WEBVTT - How AI Assistants Can Transform Education 0:00:02.160 --> 0:00:04.760 In the world of educational research, there is a famous 0:00:04.920 --> 0:00:08.280 video of a boy named Sean. I don't mean famous 0:00:08.320 --> 0:00:10.440 in a sense that it has a million views on YouTube. 0:00:10.840 --> 0:00:12.840 I mean it in the circle of people who think 0:00:12.880 --> 0:00:16.400 about teaching and how to make teaching better. The video 0:00:16.440 --> 0:00:19.960 has been written about in journal articles and shown over 0:00:20.000 --> 0:00:23.560 and again in college classrooms. It's a ten minute clip 0:00:24.000 --> 0:00:27.240 of a third grade class somewhere in Michigan. It was 0:00:27.280 --> 0:00:30.040 filmed in January of nineteen ninety, so the video is 0:00:30.080 --> 0:00:33.720 a bit grainy. The teacher's name is Deborah Lowenburg Ball. 0:00:34.080 --> 0:00:37.080 She's a professor at Michigan State University who is part 0:00:37.120 --> 0:00:39.839 of her research teaches a one hour math class at 0:00:39.880 --> 0:00:43.400 a local elementary school on the day in question. Miss 0:00:43.479 --> 0:00:47.160 Ball begins by asking her students about the previous day's lesson, 0:00:47.560 --> 0:00:50.320 which was about even and odd numbers. 0:00:50.520 --> 0:00:52.040 And I would like to hear from as many people 0:00:52.080 --> 0:00:55.040 as possible what comments you had, reactions you had to 0:00:55.080 --> 0:00:56.920 be in that meeting yesterday. 0:00:57.200 --> 0:01:00.120 A little boy with black hair raises his hand. His 0:01:00.200 --> 0:01:00.760 name is Sean. 0:01:01.520 --> 0:01:04.840 I know I don't have anything about the meeting yesterday, 0:01:04.920 --> 0:01:06.560 but I was just thinking about. 0:01:06.400 --> 0:01:09.200 Sit Sean was thinking about the number six. 0:01:10.440 --> 0:01:12.600 I was thinking that it's a it's an ID. It 0:01:12.640 --> 0:01:17.959 can be an odd number two because there could be two, two, four, six, two, 0:01:18.400 --> 0:01:25.680 three tudents and two threes, it would be an add anthonisina. 0:01:28.800 --> 0:01:29.280 She thinks. 0:01:29.280 --> 0:01:32.040 To make it takes me two things. 0:01:32.080 --> 0:01:35.920 Sean doesn't understand what odd and even means. He thinks 0:01:35.959 --> 0:01:38.080 that just because you can break down six in an 0:01:38.160 --> 0:01:41.160 odd number of parts and an even number of parts, 0:01:41.640 --> 0:01:45.760 that six must exist in some magical middle category. And 0:01:45.800 --> 0:01:48.520 when you listen to the Sean videotape, you keep waiting 0:01:48.560 --> 0:01:51.800 for the teacher to say, oh, no, Sean, you misunderstand. 0:01:52.440 --> 0:01:55.400 But Deborah Ball doesn't do that. She never tells him 0:01:55.400 --> 0:02:00.559 he's wrong. Instead, she simply asks him to explain his thinking. 0:02:00.600 --> 0:02:02.520 Me and the two things that you put together to 0:02:02.560 --> 0:02:05.040 make it were odd, right, three and three of each 0:02:05.080 --> 0:02:08.120 child any think proba. 0:02:09.360 --> 0:02:13.080 Two or easy bald and asked the class to give 0:02:13.120 --> 0:02:16.440 their views. Other students jump up and explain their theories 0:02:16.480 --> 0:02:19.680 on the blackboard. For the next fifteen minutes, she definitely 0:02:19.680 --> 0:02:22.640 guides the class through an in depth investigation of what 0:02:22.720 --> 0:02:27.200 she calls Shawn numbers, until Sewn himself realizes that the 0:02:27.240 --> 0:02:29.880 real meaning of odd and even is something different than 0:02:29.880 --> 0:02:32.880 he had imagined, and now he gets it. 0:02:33.320 --> 0:02:35.959 I've been thank you for waring in love. 0:02:36.880 --> 0:02:39.400 I don't want to focus just on how little Sean 0:02:39.720 --> 0:02:42.239 finally made his own way to the right answer. I'm 0:02:42.240 --> 0:02:44.560 interested in what his teacher did to get him there. 0:02:45.320 --> 0:02:49.280 Deborah Ball worked magic. She never told Sean the right answer, 0:02:49.919 --> 0:02:51.960 She just led him to a place where he could 0:02:52.000 --> 0:02:56.880 discover it for himself. My name is Malcolm Glawo. This 0:02:57.000 --> 0:02:59.760 is season six of Smart Talks with IBM, where we 0:02:59.800 --> 0:03:02.360 are for our listeners a glimpse behind the curtain of 0:03:02.400 --> 0:03:07.240 the world of technology and artificial intelligence. In this season, 0:03:07.560 --> 0:03:10.440 we're going to visit companies as varied as Lail and 0:03:10.520 --> 0:03:14.639 Ferrari and tell stories of how they're using artificial intelligence 0:03:15.000 --> 0:03:18.799 and data to transform the way they do business. This 0:03:18.919 --> 0:03:21.960 episode is about the promise of a radical new idea 0:03:22.040 --> 0:03:25.560 called responsive teaching, the kind of teaching that took place 0:03:25.600 --> 0:03:29.799 that day in Shawn's classroom, and whether artificial intelligence can 0:03:29.840 --> 0:03:32.760 help us train the next generation of teachers to be 0:03:32.800 --> 0:03:41.080 as good as Deborah Ball. Before we talk about how 0:03:41.120 --> 0:03:44.080 AI could transform the way we train teachers, I want 0:03:44.120 --> 0:03:46.200 to go back for a moment to the famous video 0:03:46.200 --> 0:03:50.600 of Sean. In the video, the teacher Deborah Ball doesn't 0:03:50.600 --> 0:03:53.880 have a predetermined plan that she's imposing on the class. 0:03:54.320 --> 0:03:58.360 She's improvising, making up her approach as she goes along, 0:03:58.600 --> 0:04:03.000 responding to her students odd theory about the number six. Second, 0:04:03.520 --> 0:04:08.720 she's taking Sean seriously. She's not dismissing his theory. She's 0:04:08.880 --> 0:04:11.720 listening to him and trying to understand the problem from 0:04:11.720 --> 0:04:16.599 his perspective. And Thirdly, and most importantly, she's not force 0:04:16.680 --> 0:04:20.679 feeding him the right answer. She's being patient. She's waiting 0:04:20.720 --> 0:04:23.800 to see if with just the right subtle hints, he 0:04:23.800 --> 0:04:27.600 can get to the right answer on his own. Improvisation, 0:04:28.240 --> 0:04:32.240 empathy patients. That's responsive teaching. 0:04:32.560 --> 0:04:35.000 What I think about in terms of responsiveness is more 0:04:35.120 --> 0:04:38.719 like I think that students need to have a sense 0:04:38.880 --> 0:04:43.680 of agency in what happens in the classroom, and like 0:04:43.720 --> 0:04:49.520 authentic agency where they can be legitimized as knowers. 0:04:50.240 --> 0:04:53.640 I spoke to a physicist at Seattle Pacific University named 0:04:53.680 --> 0:04:58.080 Amy Robertson, a longtime advocate for responsive teaching. She uses 0:04:58.120 --> 0:04:59.719 the Sean video in her classroom. 0:05:00.080 --> 0:05:02.200 You have to trust that kids have a way of 0:05:02.240 --> 0:05:05.440 doing that, and that like heard. What she mostly did 0:05:05.520 --> 0:05:08.080 was to facilitate a conversation and to say you have 0:05:08.160 --> 0:05:09.240 to listen to them talk. 0:05:09.680 --> 0:05:12.200 No one told him he was wrong, that's right. And 0:05:12.240 --> 0:05:15.279 then he goes, He goes, I didn't think of it 0:05:15.320 --> 0:05:20.040 that way. Again, I thank you for leaving alone. You've 0:05:20.120 --> 0:05:23.599 expanded my understanding. Thank you for bringing it up again. 0:05:23.680 --> 0:05:26.240 It's like this, I love. 0:05:27.120 --> 0:05:30.159 I know responsive teaching, as I think about it is 0:05:30.240 --> 0:05:34.080 kind of rooted in this, like Eleanor Duckworth's work around 0:05:34.160 --> 0:05:36.800 the Having of Wonderful Ideas, where she says, like the 0:05:36.839 --> 0:05:40.760 goal of education is for students to have wonderful ideas 0:05:40.760 --> 0:05:42.240 and to have a good time having them. 0:05:42.839 --> 0:05:45.799 I love that. I've never heard that. What a beautiful, 0:05:46.360 --> 0:05:49.880 succinct way of summing up the purpose of education. Yes, 0:05:50.760 --> 0:05:54.480 responsive teaching is beautiful. It's rare to find a new 0:05:54.520 --> 0:05:58.200 teaching idea that everyone loves. This is one of those 0:05:58.279 --> 0:06:01.880 rare ideas. Watching the Ball classroom, all I could think was, 0:06:02.120 --> 0:06:04.719 I really really hope my daughters get to experience a 0:06:04.760 --> 0:06:08.359 math class like that. Far too many kids are convincing 0:06:08.400 --> 0:06:11.599 themselves at far too young an age that math isn't 0:06:11.600 --> 0:06:15.159 for them, and responsive teaching is a way to solve 0:06:15.160 --> 0:06:19.960 that problem. But here is the issue. It's really really 0:06:20.000 --> 0:06:23.920 hard to teach responsive teaching. Robertson says that teaching exists 0:06:24.000 --> 0:06:27.080 in a cultural environment where the teacher is expected to 0:06:27.080 --> 0:06:29.880 be the source of truth. That teaching is about the 0:06:30.040 --> 0:06:33.480 immediate correction of error and not letting a child wander 0:06:33.560 --> 0:06:38.599 down the pathway of their own Misunderstanding. Responsive teaching is 0:06:38.839 --> 0:06:42.479 deeply counterintuitive, and the only way to understand its beauty 0:06:42.839 --> 0:06:47.400 is to do it over and over again. Aspiring teachers 0:06:47.800 --> 0:06:51.599 need a way to practice. For as long as there 0:06:51.640 --> 0:06:55.080 has been technology, people have turned to digital machines to 0:06:55.120 --> 0:06:58.800 solve problems. My father was a mathematician and I remember 0:06:58.839 --> 0:07:01.320 him coming home in the nineteen seventies with a big 0:07:01.360 --> 0:07:04.479 stack of computer cards in his briefcase that he used 0:07:04.480 --> 0:07:07.560 to program the main frame back of the office. Today, 0:07:07.880 --> 0:07:12.040 with the rise of artificial intelligence, the scale and complexity 0:07:12.240 --> 0:07:15.480 of the problems technology can help us solve has jumped 0:07:15.480 --> 0:07:18.640 by many orders of magnitude. You must have worked with 0:07:18.720 --> 0:07:22.800 a million customers who are experimenting with llm's. Has there 0:07:22.880 --> 0:07:25.720 been one use case that you were like, WHOA, I 0:07:25.720 --> 0:07:29.920 had no idea or just simply that's clever. I'm speaking 0:07:29.960 --> 0:07:33.280 to Brian Bissell, who works out of IBM's Manhattan office. 0:07:33.640 --> 0:07:36.840 He helps IBM customers discover how best to get AI 0:07:36.960 --> 0:07:37.800 to work for them. 0:07:38.560 --> 0:07:40.200 There is one, but I don't think I can talk 0:07:40.240 --> 0:07:41.120 about it unfortunately. 0:07:41.400 --> 0:07:45.080 Wait, you can't tease me like that, can you wait? 0:07:45.840 --> 0:07:47.600 Disguise disguise it for me. 0:07:47.720 --> 0:07:50.120 Just give me a general It was about the ability 0:07:50.120 --> 0:07:53.360 to pull certain types of information out of documents that 0:07:53.400 --> 0:07:56.680 you wouldn't think you would be able to get the 0:07:56.840 --> 0:07:59.720 model to do, and be able to do that at 0:07:59.800 --> 0:08:01.240 a very large scale. 0:08:01.560 --> 0:08:04.200 Bissil's point was that we are well past the stage 0:08:04.200 --> 0:08:07.760 where anyone wonders whether AI can be useful. The real 0:08:07.840 --> 0:08:10.880 question now is what problems do we want to use 0:08:10.920 --> 0:08:13.480 it to solve? Where it can make the biggest difference, 0:08:14.160 --> 0:08:17.560 And Bissil saw lots of opportunities in education. 0:08:18.480 --> 0:08:21.600 I have two kids, one in middle school and one 0:08:21.600 --> 0:08:24.880 who just graduated high school, and I'm well aware of 0:08:24.960 --> 0:08:28.600 students using things like chat GPT to do their homework. 0:08:28.680 --> 0:08:34.800 And it's very easy to take tools like that and 0:08:34.840 --> 0:08:38.560 even IBM's own large language models, and just take a problem, 0:08:38.880 --> 0:08:42.640 a piece of homework, something you want written, and drop 0:08:42.679 --> 0:08:44.560 it into that and have it generate the answer for 0:08:44.640 --> 0:08:48.959 you and the student. The user in that case hasn't 0:08:49.000 --> 0:08:51.480 done any work, they haven't put any real thought into it. 0:08:52.040 --> 0:08:55.959 To Bissil, that's the wrong use of AI. That's technology 0:08:56.200 --> 0:09:00.079 making is dumber. What we really want is technology that 0:09:00.160 --> 0:09:02.880 makes us smarter. This will explain to me that there 0:09:02.880 --> 0:09:05.760 are now two big tools being used for AI productivity, 0:09:06.240 --> 0:09:11.719 AI agents and AI assistants. Let's start with AI agents. 0:09:11.800 --> 0:09:15.600 AI agents can reason plan and collaborate with other AI 0:09:15.679 --> 0:09:20.160 tools to autonomously perform tasks for a user. This will 0:09:20.200 --> 0:09:22.800 gave me an example of how college freshmen might use 0:09:22.800 --> 0:09:23.520 an AI agent. 0:09:23.960 --> 0:09:26.440 As a new student, you may not know how do 0:09:26.480 --> 0:09:28.480 I do with my health and wellness issue? Some of 0:09:28.520 --> 0:09:30.360 your credits are going to get for this given class. 0:09:31.080 --> 0:09:34.600 You could talk to someone and find out some of that, 0:09:35.600 --> 0:09:37.160 but maybe it's a little bit sensitive and you don't 0:09:37.200 --> 0:09:37.719 want to do that. 0:09:38.240 --> 0:09:41.360 Bisill told me you could build an AI agent, a 0:09:41.440 --> 0:09:44.960 resource for new students that helps them navigate a new campus, 0:09:45.200 --> 0:09:48.440 register for classes, access the services they need, and even 0:09:48.480 --> 0:09:52.280 schedule appointments on their behalf, which in turn buys them 0:09:52.320 --> 0:09:54.640 more time to focus on their actual schoolwork. 0:09:54.920 --> 0:09:59.800 We can see patterns of how agents and assistants can 0:09:59.840 --> 0:10:04.319 help help employees and customers and end users be more productive, 0:10:04.520 --> 0:10:08.400 automate workflows so they're not doing certain types of repetitive 0:10:08.480 --> 0:10:12.680 work over and over again, and streamlining their lives and 0:10:12.720 --> 0:10:16.640 making data more accessible to them twenty four hours a day. 0:10:17.000 --> 0:10:20.720 But Bissel says you can also use AI assistance in 0:10:20.760 --> 0:10:24.839 the education space. AI assistants are reactive as opposed to 0:10:24.920 --> 0:10:29.160 AI agents, which are proactive. AI assistants only perform tasks 0:10:29.360 --> 0:10:34.240 at your request. They're programmed to answer your questions, and 0:10:34.280 --> 0:10:37.400 as it turns out, AI assistants are now being used 0:10:37.720 --> 0:10:41.480 to further the responsive teaching revolution, which is why I 0:10:41.559 --> 0:10:44.520 found myself on a beautiful Georgia spring day not long 0:10:44.559 --> 0:10:48.720 ago on the campus of Kansas State University, sitting in 0:10:48.720 --> 0:10:52.960 the classroom with two researchers, one of them Professor Dabe Lee. 0:10:53.160 --> 0:10:55.640 Let's go into the journey of building this thing. You 0:10:56.200 --> 0:10:58.840 started by taking a course. What was the course you took? 0:11:00.040 --> 0:11:05.280 So it was offered by Coursera, It was designed by IBM. 0:11:05.480 --> 0:11:08.800 It was AI Foundation for everyone. 0:11:12.080 --> 0:11:14.760 In her AI Foundation's course, Lee learned how to build 0:11:14.760 --> 0:11:18.480 an AI assistant using IBM watson X that course took 0:11:18.480 --> 0:11:19.719 how long to take. 0:11:20.280 --> 0:11:22.679 It was not to know it was like fourteen mix. 0:11:23.960 --> 0:11:27.400 Lee's idea was to train an AI assistant on classroom 0:11:27.480 --> 0:11:30.840 data to play the role of Sean, a digital persona 0:11:31.080 --> 0:11:33.559 of a nine year old who likes math but doesn't 0:11:33.559 --> 0:11:37.600 always understand math, and that AI assistant, she thought could 0:11:37.640 --> 0:11:40.760 be used to train preservice teachers or teachers in training 0:11:41.000 --> 0:11:43.600 who are preparing to enter one of the most challenging 0:11:43.640 --> 0:11:45.400 professions in the modern world. 0:11:46.320 --> 0:11:49.319 So when you think about the teacher education and a 0:11:49.520 --> 0:11:55.160 major challenge that teacher education face is that we need 0:11:55.400 --> 0:12:00.360 children to practice with. We need instructors who if the 0:12:01.240 --> 0:12:05.480 instruction on the pedagogical skills. So when you look at 0:12:05.480 --> 0:12:09.559 the teacher education program, we have coursework in field experience, 0:12:10.080 --> 0:12:14.480 and in those two areas there is something missing all 0:12:14.520 --> 0:12:14.959 the time. 0:12:16.120 --> 0:12:19.520 Li says that pre service teachers often lack access to 0:12:19.559 --> 0:12:23.040 both students and experienced teachers during their education. 0:12:23.760 --> 0:12:28.160 So what we try to resolve is that we have 0:12:28.280 --> 0:12:31.480 this virtual student for pre service teacher to work with 0:12:32.280 --> 0:12:36.520 so that they can practice their responsive teaching skills. 0:12:36.720 --> 0:12:40.439 The first AI assistant Lee created is g Wu. G 0:12:40.600 --> 0:12:43.440 Wu emulates the persona of a nine year old third 0:12:43.440 --> 0:12:46.960 grade girl. Then, with the help of one of her collaborators, 0:12:47.040 --> 0:12:50.720 a researcher at Canazon named Sean English, she created two 0:12:50.760 --> 0:12:55.240 more AI assistants, Gabriel and Noah, each of which have 0:12:55.360 --> 0:13:00.920 their own distinctive characteristics. So how are Gabriel and Noah different. 0:13:00.600 --> 0:13:05.839 From ji Wu gab Real my first one? He is 0:13:06.000 --> 0:13:09.120 very short answered. If you ask an open ended question, 0:13:09.320 --> 0:13:13.120 he will answer it in a close way. So I 0:13:13.360 --> 0:13:17.120 use that characteristic. And that's the problem that most teachers 0:13:17.320 --> 0:13:22.240 actually base. They're asked children who are shay, who are reserved, 0:13:22.480 --> 0:13:26.480 and who would not sure much of their thoughts. So 0:13:26.760 --> 0:13:30.439 we wanted that characteristic in some characters, and we use 0:13:30.559 --> 0:13:32.760 gabrielle to have that characteristic. 0:13:34.200 --> 0:13:37.240 And Noah. What'snawah's personality? 0:13:38.040 --> 0:13:41.760 How do he playful, cheering, right and energetic? 0:13:42.600 --> 0:13:45.280 That's Sean English, Professor Lee's fellow researcher. 0:13:45.840 --> 0:13:52.000 And Jiwu Ju is articulated and kind of smart, but 0:13:52.559 --> 0:13:54.560 she has her own way of thinking. 0:13:54.960 --> 0:13:56.880 I would end up spending a lot of time with 0:13:57.000 --> 0:14:00.680 ji Wu. She's something of a character. What Sean about 0:14:00.679 --> 0:14:04.199 the process of creating these AI assistants? What does building 0:14:04.280 --> 0:14:09.880 the content side of the AI assistant entail Sean what it. 0:14:09.800 --> 0:14:14.000 Sets up a series of actions, effectively, which are response cases. 0:14:14.120 --> 0:14:16.520 You can kind of think of them as you have 0:14:16.559 --> 0:14:20.240 a series of questions that you tie to an intent, 0:14:20.640 --> 0:14:23.920 and then that intent has reactions from the bot, and 0:14:23.960 --> 0:14:27.040 so effectively, if we were looking to say, make a 0:14:27.040 --> 0:14:29.320 hello action, we would have all the different ways that 0:14:29.360 --> 0:14:31.600 people could say Hello, Hello, what's up, how you doing, 0:14:31.640 --> 0:14:32.640 and all that kind of stuff. 0:14:32.960 --> 0:14:36.840 Sean says, the longer the list of potential responses, the better. 0:14:37.360 --> 0:14:41.240 But AI's responses don't just follow the list. The AI 0:14:41.280 --> 0:14:44.840 assistant uses those suggested responses to come up with a 0:14:44.960 --> 0:14:48.840 universe of other responses, and in that process sometimes it 0:14:48.920 --> 0:14:51.200 comes up with things that just don't make sense. 0:14:51.320 --> 0:14:54.760 And from a technological standpoint, while AI is a fantastic tool, 0:14:54.920 --> 0:14:57.360 AI can hallucinate, which I mean just give things that 0:14:57.360 --> 0:15:00.440 it's just straight up made up. There's a famous example 0:15:00.480 --> 0:15:02.360 of this called the three rs is where you ask 0:15:02.440 --> 0:15:05.359 a popular large language model how many RS are in strawberry, 0:15:05.640 --> 0:15:07.720 and it gives you the wrong answer and he repeats 0:15:07.720 --> 0:15:10.520 that result repetitively. You always want to have a human 0:15:11.080 --> 0:15:13.040 interacting with the system to be able to go Hey, 0:15:13.680 --> 0:15:15.840 that's a little crazy. I don't think that's exactly what 0:15:15.840 --> 0:15:16.640 we're going for here. 0:15:17.280 --> 0:15:19.280 That's why it's good to have someone like Sean English 0:15:19.320 --> 0:15:21.400 around to step in and get the model back on track, 0:15:21.800 --> 0:15:24.920 and over time, when the model has enough training, it's 0:15:25.000 --> 0:15:31.360 ready for the teachers in training. One of the rollouts 0:15:31.360 --> 0:15:34.480 of Jiwu, Gabriel and Noah was with the teacher training 0:15:34.520 --> 0:15:36.479 program at the University of Missouri. 0:15:36.800 --> 0:15:39.160 I was just kind of excited to see what the 0:15:39.240 --> 0:15:41.600 program was and what it was going to be doing. 0:15:42.080 --> 0:15:45.160 This is Logan Hovis, a junior at Missouri on the 0:15:45.200 --> 0:15:47.480 path to becoming an elementary school teacher. 0:15:47.880 --> 0:15:50.400 Obviously a little skeptical when he said it was sosed to, 0:15:50.440 --> 0:15:53.360 you know, be like talking to a student. You're like, 0:15:53.400 --> 0:15:55.800 there's no way this AI thing is going to totally 0:15:55.840 --> 0:15:58.480 sound like a second grader or a third grader, Like 0:15:58.560 --> 0:16:01.080 it's going to sound like an adult, or it's going 0:16:01.120 --> 0:16:02.960 to sound like a robot that knows all the answers. 0:16:03.480 --> 0:16:06.200 And it really didn't. It really was like talking to 0:16:06.280 --> 0:16:08.920 a child. It was very very well developed in the 0:16:08.960 --> 0:16:10.800 way that you really sit there and you feel like 0:16:10.840 --> 0:16:11.840 you're talking to a kid. 0:16:12.440 --> 0:16:15.640 Her point wasn't that Jiwu and her fellow avatars were 0:16:15.720 --> 0:16:19.440 equivalent to real kids, of course not, but for someone 0:16:19.560 --> 0:16:22.480 starting out, someone who was already nervous about being plunged 0:16:22.520 --> 0:16:25.200 into a classroom of nine year olds, Geewu was like 0:16:25.240 --> 0:16:27.160 a warm up before a baseball game. 0:16:27.520 --> 0:16:29.640 What I can think of is like, you know how, 0:16:30.040 --> 0:16:32.360 when you're at batting practice for baseball or softball, you 0:16:32.400 --> 0:16:35.360 have those automatic pitchers that throw them because you're working 0:16:35.360 --> 0:16:38.000 on your skill as the hitter. What can I do differently? 0:16:38.040 --> 0:16:41.040 What am I doing wrong? But that doesn't replace the 0:16:41.080 --> 0:16:42.800 game and what you do in a game. But this 0:16:42.960 --> 0:16:45.680 is you getting to practice your own skills to be 0:16:45.720 --> 0:16:47.240 better when you go in a game. And I think 0:16:47.240 --> 0:16:50.280 that's kind of what the AI software feels like for us. 0:16:52.200 --> 0:16:54.840 In batting practice, the pitches don't come as hard and 0:16:54.880 --> 0:16:57.440 fast as the pitch is in a real game, but 0:16:57.480 --> 0:16:59.360 you get to stand at the plate and the picture 0:16:59.360 --> 0:17:03.040 throws you dozens of balls over and over again in 0:17:03.080 --> 0:17:05.880 a concentrated block that allows you to work on your 0:17:05.880 --> 0:17:08.440 swing closely and carefully. 0:17:09.080 --> 0:17:11.520 There's a lot less stimulus going on around because the 0:17:11.520 --> 0:17:14.280 classroom is very, very busy. It's wonderful, it's beautiful, but 0:17:14.320 --> 0:17:16.800 it's very, very busy, so sometimes it's hard to keep 0:17:17.359 --> 0:17:20.640 you know, that focus in on the tasks that they're 0:17:20.640 --> 0:17:23.240 doing at hand, and also in the teacher setting, you're 0:17:23.240 --> 0:17:25.879 also kind of always looking around making sure that other 0:17:25.920 --> 0:17:27.960 students are doing what they're supposed to be doing, but 0:17:28.040 --> 0:17:30.200 also like if they need any help, if everything's going 0:17:30.240 --> 0:17:35.440 okay in the classroom. So being on the Jiwu chat, 0:17:36.000 --> 0:17:37.879 it was just nice that you didn't have to do 0:17:38.040 --> 0:17:40.360 any of the extra work to keep the focus on there, 0:17:40.800 --> 0:17:44.320 and it also felt you didn't have to feel the 0:17:44.359 --> 0:17:48.000 student's nervousness of being one on one with you. And 0:17:48.160 --> 0:17:50.160 also as a tea shirt, it was a lot less 0:17:50.200 --> 0:17:53.240 pressure too, because I was like, Okay, I'm taking this serious. 0:17:53.320 --> 0:17:55.080 This is a student I'm questioning. 0:17:55.400 --> 0:17:57.760 But I also know I'm probably not going to hurt 0:17:57.760 --> 0:18:00.879 someone's feelings right now, and that's terrify. I'm going to 0:18:00.880 --> 0:18:04.399 ask the wrong question and have set the child because 0:18:04.400 --> 0:18:05.159 I've done that. 0:18:06.160 --> 0:18:08.639 We think of the typical use of AI as a 0:18:08.680 --> 0:18:11.440 tool for speeding things up. That's what we always hear 0:18:11.680 --> 0:18:14.639 that the introduction of AI to problem X gave an 0:18:14.640 --> 0:18:18.879 answer in minutes. When solving problem X used to take weeks, 0:18:19.440 --> 0:18:23.360 but we shouldn't forget another use that it allows us 0:18:23.680 --> 0:18:27.120 to slow things down. Hovis, if she wanted to, could 0:18:27.119 --> 0:18:30.280 spend a whole weekend practicing with g Wu. A real 0:18:30.400 --> 0:18:33.040 nine year old will get frustrated on board with the 0:18:33.080 --> 0:18:37.080 fumbling novice after ten minutes, but g Wu g Wu 0:18:37.240 --> 0:18:40.040 will happily answer questions for as long as it takes 0:18:40.040 --> 0:18:42.680 for the people who want to learn to be responsive 0:18:43.240 --> 0:18:47.520 to learn how to be responsive. At the end of 0:18:47.520 --> 0:18:51.000 my time at Kenesas State, Sean and Dabe led me 0:18:51.080 --> 0:18:54.080 to a small table where Dabe had set up her laptop. 0:18:54.680 --> 0:18:57.000 In the corner of the screen was a chat box 0:18:57.280 --> 0:18:59.600 of the sort we've all seen and used a thousand times. 0:19:00.480 --> 0:19:03.919 Jiwu began she had been given a math problem. 0:19:06.480 --> 0:19:10.520 Who are of three fourth cup of a flower to 0:19:10.680 --> 0:19:16.040 the ball? Thanks to the added another three six is cup. 0:19:16.680 --> 0:19:22.000 It's a cotal amount of flower. Who use greater or 0:19:22.359 --> 0:19:26.920 dan or a less than war cup? How much flower 0:19:27.040 --> 0:19:27.919 they can loose? 0:19:28.240 --> 0:19:31.359 That's a simulation of Jiwu speaking. We pause her for 0:19:31.359 --> 0:19:36.840 a second. So Jeewu is trying to solve this problem, 0:19:36.880 --> 0:19:39.520 and the first thing she does is she draws a 0:19:39.560 --> 0:19:43.159 rectangle on the screen. This is a common tactic of 0:19:43.240 --> 0:19:47.320 nine year olds. Try to visualize the fractions, and she 0:19:47.400 --> 0:19:52.800 divides it into four pieces. And now she's gonna color 0:19:52.840 --> 0:19:55.480 in three of the four pieces. Yes, so she's representing. 0:19:55.520 --> 0:19:58.520 This is quite good. She's representing three quarters on the screen. 0:20:00.200 --> 0:20:00.480 This is. 0:20:02.600 --> 0:20:03.480 Three six. 0:20:05.760 --> 0:20:11.240 So now, Jiwu, there's another rectangle with six boxes and 0:20:11.359 --> 0:20:12.600 colors in three of them. 0:20:12.920 --> 0:20:19.720 Okay, the together that makes sikes come off. 0:20:21.280 --> 0:20:25.240 So then she counts up all the colored boxes and 0:20:25.520 --> 0:20:28.600 that's her numerator, and counts up the total number of 0:20:28.600 --> 0:20:32.280 boxes and that's her denominator. Ji Wu had counted the 0:20:32.320 --> 0:20:35.919 colored boxes and landed on an answer. When you add 0:20:36.119 --> 0:20:38.800 three quarters of a cup and three sixth of a cup, 0:20:39.400 --> 0:20:42.520 you get six tenths of a cup. So, according to 0:20:42.640 --> 0:20:45.560 ji Wu, Martin has less than one cup. And she 0:20:45.640 --> 0:20:46.800 thinks she solved the problem. 0:20:46.920 --> 0:20:49.600 Yes, okay, so it's less than one cup. 0:20:49.960 --> 0:20:53.200 Yeah, so she says it's less than one cup. Now, 0:20:53.320 --> 0:20:56.240 oh my god, this is hard. So the question is 0:20:56.240 --> 0:20:59.600 what do I, as a teacher say to Jiwu. We 0:20:59.600 --> 0:21:02.679 were off. The rules were simple. I couldn't give ji 0:21:02.760 --> 0:21:05.320 Wu the answer or explain to her what she was 0:21:05.359 --> 0:21:08.240 doing wrong. I had to be Deborah Ble. I had 0:21:08.280 --> 0:21:11.639 to help her find the way herself. The chat box 0:21:11.720 --> 0:21:13.800 in the corner of the screen was waiting for my 0:21:13.840 --> 0:21:17.119 first question. I thought for a moment and started typing, 0:21:17.400 --> 0:21:20.480 do you think the boxes in the red rectangle are 0:21:20.520 --> 0:21:23.560 the same size as the boxes in the blue rectangle? 0:21:24.359 --> 0:21:26.760 Then I turned to Sean and dabey, is that a 0:21:26.760 --> 0:21:27.280 good question? 0:21:27.560 --> 0:21:28.399 Yeah? 0:21:28.480 --> 0:21:29.000 Serious thing. 0:21:30.000 --> 0:21:31.439 Yeah, that's a good question. 0:21:32.000 --> 0:21:36.080 Jewu doesn't mess around. She answers immediately. So Ju says, 0:21:36.119 --> 0:21:38.280 the blue and red pieces are not the same sizes. 0:21:39.080 --> 0:21:43.400 Oh, so you understand now Ju knows that side differences. 0:21:44.600 --> 0:21:48.440 So she's pretty smart here. Yeah. Then I asked, if 0:21:48.480 --> 0:21:51.000 they are not the same size, do you think you 0:21:51.000 --> 0:21:55.720 can add them together? Ji Wu answered right away. Jiwu says, 0:21:56.760 --> 0:21:58.720 I have learned that I could add any numbers in 0:21:58.800 --> 0:22:01.600 grade too, So three p three is six and four 0:22:01.640 --> 0:22:02.440 to six is ten. 0:22:02.760 --> 0:22:07.199 Yeah. So she is using the knowledge of edding intiquers 0:22:07.240 --> 0:22:09.160 into edding fractions. 0:22:09.560 --> 0:22:13.520 Now I'm stumped. So now I have to somehow lead 0:22:13.640 --> 0:22:16.159 her to figure out a way to get her to 0:22:16.280 --> 0:22:20.840 understand that we're dealing with a different kind of problem, 0:22:21.160 --> 0:22:24.359 a harder problem. Amy Robertson had told me that learning 0:22:24.400 --> 0:22:28.159 how to do responsive teaching properly was really hard, and 0:22:28.160 --> 0:22:31.119 now I understood why I had to put my mind 0:22:31.520 --> 0:22:33.600 inside the mind of a nine year old. I had 0:22:33.600 --> 0:22:37.159 to internalize her knowledge base and assumptions, and keep in mind, 0:22:37.440 --> 0:22:40.959 I haven't been nine for a very long time. I 0:22:41.000 --> 0:22:44.200