prompts and prompt engineering as a career. 50 00:05:35,320 --> 00:05:43,520 Multiple comments I saw is that you can just ask Charge GPT to create prompts for you or optimize prompts for you. 51 00:05:43,520 --> 00:05:46,820 And I was like, yeah, of course it's logical that it does, 52 00:05:46,820 --> 00:05:53,920 but we also know with hallucinations that Charge GPT will go extra miles to sound extremely convincing. 53 00:05:53,920 --> 00:05:56,720 So I went and I was like, oh, optimize prompt for me journey. 54 00:05:56,720 --> 00:05:57,420 Okay, fine. 55 00:05:57,420 --> 00:06:04,720 Like I don't have any baseline to let's say evaluate that besides prompt perfect optimization or describe in image. 56 00:06:04,720 --> 00:06:13,420 But then I was like, okay, let me put a research hat and be like, let's ask to optimize prompt and create prompt for 57 00:06:13,420 --> 00:06:18,720 Wacuna text to image model, which is completely made up. 58 00:06:18,720 --> 00:06:25,020 And yeah, it went and was like, yeah, the best optimization tactics for Wacuna image models are this and that. 59 00:06:25,020 --> 00:06:33,820 And it created a prompt for this model and even told me that these tactics work very well for Wacuna image model. 60 00:06:33,820 --> 00:06:45,820 So just to put it out there that yeah, Charge GPT will go ahead and create amazing prompts and optimize prompts for whatever model you come up with. 61 00:06:45,820 --> 00:06:52,620 So it's just a question, you know, what is good enough and what is really optimized and what you can really do. 62 00:06:52,620 --> 00:06:59,120 So I would always, you know, lean to the companies who specialize on this specific issue or a problem solving. 63 00:06:59,120 --> 00:07:00,220 Exactly. 64 00:07:00,220 --> 00:07:02,720 It's not going to truly performing optimization. 65 00:07:02,720 --> 00:07:06,020 It's just going to reconfigure the prompt. 66 00:07:06,020 --> 00:07:13,720 Whereas like some of the tools that you're given at prompt perfect where it can add different measures of quality, shorten it, elongate it. 67 00:07:13,720 --> 00:07:21,120 That's actually done with an algorithm or a function that is improved over time and calculated and measured. 68 00:07:21,120 --> 00:07:25,420 That's the way you actually truly want to optimize a prompt like that. 69 00:07:25,420 --> 00:07:39,120 Well, speaking of optimization, we came across a little happenstance this week that made me smile a little bit with regards to AI in the classroom, specifically computer science classrooms. 70 00:07:39,120 --> 00:07:47,720 Not sure folks are aware, but Harvard has a introduction to computer science course called CS50 and the social network movie. 71 00:07:47,720 --> 00:07:56,620 It talks about this a little bit, you know, with Zuckerberg and in this course, it's a very famous and widely lauded computer science course. 72 00:07:56,620 --> 00:08:01,120 Five thousand students at Harvard sign up for it every year. 73 00:08:01,120 --> 00:08:07,120 And one great thing about it before I get into the article, the entire course is free on YouTube. 74 00:08:07,120 --> 00:08:10,120 We have a YouTube channel, one million. 75 00:08:10,120 --> 00:08:13,120 One point five million just for this course. 76 00:08:13,120 --> 00:08:15,220 And the profile picture is a cat. 77 00:08:15,220 --> 00:08:23,020 The primary instructor, Professor Mandlin, easily the most engaging instructor I've ever, ever watched, ever witnessed. 78 00:08:23,020 --> 00:08:35,520 Even if you have no interest in coding computers, anything at all, it's undeniable that they pretty much are kind of foundational to our entire human experience now, 79 00:08:35,520 --> 00:08:40,220 be it our phones, be it the technology in our cars, TVs, everything around us. 80 00:08:40,220 --> 00:08:47,720 So I would encourage everyone to just watch at least the first video in the CS50 series. 81 00:08:47,720 --> 00:08:52,020 I think it's a series of about 20 ish lectures that are on there. 82 00:08:52,020 --> 00:08:54,820 Do yourself a favor. Watch the first one. 83 00:08:54,820 --> 00:09:02,120 It was not until I found this course in grad school or I was in comp methods one just in over my head. 84 00:09:02,120 --> 00:09:03,820 I'm just like, what is going on? 85 00:09:03,820 --> 00:09:05,520 What is this like Python? 86 00:09:05,520 --> 00:09:09,120 You're telling me about, you know, functions and lists and dictionaries. 87 00:09:09,120 --> 00:09:11,420 And I'm just like, it's not gelling. 88 00:09:11,420 --> 00:09:16,320 And I just happen to find this course and Professor Mandlin is just so engaging. 89 00:09:16,320 --> 00:09:21,320 And just throughout the course of this by like the third or fourth video, it just like clicked for me. 90 00:09:21,320 --> 00:09:26,420 It kind of gelled. I'm just like, oh, that's how you mean object oriented programming. 91 00:09:26,420 --> 00:09:30,120 That's how binary works. It's got some great examples. 92 00:09:30,120 --> 00:09:33,720 I think really anyone would find it terrific. 93 00:09:33,720 --> 00:09:36,320 So I can't give a ringing enough endorsement for it. 94 00:09:36,320 --> 00:09:39,820 But the reason I'm even telling this story is within CS50. 95 00:09:39,820 --> 00:09:44,320 Now the students are going to have a chat bot teacher. 96 00:09:44,320 --> 00:09:48,420 The AI teacher is going to teach part of the class beginning in September. 97 00:09:48,420 --> 00:09:54,320 Is it just that AI avatar or is chat bot and personalized learning path? 98 00:09:54,320 --> 00:09:59,220 And another thing is this for free with the AI teacher. 99 00:09:59,220 --> 00:10:05,320 This whole course is online and also taught through their EDX platform listed on there all for free. 100 00:10:05,320 --> 00:10:13,820 I have to imagine that the parents of these undergraduate students taking CS50 don't feel too great about an AI chat bot teaching their students 101 00:10:13,820 --> 00:10:18,320 considering they're paying the tuition fees that I'm sure are quite substantial in Harvard these days. 102 00:10:18,320 --> 00:10:19,020 That's true. 103 00:10:19,020 --> 00:10:24,820 Which, you know, but I think the true intent and there's a quote here from Professor Mandlin. 104 00:10:24,820 --> 00:10:33,420 Our hope is that through AI, we can eventually approximate a one to one teacher to student ratio for every student in CS50 105 00:10:33,420 --> 00:10:41,120 by providing them with the software based tools that 24 7 can support their learning at a pace and in a style that works best for them individually. 106 00:10:41,120 --> 00:10:43,320 Personalized learning. Here you go. 107 00:10:43,320 --> 00:10:46,420 Personalized learning tailored to their needs. 108 00:10:46,420 --> 00:10:52,720 Their chat bot delivers examples that resonate with them that are in their learning style. 109 00:10:52,720 --> 00:11:02,920 Like, you know, perhaps they have some sort of like survey at the beginning that students can take to kind of give the chat bot a persona in terms of like how it's going to give them personalized instruction. 110 00:11:02,920 --> 00:11:04,520 But this is the way it's going. 111 00:11:04,520 --> 00:11:17,320 You know, the generative AI think is going to transform tons of classrooms, the entire learning experience and in a very good way, you know, a recall based learning paradigm. 112 00:11:17,320 --> 00:11:18,820 Like it's always been. 113 00:11:18,820 --> 00:11:22,820 It's not as necessary as it used to be back in the day. 114 00:11:22,820 --> 00:11:39,520 Back in like the 1817 1600s, like you needed a recall based kind of learning style because the centers of knowledge were cities and the people that learned everything would have to go take this to areas where there weren't a lot of books. 115 00:11:39,520 --> 00:11:47,820 There weren't a lot of population and recall facts and lessons and stories and things that are happening. 116 00:11:47,820 --> 00:11:55,120 So like that's that kind of roughly built this entire framework of how at least probably you and I came up learning. 117 00:11:55,120 --> 00:11:59,220 And here we have something where you could just kind of completely circumvent that. 118 00:11:59,220 --> 00:12:05,920 You know, I don't think the need for recall is as important anymore because we can call upon information. 119 00:12:05,920 --> 00:12:08,320 It's at our fingertips at all times. 120 00:12:08,320 --> 00:12:14,520 So rather than being just a game of memorization game of to test your understanding. 121 00:12:14,520 --> 00:12:23,720 Now we have these tools that can be all right, no matter what we're going to teach you in a way that will resonate with you from a accessibility perspective. 122 00:12:23,720 --> 00:12:24,820 That's tremendous. 123 00:12:24,820 --> 00:12:36,220 But also that opens up a means of constantly teaching like critical thinking skills throughout every single, you know, every single lesson, every single topic or course. 124 00:12:36,220 --> 00:12:51,320 That to me is the true kind of magic of this, these generative AIs as tutors, because you're reinforcing a lot of different measures and practices of critical thinking that can be applied to every aspect of someone's life. 125 00:12:51,320 --> 00:12:54,520 For me, just hearing you talk comes two points. 126 00:12:54,520 --> 00:12:57,120 One is about parents paying this tuition. 127 00:12:57,120 --> 00:13:04,420 I hope finally people realize what they pay money when they send their kids to university. 128 00:13:04,420 --> 00:13:07,020 And I had this very interesting experience. 129 00:13:07,020 --> 00:13:22,020 So I don't know if I told you, but before pandemic hit, I was visiting Paris and I was planning and very much on a go to study MBA at INSEAD. 130 00:13:22,020 --> 00:13:24,220 It's a business management school. 131 00:13:24,220 --> 00:13:28,720 It's always ranked at the top, at least in Europe, if not worldwide. 132 00:13:28,720 --> 00:13:31,720 I think anybody at McKinsey would know that. 133 00:13:31,720 --> 00:13:39,720 And I had some consultation calls and one lady and I had very good conversation with her. 134 00:13:39,720 --> 00:13:55,720 And basically she was saying to me she did INSEAD back in the days and it cost over 100,000, which is crazy considering that, you know, all my educations were free, even though in prestigious schools, because this is just how like, I don't know, we are used to. 135 00:13:55,720 --> 00:13:57,420 But this one specifically you pay. 136 00:13:57,420 --> 00:13:59,320 Not here in the States, not in the States. 137 00:13:59,320 --> 00:14:00,720 I know, I know. 138 00:14:00,720 --> 00:14:11,820 But she told me this thing that when it comes down to education, especially these days, you don't pay for education. 139 00:14:11,820 --> 00:14:17,820 You pay for networking opportunities and access to alumni network. 140 00:14:17,820 --> 00:14:19,320 This is the two things. 141 00:14:19,320 --> 00:14:26,320 Yeah, you can throw in experience, but I think especially at MBA and I was looking at the executive master's degree. 142 00:14:26,320 --> 00:14:33,920 You come with an experience, you already experience, you know, a bachelor, you already experience corporate world or owning your own business. 143 00:14:33,920 --> 00:14:44,220 So this whole experience going back to the school maybe is not the first touch point where you, you know, learn how to live in a dorm and all those things. 144 00:14:44,220 --> 00:14:48,820 And then it went pandemic hit and I took this money and bought apartment. 145 00:14:48,820 --> 00:14:56,020 And I can tell you, I don't regret right now that I didn't continue in Seattle, even though it's amazing. 146 00:14:56,020 --> 00:15:03,520 It's still kind of on my list, but maybe down the line when 100,000 is not this ginormous number. 147 00:15:03,520 --> 00:15:10,620 And the second point about education in general, which we both are parents, so it's super relevant. 148 00:15:10,620 --> 00:15:15,420 I think it's rolling in your head how the whole education system is going to be looking in the future. 149 00:15:15,420 --> 00:15:21,620 And that we are seeing this transition, especially in a Harvard University scale. 150 00:15:21,620 --> 00:15:27,620 Like I want to understand, you know, some school somewhere, but this is Harvard putting out a statement like that. 151 00:15:27,620 --> 00:15:37,620 Like they are standing for something and that education system that it goes, as you mentioned, all the way back to 18, 19th century. 152 00:15:37,620 --> 00:15:43,620 But every Western education system is based on Prussian education system. 153 00:15:43,620 --> 00:15:49,620 And Prussian education system was created back then in Germany, like current Germany. 154 00:15:49,620 --> 00:15:57,420 And it was primarily created because Germans didn't listen to their soldiers during the Napoleonic wars. 155 00:15:57,420 --> 00:15:59,920 So what can you do? 156 00:15:59,920 --> 00:16:12,120 The idea was that you gather kids from age 6 to 16 and you put them in the classrooms and you teach them to listen, to follow rigid tables, to remember instructions. 157 00:16:12,120 --> 00:16:23,420 The insane aspect of that is that nothing really changed throughout all the industrial revolutions, even just automation revolution, right? 158 00:16:23,420 --> 00:16:27,820 Kids still go to school, sit in the tables, memorize text. 159 00:16:27,820 --> 00:16:35,420 And this is the way we are learning. And by the way, back then it was just for boys because of the main purpose and the reason it was free. 160 00:16:35,420 --> 00:16:39,720 And this is why it kind of was populating in Europe, free education, right? 161 00:16:39,720 --> 00:16:46,420 It was for boys because the idea was to send them to military afterwards. 162 00:16:46,420 --> 00:16:51,320 Right. And the girls were taught moms or maids. 163 00:16:51,320 --> 00:16:56,320 And then only later with industrial revolution that we need workers in the factory. 164 00:16:56,320 --> 00:17:02,220 So we need administration workers. We need people to operate sewing machines. 165 00:17:02,220 --> 00:17:09,220 Therefore, girls started going to school to memorize tables and follow orders. 166 00:17:09,220 --> 00:17:17,420 Yeah, so I am incredibly excited for education system to take different shape and form. 167 00:17:17,420 --> 00:17:29,120 What better endorsement is there than like, hey, our entire system is based on kind of a need for soldiers to listen to their superior officers on the battlefield. 168 00:17:29,120 --> 00:17:35,320 Maybe it's time to reinvent the wheel here a little bit with regard to how our young folks are educated. 169 00:17:35,320 --> 00:17:48,520 And just the point that when you have small children, I find it fascinating that a lot of times AI and how you train AIs refer to how you interact with kids and how you observe kids learning. 170 00:17:48,520 --> 00:17:59,720 And this is the truth that you adjust the information or the tools or the toys to introduce to the kid based on their development, based on their interest. 171 00:17:59,720 --> 00:18:11,720 And this is how you follow through their education. But you do that as a parent or if you're trying to be active parent with educating against and then you place them in the school and then put a stamp on them. 172 00:18:11,720 --> 00:18:19,420 But OK, now you have to learn this. Sit here. Listen, even though if it's not relevant, not interesting at all. 173 00:18:19,420 --> 00:18:25,520 I just so can't imagine going back to sitting in a classroom for for 12 years. 174 00:18:25,520 --> 00:18:30,820 I was kicking around the idea of getting a PhD and my wife's like, no, you've been a student your entire adult life. 175 00:18:30,820 --> 00:18:39,920 You better rest. But now with the onset of generative AI, why would you the same way unless you just want the pedigree? 176 00:18:39,920 --> 00:18:43,320 I mean, there's we found an interesting research paper this week to that. 177 00:18:43,320 --> 00:18:50,920 Like, hey, here is here's here teachers. Here is some direct things that you can do with prompts and prompt engineering in the classroom. 178 00:18:50,920 --> 00:18:58,620 Like right now to help things like your lesson plans, your, you know, the types of questions, the way you evaluate things. 179 00:18:58,620 --> 00:19:06,520 I think it was a it was a study by a couple of professors at the Wharton School of Interactive Education and University of Pennsylvania. 180 00:19:06,520 --> 00:19:11,220 And this is an important topic to me. I'm actually with some of my partners at SynthBinds. 181 00:19:11,220 --> 00:19:19,720 We're actively developing a course on AI for educators that we hope to get deployed in front of some teachers before the upcoming school year, 182 00:19:19,720 --> 00:19:25,420 because there's just so many things that they can do to better evaluate not only, you know, 183 00:19:25,420 --> 00:19:32,720 offload some of the burden of lesson planning and creating content for their students, but like also to educate them on it. 184 00:19:32,720 --> 00:19:39,220 Hey, this is a very real thing that your students will be using, you know, maybe let's have an understanding of it. 185 00:19:39,220 --> 00:19:48,820 It's limitations. And hopefully, you know, they work for a district that is open minded and embraces some of this stuff, 186 00:19:48,820 --> 00:19:55,820 as opposed to puts their head in the sand and pretend pretends it doesn't happen and says, no, you can't use any of this stuff. 187 00:19:55,820 --> 00:20:02,920 OK, I have a question to you. So then I was doing my keynote at this tech conference in Berlin. 188 00:20:02,920 --> 00:20:07,520 Bunch of questions. But the last question was like this cherry on top. 189 00:20:07,520 --> 00:20:18,320 And it was a grown man. And he read about me and he was like, oh, your motivation to go full on in the direction of AI and to emerge 190 00:20:18,320 --> 00:20:22,020 in technologies, which is maybe not going to be just AI. Right. 191 00:20:22,020 --> 00:20:32,220 But that direction was because I want to be part of my children's life and I want to be the reference point and I want to understand what type of world we live. 192 00:20:32,220 --> 00:20:35,520 So he referenced that and it was like standing with a huge smile. 193 00:20:35,520 --> 00:20:41,620 But then he asked me that he has teenagers are like eight, twelve and I think fourteen. 194 00:20:41,620 --> 00:20:50,820 And the question was that how now, right now, can I motivate my kids to learn this and start using this tool? 195 00:20:50,820 --> 00:20:57,520 This what he told that his kids are like, oh, yeah, it's dad talking about his stuff at work. 196 00:20:57,520 --> 00:21:02,020 Like, I don't want to deal or touch this techie stuff. 197 00:21:02,020 --> 00:21:10,320 And it also is reflected to when you look at the usage of chat GPT or image and narrative tools. 198 00:21:10,320 --> 00:21:17,520 It leans towards older segment of population, not teenagers who would be jumping on. 199 00:21:17,520 --> 00:21:19,720 Oh, my God, this is exciting new tech. 200 00:21:19,720 --> 00:21:22,520 So the question to you. So this is the premise. 201 00:21:22,520 --> 00:21:28,020 The question to you is, how do you imagine parents right now? 202 00:21:28,020 --> 00:21:30,720 How would they deal about this? 203 00:21:30,720 --> 00:21:37,920 How where do you even start educating your kids or motivating them to learn and to have this advantage even in school? 204 00:21:37,920 --> 00:21:46,820 Right. Or in university going forward versus all the other aspects that they need to know, like hallucinations, how to prompt. 205 00:21:46,820 --> 00:21:51,520 Well, security issues, personal data, you know, leaks. 206 00:21:51,520 --> 00:21:54,420 Would you jump to teach your kids? Yeah. 207 00:21:54,420 --> 00:22:00,920 OK. Oh, I mean, as both of us being optimistic, you know, generative A.I. 208 00:22:00,920 --> 00:22:04,720 optimists in terms of how this is going to shake out. 209 00:22:04,720 --> 00:22:10,820 I think so much of it goes back to a just the opportunities that abound. 210 00:22:10,820 --> 00:22:21,020 It provides a wealth of opportunities, a wealth of means to connect to teach your kids about critical thinking and problem solving. 211 00:22:21,020 --> 00:22:25,620 You know, being able to go like, hey, well, first off, from a moral standpoint, 212 00:22:25,620 --> 00:22:30,320 if you're going to put something into chat GPT and say, hey, write me a book report about this, 213 00:22:30,320 --> 00:22:36,920 like that's an opportunity to talk to them about morality and ethics being like, you know, this isn't something that you wrote. 214 00:22:36,920 --> 00:22:41,020 You shouldn't represent yourself this way because that's cheating. 215 00:22:41,020 --> 00:22:45,420 And that's an opportunity to talk about cheating with them even at a young age. 216 00:22:45,420 --> 00:22:49,720 But if they take a little bit of time to understand how these things work, 217 00:22:49,720 --> 00:22:56,320 like, for example, why they hallucinate, how the language models work, just even at a basic level, 218 00:22:56,320 --> 00:23:03,820 it kind of opens up this opportunity right away to go, OK, yeah, well, let's put this in there and see what it says about, 219 00:23:03,820 --> 00:23:07,520 you know, about to kill a mockingbird for your book report. 220 00:23:07,520 --> 00:23:13,720 And then when it comes out, let's talk about how you're going to evaluate what it said, because we know it can hallucinate a little bit. 221 00:23:13,720 --> 00:23:24,320 So invariably, if you have a discussion about going, all right, well, how can you how can you verify that the things that the language model is outputting are incorrect, 222 00:23:24,320 --> 00:23:32,820 are correct or compelling arguments? Well, you kind of have to then go figure out if it's accurate, you know, and maybe you have to read the material. 223 00:23:32,820 --> 00:23:43,720 And then along the way, they actually learn it. That to me is like the opportunity that is kind of dangling in front of parents right now as someone who's going to be a parent of a teenager eventually. 224 00:23:43,720 --> 00:23:50,020 She's not there yet. It's a conversation I'm going to have early because these things are not going away. 225 00:23:50,020 --> 00:24:06,320 I would much rather her have the tools or think about these things as to tools to do the execution part of writing, of brainstorming, of thinking, not the assemblage, not the evaluation and not the planning portion. 226 00:24:06,320 --> 00:24:14,920 Planning essentially is your prompt. What are you going to put into the prompt? How are you going to, you know, make sure that it outputs a good essay? 227 00:24:14,920 --> 00:24:21,620 All right. Well, if you just say write an essay, that's not very specific. What you're trying to make an argumentative essay, you're trying to make it persuasive essays. 228 00:24:21,620 --> 00:24:29,820 It has to be a thesis. What's the format? These are all very important questions that would go into it that that make the kid think. 229 00:24:29,820 --> 00:24:35,820 I mean, isn't that the kind of whole point of learning confronting your reality and facts and thinking about it and evaluating and see how you feel? 230 00:24:35,820 --> 00:24:43,720 That's prompt engineering right there, like thinking about it, having a plan, writing clear instructions and the better instructions you give the better result. 231 00:24:43,720 --> 00:24:52,720 That sounds like a good, good idea to me. I don't see why we want to go back any old way of just memorize this thing and then forget about it once you've done it on the test. 232 00:24:52,720 --> 00:25:02,620 So what you're talking this structure and systematic thinking and I saw this conversation going about once again, you know, prompt engineering that it's so easy, 233 00:25:02,620 --> 00:25:16,220 but it had the interesting twist saying that this just exposed how bad we are at communication and how bad we are at giving clear instructions. 234 00:25:16,220 --> 00:25:22,020 And if I think about managers I worked with, that's so true. 235 00:25:22,020 --> 00:25:31,420 You and I talked about how using more of these tools actually improve our communication, at least written, especially emails. 236 00:25:31,420 --> 00:25:34,520 And never thought going back to education. 237 00:25:34,520 --> 00:25:41,020 So what you said, developing a course for educators, for teachers and school system. 238 00:25:41,020 --> 00:25:44,320 Eventually we got computer class, right? 239 00:25:44,320 --> 00:25:56,920 Yeah, it was maybe a bit delayed and some kids who had access to computers, Elon Musk, got way, way better advantage going forward with playing with these tools. 240 00:25:56,920 --> 00:26:06,520 The same goes to Bill Gates, but some kids, let's say going back to Europe and Eastern Europe, we got our computers so late. 241 00:26:06,520 --> 00:26:12,420 Knowing how to use AI and how to apply that in your daily basis. 242 00:26:12,420 --> 00:26:14,120 Shouldn't we start talking? 243 00:26:14,120 --> 00:26:17,620 How should we educate kids going forward? 244 00:26:17,620 --> 00:26:21,120 And when this education could be a solid point, 245 00:26:21,120 --> 00:26:27,020 because one thing is to put it on parents and say that, okay, you are the parent, this is up to you. 246 00:26:27,020 --> 00:26:38,620 But if we look at the statistics, that what it's 52% of US, which is primarily using child GPT, 52% of US heard of it, 247 00:26:38,620 --> 00:26:45,620 14% tried and tried it doesn't mean mastered it and understand it and all of those aspects. 248 00:26:45,620 --> 00:26:52,420 And the courses what we are seeing are also at the level of comprehension, not of a teenager, let's say. 249 00:26:52,420 --> 00:27:02,620 So I think we need to start having conversation for the people who, you know, 14, 15, 16, whatever age it's appropriate. 250 00:27:02,620 --> 00:27:06,520 How do we introduce education to kids about this emerging technology? 251 00:27:06,520 --> 00:27:07,920 I completely agree. 252 00:27:07,920 --> 00:27:11,220 I mean, I hope to be part of that conversation. 253 00:27:11,220 --> 00:27:19,120 I will take a look at some if there is anything happening, looking at the education system as it is right now, 254 00:27:19,120 --> 00:27:22,320 how long it takes to just update books. 255 00:27:22,320 --> 00:27:24,920 You know, I think I've used this analogy before. 256 00:27:24,920 --> 00:27:30,820 As you mentioned, the computer, we gave students calculators in the 60s, 70s, 80s. 257 00:27:30,820 --> 00:27:31,920 We didn't get rid of math. 258 00:27:31,920 --> 00:27:34,820 We just changed how we evaluated them on it. 259 00:27:34,820 --> 00:27:36,120 Same is going to happen here. 260 00:27:36,120 --> 00:27:40,020 It might be a little bigger and broader and it's hard to kind of picture right now, 261 00:27:40,020 --> 00:27:41,420 but it's going to take place. 262 00:27:41,420 --> 00:27:50,220 And I think with that is a good enough place as any to say for Go2Go, I am West the Synth Mind saying happy prompting everybody. 263 00:27:50,220 --> 00:27:54,020 Happy learning prompting everybody. 264 00:27:54,020 --> 00:28:01,520 Thanks for listening to How to Talk to AI with your hosts, Go2Go and West the Synth Mind. 265 00:28:01,520 --> 00:28:08,420 As always, you can check out the show notes and links at howtotalk2.ai. 266 00:28:08,420 --> 00:28:10,820 That's all for this week's episode. 267 00:28:10,820 --> 00:28:38,820 Happy prompting, everyone.