WEBVTT NOTE This file was generated by Descript 00:00:11.906 --> 00:00:17.966 Welcome to Transformative Principle, where I help you stop putting out fires and start leading. 00:00:18.476 --> 00:00:19.976 I'm your host, Jethro Jones. 00:00:19.976 --> 00:00:22.676 You can follow me on Twitter at Jethro Jones. 00:00:35.471 --> 00:00:35.981 Alright. 00:00:35.981 --> 00:00:38.441 Welcome everybody to Transformative Principal. 00:00:38.441 --> 00:00:43.781 I am very excited to have on the podcast today, Linda Burrick. 00:00:43.841 --> 00:00:51.911 She is a behavioral scientist, uh, behavioral scientist, and she does all kinds of amazing cool things. 00:00:52.271 --> 00:01:00.101 Uh, she runs a consultancy called Linda b Learning out of Seattle, and we started talking. 00:01:00.861 --> 00:01:20.751 Uh, last week before we, uh, decided to do the interview and I was like, we definitely gotta talk, because Linda's been in the learning space for a long time, has a ton of really fascinating perspectives and information and experience about, uh, artificial intelligence. 00:01:20.751 --> 00:01:23.781 Before it was the marketing word, artificial intelligence. 00:01:23.781 --> 00:01:26.871 So Linda, welcome to Transformative principle. 00:01:26.871 --> 00:01:27.471 Great to have you 00:01:27.626 --> 00:01:30.686 Thanks so much for inviting me and having the opportunity to chat. 00:01:31.871 --> 00:01:33.941 Yeah, I'm, I'm really excited to talk with you. 00:01:33.941 --> 00:01:37.871 Our conversation last week really inspired me and made me think about some things. 00:01:37.871 --> 00:01:46.691 And I wanna start with this idea of artificial intelligence and how, uh, you mentioned that it's just a marketing term and that you've been doing this for a long time. 00:01:46.871 --> 00:01:53.861 So can you just give like a brief idea of some of the things that you've worked on, uh, already in your career? 00:01:53.996 --> 00:01:54.686 Yeah. 00:01:54.716 --> 00:01:55.196 Yeah. 00:01:55.196 --> 00:02:03.626 So when I talk about artificial intelligence, I use the term machine learning, and then sometimes I refer to the specific kind of statistical learning. 00:02:04.556 --> 00:02:05.966 You know, category that it is. 00:02:05.966 --> 00:02:07.916 So that's typically how I talk about it. 00:02:07.916 --> 00:02:13.436 But because AI is such a buzzword, I, I will use those terms, but please know, that's what I'm talking about. 00:02:13.856 --> 00:02:23.411 And so from that perspective, really, when you're building anything with a computer that helps you memorize or learn things, I. You are building like a machine learning peer. 00:02:23.591 --> 00:02:27.041 And I don't know that people think about it that way, but that's how I think about it. 00:02:27.041 --> 00:02:31.571 And that was literally the first experience I had with machine learning when I was 14 years old. 00:02:31.871 --> 00:02:38.471 And then, you know, as I got into graduate school, um, I was studying behavior analysis, so that's mostly human and. 00:02:39.206 --> 00:02:43.766 Animal learning, but we also worked with plants because plants actually can learn. 00:02:43.856 --> 00:02:46.796 And then for me, I'm like, well, machine learning, obviously, duh. 00:02:46.886 --> 00:02:58.226 So I mean, I worked on things like some of the very early, uh, voice recognition and voice synthesis, um, technologies used because pop people thought that would be a way better way of interacting or interfacing with a computer. 00:02:58.601 --> 00:03:04.211 Not realizing how difficult it actually is to match a voice when you're thinking about how that actually works. 00:03:04.811 --> 00:03:06.731 Um, and then just progressed on from there. 00:03:06.731 --> 00:03:12.191 You know, I built, I, uh, assisted Paul Allen building one of the very first online schools ever before there was a thing. 00:03:12.431 --> 00:03:15.191 Uh, built my own companies and sold it when I was 28. 00:03:15.191 --> 00:03:21.761 And then continued to work, uh, with different companies in Silicon Valley, mostly Sun Microsystems in Oracle for a substantial part of my career. 00:03:21.761 --> 00:03:23.951 But lots of other, uh, places as well. 00:03:23.951 --> 00:03:24.401 And really. 00:03:24.936 --> 00:03:31.566 My role was to get closer to the technology because I wanted to not just be a learning person who develops training. 00:03:31.926 --> 00:03:34.416 Um, you know, and that was something I always used to get is like, well. 00:03:34.796 --> 00:03:37.406 For a training person, you really don't want to do training. 00:03:37.406 --> 00:03:38.546 I'm like, oh, I do training. 00:03:38.546 --> 00:03:39.776 I just put it into the software. 00:03:39.776 --> 00:03:41.846 I'm not about like, everything has to be in a classroom. 00:03:42.056 --> 00:03:42.416 We are. 00:03:43.166 --> 00:03:47.876 That ship has sailed like decades, you know, before, even when I first started when I was a teenager. 00:03:47.876 --> 00:03:53.516 So like where we are now with virtual reality and all these other kinds of immersive technologies, the internet of thing called computing. 00:03:54.056 --> 00:03:55.706 It's opened up so much possibility. 00:03:55.736 --> 00:04:03.806 But the challenge I think, for people, especially in the K 12 space and higher ed space, is just, you know, sifting through the hype from. 00:04:04.391 --> 00:04:07.596 What can you actually do and how can it be beneficial for the populations that we're serving? 00:04:09.391 --> 00:04:24.341 Well, and what I, what I think is so interesting is you and I took a, uh, a similar path that we said, how can we make this happen more automatically so that we don't have to spend the human cycles doing the things that machines can do? 00:04:24.671 --> 00:04:26.141 Better than us. 00:04:26.351 --> 00:04:42.071 And so you talk about that idea, a little bit of outsourcing, repetition, repeated learnings, repeated practice kinds of things to technology and why that makes sense and has made sense for decades now. 00:04:42.851 --> 00:04:43.781 Yeah, well. 00:04:45.311 --> 00:04:48.371 Think about what we call these things that we're interfacing with. 00:04:48.371 --> 00:04:49.931 We call them computers. 00:04:49.931 --> 00:04:53.771 So that alone should suggest to you what they're really good at doing. 00:04:53.771 --> 00:04:57.401 What technology is, was the underpinning of what technology is really good at. 00:04:57.401 --> 00:04:59.471 And that is computing, right? 00:04:59.471 --> 00:05:00.911 Being able to put things together. 00:05:01.151 --> 00:05:08.471 So I remember back in the day, and you probably remember this too, when you were told that you know, can't rely on the calculator, you know you're not gonna have a calculator and then you know, you look at your. 00:05:08.531 --> 00:05:10.121 Smartphone and you're like, oh, really? 00:05:10.811 --> 00:05:11.201 Yeah. 00:05:11.921 --> 00:05:16.031 um, so there is so many things that are just in your hand and just in your pocket that you can do. 00:05:16.451 --> 00:05:18.131 But the challenge is, is that. 00:05:19.931 --> 00:05:34.121 We're kind of like doing the wrong things with the devices, so yeah, computing, calculating, sorting, memorization, those kind of repetitive skills where there's a specific kind of response that you're after and it's a correct response. 00:05:34.211 --> 00:05:41.591 Like I just wrote a newsletter about psychomotor learning and the whole underpinning of psychomotor learning is this notion of there's good form. 00:05:41.616 --> 00:05:44.346 So, and that's what you coach to, is what is that good form? 00:05:44.346 --> 00:05:54.756 You know, that Olympic level performance or that surgeon who's now ready to operate, or that, um, Potter who's working at the wheel shaping that pottery bowl. 00:05:54.786 --> 00:05:58.507 How do they get to the point where their motor skills are done in a way where they's. 00:05:58.941 --> 00:06:00.111 They're exhibiting good form. 00:06:00.291 --> 00:06:07.221 So for psycho motor skills, it becomes really, really obvious when we're trading up physical skills, and we even think about it when you watch Chi children develop, right? 00:06:07.221 --> 00:06:16.431 And they're, they get feedback from their natural involvement environment on how to walk, but they also get feedback as the repertoires get more complex from a skilled observer. 00:06:16.761 --> 00:06:21.231 And sometimes the computer can be a better skilled observer depending on what it is that you're training up. 00:06:21.861 --> 00:06:25.071 But if we're trying to train up things that are creative in nature, like. 00:06:25.901 --> 00:06:38.831 Oh, we could have a whole conversation about generative AI and GPTs and that whole thing because I was, again, I was very early in the front of that as well and actually had the opportunity to work for one of the bigger vendors and after reviewing their model declined. 00:06:39.251 --> 00:06:46.871 But um, but yeah, there's definitely things that computers are very good at and can be excellent coaches for. 00:06:47.606 --> 00:06:50.906 It comes down to this notion of what's the res required response? 00:06:50.906 --> 00:07:01.316 And we already get a bad rap in, in K 12 and higher ed around too much focus on memorizing bodies of curriculum as opposed to critically thinking about them. 00:07:01.316 --> 00:07:01.646 Right? 00:07:01.856 --> 00:07:10.526 And so it's not to say that computers can't be, you can't design computer programs that can challenge and, and help augment the development of, of, um. 00:07:11.186 --> 00:07:19.106 Um, creative behavior or, or, you know, problem solving and those kinds of more complex, uh, complex cognitive skills. 00:07:19.496 --> 00:07:23.276 It's just the way you get there isn't the way we're doing it right now. 00:07:23.276 --> 00:07:29.876 You don't apply this mechanistic, step-by-step procedural methodology to train a far transfer complex task. 00:07:30.356 --> 00:07:38.906 And, and quite honestly, sometimes it's like, well, do we even want to, like, there's so many things where if you consider the emotional underpinning of learning, which is present, we can talk again. 00:07:39.581 --> 00:07:44.441 Whole conversation about what that means, but the, that's where the human element really comes in. 00:07:44.711 --> 00:07:50.351 And so there's definitely ways that feedback can be reinforcing just in terms of build, building skill development. 00:07:50.441 --> 00:07:53.111 And that's motivating, but there's also the human touch. 00:07:53.141 --> 00:07:59.141 And especially in the digital world that we're in, coming outta COVID, we're almost forgetting how to be human, you know? 00:07:59.171 --> 00:08:08.081 'cause we're not spending a lot of, as much time even now in, in actual, in-person, you know, in the room kind of experiences. 00:08:08.231 --> 00:08:08.501 Yeah. 00:08:09.296 --> 00:08:10.286 Yeah, that's true. 00:08:10.346 --> 00:08:18.236 I mean, it's been five years since Covid, and yet it is still fresh on everybody's mind, so it's not like it's, um, like it's that far in the rear view mirror. 00:08:18.236 --> 00:08:24.386 But you, you started talking about these near transfer versus far transfer and organic and mechanistic skills. 00:08:24.386 --> 00:08:28.616 And so I want to talk about that a little bit, um, because that's something that I've been. 00:08:28.766 --> 00:08:39.956 Thinking a lot about that these mechanistic skills, which I'm defining as skills that you can easily say, this is how to do it. 00:08:39.956 --> 00:08:44.456 This is the right way, there's a wrong way, and I can transfer this knowledge to you. 00:08:44.456 --> 00:08:48.656 And by transferring the knowledge and you practicing, you can show that you've got it. 00:08:48.746 --> 00:08:50.936 And, and that is sufficient. 00:08:50.966 --> 00:08:54.866 And you know, that is the memorization, the, the, uh. 00:08:55.331 --> 00:08:59.891 The, the process skills, the things that we know need to happen in schools. 00:09:00.101 --> 00:09:07.151 Um, also a lot of times considered, uh, secondary, biological knowledge, I think is another term they use for it. 00:09:07.151 --> 00:09:14.861 But those are things where it's like, how do you, how do you live in this world and do the things that you need to do that are like, there's a yes and a no right? 00:09:14.861 --> 00:09:20.021 And a wrong way, binary skills, whereas organic skills are skills that are. 00:09:20.501 --> 00:09:29.381 Things that, uh, there's a right way, there's a better way, there's a really good way, there's an okay way and there's a really bad way. 00:09:29.441 --> 00:09:31.811 And there's this gradient of how to deal with it. 00:09:31.811 --> 00:09:41.471 Things like empathy, grit, perseverance, uh, patience, all all kinds of, these, what we typically call soft skills that I'm now calling organic skills. 00:09:41.711 --> 00:09:46.091 And then you, uh, brought up near transfer and far transfer skills. 00:09:46.091 --> 00:09:47.231 And how do those relate? 00:09:47.231 --> 00:09:48.311 Let's talk about that a little bit. 00:09:48.371 --> 00:09:48.941 Yeah. 00:09:48.941 --> 00:09:49.241 Yeah. 00:09:49.241 --> 00:09:52.061 So I, I touched on it a little bit, but let's go a little deeper on that. 00:09:52.061 --> 00:09:55.331 So, a near transfer skill is procedural in nature. 00:09:55.331 --> 00:10:00.131 So to your point, there's a right way to do it and everything else, right? 00:10:00.191 --> 00:10:00.941 And so. 00:10:01.916 --> 00:10:03.836 Think about like making a Starbucks latte. 00:10:03.866 --> 00:10:07.676 There is a step one, step two, step three, step four on how you do it. 00:10:07.796 --> 00:10:13.346 And yes, there's variations on it and, but for the most part there is, this is the way you build it. 00:10:13.346 --> 00:10:18.986 And that's not to say that that's representative of all lattes, that's representative of Starbucks lattes for standardization purposes. 00:10:18.986 --> 00:10:19.286 Right? 00:10:19.526 --> 00:10:24.086 So that's the idea of near transfers, uh, near transfer tasks and near transfer training. 00:10:24.086 --> 00:10:29.636 Is that the way that you teach it is the way that you, uh, the way that you perform it on the job. 00:10:30.281 --> 00:10:32.321 In real life is how you train it up. 00:10:32.621 --> 00:10:35.141 So it's not, there's no variation to it. 00:10:35.141 --> 00:10:36.911 It's always done the same kind of way. 00:10:36.911 --> 00:10:41.291 So that's what for far transfer, um, usually or near transfer usually refers to. 00:10:41.381 --> 00:10:47.141 And I should say too, that's comes from Ruth Clark, who is very prominent in technical training and corporate training. 00:10:47.141 --> 00:10:52.781 So that's comes from her, uh, Noma, Cher, that's where the, the terms near transfer and far transfer comes from. 00:10:52.961 --> 00:10:57.641 And she also talks about 'em in terms of procedures being near transfer tasks, and then. 00:10:57.891 --> 00:11:00.231 Principle based learning being far transfer tasks. 00:11:00.231 --> 00:11:05.421 So if near transfer tasks is, it's always done the same way every single time, under every condition. 00:11:05.421 --> 00:11:06.891 Step one, step two, step three. 00:11:07.461 --> 00:11:09.951 Far transfer tasks are those tasks that are more. 00:11:10.256 --> 00:11:11.336 Strategic in nature. 00:11:11.336 --> 00:11:19.436 So you form a strategy about how you're gonna approach the task at hand, meaning that depending on the context, what you're gonna do varies. 00:11:19.526 --> 00:11:23.636 And so, like a classic example that people would often give is making a sales call. 00:11:23.696 --> 00:11:31.046 It's not just step one, step two, step three because how the customer responds or how the potential prospect response matters. 00:11:31.051 --> 00:11:33.956 And you're gonna, uh, start pivoting and do things differently. 00:11:34.256 --> 00:11:35.546 So far, transfer tasks. 00:11:36.056 --> 00:11:41.516 Are also referred to oftentimes as complex cognitive tasks, whereas they, you would call a procedural task. 00:11:41.846 --> 00:11:44.576 Um, some sometimes referred to as a simple cognitive task. 00:11:44.576 --> 00:11:45.866 So these are different theorists. 00:11:45.866 --> 00:11:53.486 So that la latter term that comes from teman and markle, that's very top of mind since the psychomotor work I was just writing about comes from them as well. 00:11:53.636 --> 00:11:54.056 Right. 00:11:54.056 --> 00:11:57.566 So, and the way that they propose it is that everything builds on everything. 00:11:57.566 --> 00:12:01.886 That there's a psychomotor response, a physical response using your musculature and your, your. 00:12:02.156 --> 00:12:04.316 Um, anatomy and physiology to perform. 00:12:04.526 --> 00:12:12.326 That's a precursor for the simple cognitive tests, which are precursors for, um, complex cognitive, and they are like hardcore programmed instruction behaviorists. 00:12:12.326 --> 00:12:14.426 So that's where that particular theory comes from. 00:12:14.726 --> 00:12:29.786 But this, and, and quite honestly, it's worth revisiting programmed instruction considering the computer age we're in now because so much of that was dependent on our understanding of how computers and information processing and memory and those pieces work. 00:12:30.761 --> 00:12:42.641 And now if you look at it in the context of, like I said, cloud computing, internet of things, generative ai, you know, then it starts to take on, oh, so what can we actually program? 00:12:42.971 --> 00:12:46.121 And again, you can program things that have a definite right and a definite wrong. 00:12:46.331 --> 00:12:50.741 This really comes to bear, like when we start thinking about subjects where history. 00:12:51.296 --> 00:12:52.886 Even geography, right? 00:12:52.886 --> 00:12:59.306 Some people have, have had questioned is like, are maps really representative of what our actual world actually looks like? 00:12:59.606 --> 00:13:13.136 Or is that even a skewed response based on, you know, political and you know, the ruling power or the Victor's powers presentation of what geography is or what history is, so, 00:13:13.196 --> 00:13:16.976 Yeah, so, so you talked about these, uh, standardizable 00:13:17.201 --> 00:13:17.621 mm-hmm. 00:13:17.666 --> 00:13:24.866 and, and much of what we do in schools is really standardized information, and that's what we're testing on. 00:13:24.866 --> 00:13:25.856 That's what we're assessing. 00:13:25.856 --> 00:13:28.916 That's why we use multiple choice tests and things like that. 00:13:29.246 --> 00:13:33.356 And so there are, there are things where it makes sense to use. 00:13:33.686 --> 00:13:38.066 These mechanistic approaches to, to learning things. 00:13:38.126 --> 00:13:48.206 And so, uh, for example, um, we know how to teach, uh, reading with phonemic segmentation, oral reading fluency, those kinds of things. 00:13:48.506 --> 00:13:55.826 And, and so that makes sense for that to be something that we use a computer to teach. 00:13:56.186 --> 00:14:01.796 Whereas other things, um, like how to understand what an author is saying in the book. 00:14:02.141 --> 00:14:08.501 Is probably not going to be the, the same way that we teach someone how to read. 00:14:08.711 --> 00:14:13.241 How do you determine what kinds of things are worthwhile to use? 00:14:14.501 --> 00:14:21.821 specifically machine learning technology for learning and not use that for learning. 00:14:22.421 --> 00:14:26.021 I think for, for me, it is definitely. 00:14:26.996 --> 00:14:28.706 The context in which you're gonna use it. 00:14:28.706 --> 00:14:36.176 So I actually love using computers for peer practice partners on those tasks that you were describing. 00:14:36.206 --> 00:14:42.236 Oral reading, fluency, math, fact memorization, those things where there's definitely right, definitely wrong answers because I. 00:14:44.066 --> 00:14:45.746 One, the computer is always accurate. 00:14:45.806 --> 00:14:48.296 It always gives you correct feedback. 00:14:48.326 --> 00:14:52.526 So you actually, so it's not like, oh, you want, like a human is, has human foibles, right? 00:14:52.526 --> 00:14:53.636 They will look away. 00:14:53.636 --> 00:15:02.726 They will not listen or they will nod and say, Ah-huh, even though, so they're not as reliable as a computer is in terms of giving feedback when there's a correct response. 00:15:03.326 --> 00:15:09.206 But when there could be variation, you run into challenges, right? 00:15:09.206 --> 00:15:09.746 And so. 00:15:09.991 --> 00:15:10.411 Mm-hmm. 00:15:10.586 --> 00:15:14.756 What I've seen happen with GPT models recently is it falls into a couple categories. 00:15:14.756 --> 00:15:27.056 So early GPT, when you would ask questions of it, it would, it kind of made you feel bad about asking questions because it came back with this very like, you know, just a tone to it. 00:15:27.056 --> 00:15:31.228 Even though you know, you know, it's just uh, uh, you know, basically. 00:15:31.586 --> 00:15:33.326 A word processor on steroids, right? 00:15:33.326 --> 00:15:35.546 That's putting together the probability of this word coming. 00:15:35.546 --> 00:15:37.466 You know, that's how they're constructing these answers. 00:15:37.466 --> 00:15:43.406 And I mean, that's just a very, very simplified version of talking about how GPTs work, but that's essentially it. 00:15:43.406 --> 00:15:48.056 Like it, the fact that it's able to produce the kinds of response that it is and make you feel some kind of way about it. 00:15:48.956 --> 00:15:52.436 That's your own anthropomorphism that you're introducing into what you're seeing. 00:15:52.436 --> 00:15:53.366 But it's, it's true. 00:15:53.366 --> 00:16:02.576 Like people would comment on the, the, you know, for lack of a better term, the mansplaining, um, tone of the original versions of Chatt PT 3.5 and on. 00:16:03.056 --> 00:16:08.696 But now we're seeing in this emphasis, we were talking about empathy, you know, to make the computer. 00:16:09.086 --> 00:16:22.706 Interaction appear more empathetic is this whole thing where it's almost pandering to you and telling you how wonderful you are and how special you are and how that was a really insightful even it's like having some garbage response. 00:16:22.706 --> 00:16:22.946 You know? 00:16:22.946 --> 00:16:27.116 You could have put in any other series of words and it would've picked even that would make no sense. 00:16:27.116 --> 00:16:28.136 Near you are praising me on how. 00:16:28.196 --> 00:16:28.616 well at. 00:16:29.651 --> 00:16:32.531 And the challenge to that is that that is actually. 00:16:33.551 --> 00:16:36.041 helpful to say that it's been, that. 00:16:36.041 --> 00:16:42.461 It's like, oh, that was a great response, good catch when like, no, that don't patronize me. 00:16:42.461 --> 00:16:45.881 Don't, don't tell me that this was good when it actually wasn't. 00:16:46.211 --> 00:16:53.891 Because if I'm not familiar with what that is doing, then I might believe that and that's another big problem. 00:16:54.131 --> 00:16:54.611 So. 00:16:54.671 --> 00:17:01.031 So before we move into that arena, what you're saying really is that things like the. 00:17:01.481 --> 00:17:05.501 The, the terrible phrase we used to use is drill and kill for specific things. 00:17:05.741 --> 00:17:07.541 That's what the technology is really good 00:17:07.606 --> 00:17:08.651 It's excellent for that. 00:17:08.711 --> 00:17:09.071 Yep. 00:17:09.191 --> 00:17:20.171 that's what we should be using it for, because it can create answer correctly on those things, millions of times faster than we can do it. 00:17:20.171 --> 00:17:20.381 So 00:17:20.426 --> 00:17:21.806 And consistently. 00:17:21.896 --> 00:17:22.586 Oh, and that's, 00:17:22.631 --> 00:17:23.291 consistently 00:17:23.636 --> 00:17:29.126 that's super important too, because that's an important thing that we can come back to talking about GPTs, why they're also problematic. 00:17:29.126 --> 00:17:32.126 It's not just this pandering fe way of giving feedback. 00:17:32.306 --> 00:17:32.936 It's that. 00:17:33.201 --> 00:17:38.031 You can run a trial of exactly the same interaction and you're gonna get different feedback. 00:17:38.031 --> 00:17:51.561 So GPTs aren't consistent because they aren't, you are not giving a consistent response, and so they're not gonna give you consistent feedback because what they're giving you feedback on depends on your response versus in the drill and kill. 00:17:52.166 --> 00:18:08.066 You're always gonna have correct feedback now, and that's not to say that you can't train computers to be able to give better feedback and to be able to like for more complex cognitive skills and to be able to better help the human come to their own conclusions. 00:18:08.336 --> 00:18:11.696 But it's not done with GPT technology, like it is absolutely not. 00:18:12.491 --> 00:18:14.051 But you can code it so you can be done. 00:18:14.051 --> 00:18:25.751 I've done small language models, so as compared to large language models, I've done small language models that actually work very well to shape behavior into a behavioral repertoire or behavioral class. 00:18:26.021 --> 00:18:31.181 So when we're talking about training up things like concepts or you know, more strategic learning. 00:18:32.036 --> 00:18:35.276 We have a frame of referent, we have to have some frame of referent, right? 00:18:35.276 --> 00:18:36.836 It's the same thing when you're teaching them in school. 00:18:36.836 --> 00:18:48.476 You have your rubric of what you're looking for, but the way that you apply that rubric is gonna vary based on the circumstances under, you know, that come to bear and the what's being presented to you in the case at hand. 00:18:49.736 --> 00:18:53.336 Yeah, well, and, and that hints on my big problem. 00:18:53.636 --> 00:19:01.706 With any kind of grading in school is that it is all subjective and we try to make it objective, but it is not. 00:19:01.736 --> 00:19:05.816 And what I constantly tell my own children all the time is grades are made up. 00:19:07.319 --> 00:19:12.329 once you understand that grades are made up, then you, they can, one, stop having power over you. 00:19:12.539 --> 00:19:15.209 And two, you can understand that you are playing a game. 00:19:15.409 --> 00:19:23.299 And you need to figure out the rules that the teacher is, is going by as she or he grades your work because it is all made up. 00:19:23.299 --> 00:19:33.259 And anybody who argues that it's not and that it is actually objective, has no basis in reality, they're just absolutely wrong because every teacher makes up their own grading system. 00:19:33.289 --> 00:19:44.269 Even if they have a rubric, even if they have standards and guides from the, the district or the state or whatever it is, they still make that ultimate decision about whether or not it is. 00:19:44.684 --> 00:19:48.884 Uh, it is a certain thing and that makes it subjective. 00:19:48.944 --> 00:19:58.394 And so understanding that and being okay with it is one thing and like saying, all right, I get what's going on here and I, and I can react in a different way. 00:19:58.664 --> 00:20:07.934 But the problem is that we pretend like they are actually objective and that that is part of a, a bigger problem that we're not gonna go all into here. 00:20:08.894 --> 00:20:11.414 you brought up this example of, of GPTs. 00:20:11.879 --> 00:20:14.279 Reacting differently to the same prompt. 00:20:14.279 --> 00:20:15.509 And I actually did this. 00:20:15.509 --> 00:20:24.089 I have, I'm in a doctoral program right now and I had an assignment where we had to basically write a paper on what we've learned this, uh, this semester in class. 00:20:24.389 --> 00:20:29.639 And to me, this assignment was just really tedious and I did not. 00:20:29.954 --> 00:20:42.254 Want to do it because it, it did not, I, I didn't see the value in it, and so I decided to make it meaningful to me, which is often how I approach, uh, my assignments in school, is I figure out a way to make it meaningful. 00:20:42.254 --> 00:20:55.694 So what I did is I put it into two different gpt to see how it would write it, uh, based on my notes, what my dissertation topic is, um, and what, what other things I could feed all my assignments and stuff. 00:20:55.694 --> 00:20:57.224 So I put all my assignments in there. 00:20:57.494 --> 00:21:04.934 I put my notes that were very like basic chicken scratch notes that I had just typed into my note taking system. 00:21:05.924 --> 00:21:09.824 And then a few different documents also to like give context to it. 00:21:10.124 --> 00:21:19.124 And then I said with one that I had already trained to have my voice and sound like me as best it could, uh, and to know what next word it should predict for 00:21:19.219 --> 00:21:19.639 Mm-hmm. 00:21:19.994 --> 00:21:23.774 on, on that I use that to, to write this. 00:21:24.419 --> 00:21:29.939 Paper and then I used another one that I hadn't trained on all my information and I just said, here's the stuff. 00:21:29.939 --> 00:21:32.849 Now make it how it should, like how it makes sense. 00:21:33.149 --> 00:21:34.229 The second one. 00:21:34.619 --> 00:21:36.929 Um, made it very academic. 00:21:37.259 --> 00:21:41.549 Included the citations for my other assignments and said, this is where I talked about this. 00:21:41.789 --> 00:21:44.429 And it sounded very much like an academic thing. 00:21:44.639 --> 00:21:46.589 The one that I had trained to sound like me. 00:21:46.799 --> 00:21:57.569 I had to do way more editing because one, it didn't sound like me, and two, it, uh, it got a bunch of things wrong, had more hallucination, but that other one was very academic. 00:21:57.569 --> 00:22:00.839 I did, I had to do very little editing because. 00:22:01.169 --> 00:22:04.289 It already didn't sound like me because it wasn't trying to sound like 00:22:04.334 --> 00:22:04.694 Yeah. 00:22:04.709 --> 00:22:10.139 one that was trying to sound like me made mistakes and hallucinated and put things in that weren't there. 00:22:10.379 --> 00:22:13.139 And it was so fascinating because I did the same exact prompt. 00:22:13.139 --> 00:22:23.549 I just copied it from one or the other and, and let it do its thing and, and then as I was giving it feedback, going paragraph by paragraph, which I also did, then it was. 00:22:24.509 --> 00:22:31.649 It would, it would give different responses based on what I said, and then it would add stuff in that weren't there and things like that. 00:22:31.649 --> 00:22:36.419 And so you can't, can't think that this is a perfect technology. 00:22:36.659 --> 00:22:43.709 However, it was pretty amazing that it could do what it could do and that it could generate this stuff from what I had said. 00:22:44.039 --> 00:22:46.619 And, and so that part was pretty cool. 00:22:46.619 --> 00:22:48.719 But if you don't understand what it's really doing. 00:22:49.079 --> 00:22:58.649 It's so easy to think that it's either magic or that it's always right and, and that are, those are the two big problems in my mind with it, is that it's not always right. 00:22:58.649 --> 00:23:02.129 It's basically just guessing and doing a good job of it most of the time. 00:23:02.129 --> 00:23:05.159 But it's really just guessing and it doesn't actually know the right answer. 00:23:05.664 --> 00:23:06.084 Yeah. 00:23:06.144 --> 00:23:06.624 But it does. 00:23:06.654 --> 00:23:07.254 But yeah. 00:23:07.254 --> 00:23:08.154 But again. 00:23:08.969 --> 00:23:10.739 Don't look behind the curtain, you know? 00:23:10.739 --> 00:23:11.999 And it all seems like magic. 00:23:12.359 --> 00:23:19.019 Yeah, it's a, I mean, don't get me wrong, I've grown up with the evolution of this, so I think it's super cool that we can do this. 00:23:19.019 --> 00:23:21.389 And you were mentioning about training up your own voices. 00:23:21.389 --> 00:23:35.579 That was literally some of the first work that I did in this area was working for, like, I, I left Oracle in 2019, so I'd take on a few gigs working for some social media influencers, and they were just telling me the big problems about, ah, I can't respond to everybody and I don't, I'm losing followers because I'm not responding. 00:23:35.579 --> 00:23:36.119 I'm all like. 00:23:36.494 --> 00:23:37.334 Cool story, bro. 00:23:37.334 --> 00:23:39.134 I could totally make a bot that could respond. 00:23:39.134 --> 00:23:40.724 That sounds just like how you respond. 00:23:40.724 --> 00:23:47.564 And I did be, but again, the kind of responses were very much tailored to whatever, you know, what they would've said. 00:23:47.564 --> 00:23:53.384 So I had, and we had tons and tons of data to build with, and it was just a, it was a small, uh, language model. 00:23:53.564 --> 00:23:55.604 I didn't need to scrape the entire internet. 00:23:55.784 --> 00:23:58.364 The reason that your paper sounded so good in academic was. 00:23:58.394 --> 00:24:08.744 All of those academic models that had to draw from, in addition to, as opposed to, you know, there's a time when you wanna narrow the field in terms of what data you wanna draw from, and then there's a time where you wanna widen the field. 00:24:08.804 --> 00:24:10.154 And so that's the difference. 00:24:10.154 --> 00:24:22.064 Like, like with search, you wanna widen the field, you know, you wanna be able to find the thing that you're looking for, but you wanna be able to wi, it's sort of like there's a concept when you play sports about having wide focus and then narrowing your focus. 00:24:22.064 --> 00:24:26.774 So when you're out on the field and you're sort of taking everything in and then there's something that you need to beeline on. 00:24:27.314 --> 00:24:28.574 That's how search should work. 00:24:29.234 --> 00:24:31.934 It doesn't, it's because it's, it's not built for that. 00:24:31.934 --> 00:24:33.674 It's not built to really search and be helpful. 00:24:33.674 --> 00:24:45.434 I mean, I think that was the intention originally, but we're so far from that where we've gone with what we can do with the compute capability that we have now and, and, and the connectivity that we have now. 00:24:45.434 --> 00:24:50.354 I mean, it just, it's frustrating and having spent like, you know, 15 years working in Silicon Valley and just. 00:24:50.924 --> 00:24:58.004 I know why we are where we are, but I'm just, I, I just wanna shake some of my, all the tech bros I've worked with, I'm like, why though? 00:24:58.124 --> 00:24:59.624 You're not the only one who has ideas. 00:24:59.624 --> 00:25:00.074 And guess what? 00:25:00.074 --> 00:25:04.094 Lots of people think way more creatively than you because you think you're the default. 00:25:04.094 --> 00:25:07.724 And that, you know, like I run into it constantly and I'm like. 00:25:07.944 --> 00:25:08.724 Oh my God. 00:25:08.724 --> 00:25:11.364 Do you know how, how much hubris you have? 00:25:11.364 --> 00:25:14.874 Do you, do you recognize like how much you just center yourself? 00:25:14.874 --> 00:25:16.554 Like just by default. 00:25:16.554 --> 00:25:17.304 I'm like, oh my. 00:25:17.394 --> 00:25:18.444 Like really, dude? 00:25:18.444 --> 00:25:23.814 And now you've built this megaphone that just amplifies all this out and said, this represents the world. 00:25:24.024 --> 00:25:25.824 And I'm like, no. 00:25:25.824 --> 00:25:26.274 Oh, no. 00:25:26.634 --> 00:25:28.944 Talk about making people feel not seen, I mean, 00:25:29.229 --> 00:25:30.629 Yeah, exactly. 00:25:30.629 --> 00:25:36.149 So, so with what we have access to now, focusing on schools, 00:25:36.174 --> 00:25:36.474 mm-hmm. 00:25:36.509 --> 00:25:42.719 where should we be putting our time, energy, and our money in investing in using these tools? 00:25:42.719 --> 00:25:44.789 And what should that look like in your opinion? 00:25:45.509 --> 00:25:49.199 I am so I loved your take on grades. 00:25:49.259 --> 00:25:56.069 The challenge with grades, of course is that is not that, you know, yes, we can agree that they're subjective and that, you know, it's, they're inconsistent. 00:25:56.189 --> 00:25:59.339 However, they have a massive impact on people's lives. 00:26:00.134 --> 00:26:10.544 And so they, for that reason, that's why you can't just like go, oh yeah, I don't care about whether I get the A, but I have noticed being that I'm not educated formally until I got to graduate school in the United States. 00:26:10.934 --> 00:26:12.494 Big cultural differences with that too. 00:26:12.794 --> 00:26:23.564 So I think when you're thinking about how to implement technology in a classroom, it's gonna be completely based on what is the culture that's already there and how does the technology support the culture. 00:26:24.554 --> 00:26:26.624 I promise I did not pay her to say that. 00:26:28.874 --> 00:26:31.424 I'm just saying that Yeah. 00:26:31.664 --> 00:26:35.534 For, for anybody to say like, this is the answer to. 00:26:36.104 --> 00:26:40.214 Schools, problems with technology and say they have that solution, they just don't. 00:26:40.244 --> 00:26:42.764 It has to be focused on what your culture is. 00:26:43.214 --> 00:27:04.364 And I, I don't have my book right here, but, but that is literally what I talk about in both the books that I've written, how to be a Transformative Principal in school X is you have to design your school for the people that are there, not for who you hope would be there, not for who was there 20 years ago, but who is actually there right now, which means that it has to be a continuous, ongoing. 00:27:04.664 --> 00:27:08.654 Design process for you to make it work for those people. 00:27:08.744 --> 00:27:17.024 And it's going to change because that is the nature of schools and, and it has to, so, so, I, I like that you said that, but go deeper. 00:27:17.024 --> 00:27:20.894 What do you, what are the things that you have to be looking for and paying attention? 00:27:20.894 --> 00:27:24.344 Maybe give a few examples of, of what that looks like in a school. 00:27:24.704 --> 00:27:25.454 So. 00:27:26.324 --> 00:27:31.214 There's already a predisposition in society to use technology as a babysitter. 00:27:31.244 --> 00:27:39.074 So recognizing what your learners are coming in with and what their experience is with technology versus what you can actually support. 00:27:39.104 --> 00:27:46.424 So for example, I moved at the Seattle area in 1994 to do my one year doctoral internship at a school for children with learning differences. 00:27:46.634 --> 00:27:52.724 And they had a big part of the way that they did their generative instructive approach included drill and practice. 00:27:52.724 --> 00:27:53.054 So. 00:27:53.164 --> 00:28:07.894 For me, seeing all these reams of paper being wasted hurt my heart from, you know, just from a conservationist perspective, but also knowing that, well, I thought it was cool that they did peer learning and they, they were coaching each other and there's a substantial amount of their days where they function as peer coach. 00:28:07.894 --> 00:28:08.434 Love that. 00:28:08.434 --> 00:28:09.364 For a lot of reasons. 00:28:10.204 --> 00:28:11.584 I didn't love that. 00:28:11.584 --> 00:28:14.404 They could have had that relationship done more effectively with. 00:28:15.374 --> 00:28:20.114 A computer and using technology even in 1994, we could have, that's was entirely possible. 00:28:20.144 --> 00:28:25.394 And then use the time, uh, where kids can be together more pro socially. 00:28:26.174 --> 00:28:29.924 'cause that was the other reason why some of the children were in there with, with learning differences. 00:28:30.104 --> 00:28:31.724 It wasn't just that they, it was. 00:28:32.924 --> 00:28:35.684 It's sort of a chicken and egg for me because I, I worked with the older kids. 00:28:35.684 --> 00:28:36.824 It's like, what happened first? 00:28:36.824 --> 00:28:38.684 Did they have the, the learning differences first? 00:28:38.684 --> 00:28:41.774 And that led to social issues or vice versa, right? 00:28:41.924 --> 00:28:45.464 And so, um, so that's, that was a lot of what I saw. 00:28:45.464 --> 00:28:51.914 And while I worked there for five years after I finished my, um, my internship, that was my contention all the time. 00:28:51.914 --> 00:28:56.234 And plus I wanted to do more complex things, uh, with the actual component composite. 00:28:56.254 --> 00:29:07.774 Building that you can do with generative instruction and how that really can make these huge cre uh, huge leaps so that you can do some really fun work in person, together with humans. 00:29:07.834 --> 00:29:08.104 Right? 00:29:08.104 --> 00:29:18.064 So problem solving group, like, so much of what I loved about Morningside, I really loved, but the stuff that sat in my craw, I just designed a better version of, and that was the first company that I sold to a German company. 00:29:18.244 --> 00:29:18.664 So, 00:29:18.779 --> 00:29:19.069 Yeah. 00:29:19.174 --> 00:29:19.354 yeah. 00:29:19.354 --> 00:29:21.364 So I think it's really, but learning your co, like. 00:29:22.259 --> 00:29:26.579 I love that you said that, that, um, it, your culture is not a static thing. 00:29:26.579 --> 00:29:27.659 It's not this one and done. 00:29:27.659 --> 00:29:28.589 You design it, you're done. 00:29:28.769 --> 00:29:29.639 No, of course not. 00:29:29.909 --> 00:29:39.059 You're gonna see different things happening based on the demographics that come into your school, but then also world events and just you as an individual, human evolving as well. 00:29:39.059 --> 00:29:46.949 Like when teachers don't consider their own impact, like what they're bringing in and reflecting back, like that student in front of you is actually a reflection of you. 00:29:47.154 --> 00:29:52.884 And how much are you actually putting on that student instead of realizing, no, that's you putting it on them. 00:29:53.214 --> 00:29:53.514 You know? 00:29:53.844 --> 00:29:56.364 So being able to, to pivot as well. 00:29:56.634 --> 00:30:00.324 But I think really it's just, it's being really, and teachers know this stuff. 00:30:00.534 --> 00:30:08.934 I mean, a lot of times, I guess sometimes folks get jaded, but I have one amazing client who is like 28-year-old or 28 year veteran in the classroom. 00:30:09.204 --> 00:30:10.944 Uh, grades two through four. 00:30:11.334 --> 00:30:14.004 And she has written the most amazing. 00:30:14.254 --> 00:30:18.094 Technology, the most amazing curriculum that I just adore. 00:30:18.094 --> 00:30:24.874 That does, that, illustrates this use beautifully about how you can, and then also capitalizes on lots of other technology that we have available now. 00:30:24.964 --> 00:30:27.094 Like that's the other thing too, is like it's not just. 00:30:27.734 --> 00:30:29.834 What are some past really good uses? 00:30:29.924 --> 00:30:31.934 It's like imagining what else can come. 00:30:31.934 --> 00:30:40.304 Like I love also using technology to, uh, as a springboard, like getting kids interested in STEM from things they're already interested in. 00:30:40.304 --> 00:30:43.994 Like there's this, um, uh, you know, stem to dance group, right? 00:30:43.994 --> 00:30:47.174 Which is focusing on dance, which a lot of children love. 00:30:47.714 --> 00:30:48.614 Organically. 00:30:48.794 --> 00:30:51.524 And then what is the technology pieces that you can attach to that? 00:30:51.524 --> 00:30:55.094 So that's really more of like this doy ask progressive education perspective, right? 00:30:55.124 --> 00:30:56.744 Rather than the standardized curriculum. 00:30:57.014 --> 00:31:01.124 It's like what are the things that really grab the children's interest, interests? 00:31:01.124 --> 00:31:02.384 What do they love to do? 00:31:02.444 --> 00:31:07.424 And then how can we teach them those curriculum standards based on their interests, their unique passions? 00:31:07.964 --> 00:31:09.644 That's individualized instruction. 00:31:09.854 --> 00:31:09.884 Uh. 00:31:10.429 --> 00:31:14.479 Prescriptive diagnostic that you can absolutely do with computers too. 00:31:14.479 --> 00:31:18.889 So I mean, the potential to personalize instruction is amazing, but 00:31:19.124 --> 00:31:19.514 Yeah. 00:31:19.729 --> 00:31:25.219 lot more you have to under really understand all the different layers of learning, including that emotional layer to do that well. 00:31:25.519 --> 00:31:25.849 So, 00:31:25.979 --> 00:31:26.369 Yeah. 00:31:26.579 --> 00:31:26.879 Yep. 00:31:26.939 --> 00:31:28.529 And I didn't pay her to say that either. 00:31:28.559 --> 00:31:34.169 I mean, you're the reason why when we were talking last week and you started hinting at some of these things that you're now 00:31:34.474 --> 00:31:34.894 mm-hmm. 00:31:35.339 --> 00:31:37.799 I was like, yes, we are kindred spirits. 00:31:37.829 --> 00:31:38.999 We totally get this. 00:31:38.999 --> 00:31:44.399 And, and, and those are things that people who have listened to the show have heard me say a hundred times. 00:31:44.399 --> 00:31:45.929 And, and that's where. 00:31:46.409 --> 00:31:54.239 We should take advantage of the technology to, to give ourselves and students more opportunities to be human. 00:31:54.614 --> 00:31:59.714 With each other and, and be, and be the thing that the technology can't be. 00:31:59.834 --> 00:32:00.254 And 00:32:00.384 --> 00:32:00.804 Mm-hmm. 00:32:00.824 --> 00:32:02.444 that the technology can do better than us. 00:32:02.444 --> 00:32:04.394 We should have it do as soon as possible. 00:32:04.664 --> 00:32:17.624 But there are so many things that the technology will never be able to do because it is not human and, and there's not value there, but that we can do and that we should pursue and continue doing because it is. 00:32:18.039 --> 00:32:21.999 It's, it's what makes us unique and what makes us human and, and not machine. 00:32:21.999 --> 00:32:26.229 So, uh, this, this was awesome and I, I would love to interview that. 00:32:26.229 --> 00:32:26.259 Uh. 00:32:27.164 --> 00:32:32.054 That teacher you were talking about, uh, that created that, uh, curriculum for kids. 00:32:32.054 --> 00:32:33.704 That would be, that would be interesting. 00:32:33.704 --> 00:32:35.174 I'll talk to you more about that afterward. 00:32:35.414 --> 00:32:35.714 Yeah. 00:32:36.374 --> 00:32:37.904 um, Linda, this was awesome. 00:32:37.904 --> 00:32:51.224 I wanna let make sure people know they can check, uh, your stuff out at Linda b learning, uh, dot com, linda b learning.com for Linda Burrick, and, uh, any parting words or anything you'd like to say before we sign off? 00:32:51.959 --> 00:32:59.639 I. Just wanna encourage teachers, especially those who are a little more maybe reticent when it comes to technology. 00:32:59.819 --> 00:33:05.729 Don't be afraid of it, and let the kids lead too, because that can take you to some really, really interesting direction. 00:33:05.729 --> 00:33:13.704 So rather than thinking about fighting against the technology and that it's something that we have to overcome, or like the whole cheating aspect, nah, lean into it. 00:33:13.704 --> 00:33:15.749 Get a little more creative of what we can do instead. 00:33:16.499 --> 00:33:16.979 Yes. 00:33:17.039 --> 00:33:18.389 Oh, preach. 00:33:18.869 --> 00:33:22.079 Alright, Linda, this has been awesome having you on Transformative principle. 00:33:22.079 --> 00:33:23.009 Thank you so much. 00:33:23.039 --> 00:33:27.959 Again, Linda b learning.com for her newsletter and it's been great chatting with you. 00:33:27.959 --> 00:33:28.739 Thanks so much for your 00:33:28.889 --> 00:33:29.279 You bet. 00:33:29.279 --> 00:33:29.699 Thanks.