0:00 This is the second story in our series
0:02 on artificial intelligence being applied here and now,
0:05 where it matters most.
0:07 Last week, we brought you the story
0:09 of the AI used by your doctor.
0:11 This week, it's about your teacher,
0:13 the one you depend on to educate your children,
0:16 the one we all depend on
0:17 to prepare the next generation of workers,
0:19 the one that we all remember from our own childhood.
0:24 -Whenever I talk to people
0:26 about their experiences in education,
0:27 they always name this one teacher
0:29 who was so influential,
0:31 and they got into this discipline
0:33 because this teacher was amazing and inspired them.
0:36 Where students said, "What?"
0:38 Westin: Professor Shamya Karumbaiah
0:39 now teaches those teachers
0:41 as a professor at the University of Wisconsin.
0:44 She lives at the intersection of education and tech,
0:48 focusing on the application of artificial intelligence
0:50 in America's classrooms.
0:53 -We know that teacher in the front of the classroom
0:55 and students in front of their laptops
0:57 is not a model that works. It's broken.
0:58 Last couple of years, especially,
1:00 has been very exciting.
1:02 And I would say the big change
1:03 is the ability for teachers to customize
1:06 what's happening with AI.
1:08 -So why don't you look for on your Kira dashboard...
1:11 Westin: Lance Key is one of those classroom teachers
1:13 adopting and adapting AI
1:16 as he teaches computer science
1:17 in the Putnam County School System
1:19 just east of Nashville, Tennessee.
1:22 -Computer science curriculum, they've really gamified it,
1:24 and they've made it colorful, and engaging,
1:27 and exciting for the students.
1:28 And also, you know, inside of there,
1:30 if they try to do things that they can't,
1:32 it again will redirect them.
1:33 It'll give them, "Oh, there's an error here."
1:35 And then they can go to the chatbot
1:36 and try to work through what the error is.
1:39 And then it'll redirect them to where they need to go.
1:41 -I think the big opportunity
1:43 that stems from AI in the classroom
1:45 is the ability to personalize instruction
1:47 to individual students.
1:48 Westin: Andrea Pasinetti is co-founder and CEO
1:52 of the company Kira.
1:53 It provides the education software platform
1:56 to Lance Key's classroom,
1:57 as well as to many of the largest school districts
2:00 across the United States.
2:02 -With AI, a teacher has an unlimited number
2:06 of teaching assistants to support that process.
2:09 And the teacher can provide guidance,
2:11 guardrails and guidelines
2:12 on how to provide that support to students.
2:14 That looks like students working on a terminal
2:16 and having an AI tutor
2:18 that they can query either in written text,
2:20 by typing on a keyboard, or in spoken word,
2:23 by speaking directly to the terminal on the computer.
2:25 -The AI chatbot.
2:27 Westin: That personalization is something that Key
2:29 is experiencing every day in his Tennessee classroom.
2:33 -It allows me to have a more personalized relationship
2:36 with my students in the classroom.
2:38 Before, there was one of me.
2:40 And for every question that came up,
2:42 I was like bouncing around the room,
2:44 just answering questions all the time.
2:46 But now we've got, you know,
2:47 a tutor that can help them along the way.
2:50 And I can build personalized tutors for them, too.
2:53 So if it's on a specific content area,
2:55 I can say, "OK, today we're working
2:57 "on solving two step equations.
2:59 "Here's a two step tutor
3:00 "that my students can work with alone."2
3:02 And they can ask it questions.
3:03 So then I can walk around,
3:05 and then one on one check with students.
3:07 I've also got a dashboard on the backside
3:08 that will alert me
3:10 if there's a student that's having problems.
3:12 So then I know real time that I need
3:14 to go check on Johnny or Susie.
3:16 Westin: Over and above that personalization,
3:18 AI can also provide teachers like Lance Key
3:21 with some much needed relief.
3:23 My wife previously was an English teacher,
3:26 and I recall grading essays being very daunting for her
3:31 because she would have to score 150 essays every time,
3:36 if they wrote an essay in the classroom.
3:37 So using a rubric,
3:39 being able to upload our district rubrics into Kira,
3:42 and being able to use their...
3:44 One, their grader off of that rubric,
3:47 but also their AI detection, their plagiarism detection,
3:49 and their feedback writer has been amazing.
3:52 So we can load all the papers up into that.
3:54 It will score it all for us.
3:55 It will give us all the feedback.
3:57 And the teachers can then just review it.
3:59 And I think that's the big thing that we can focus on with AI,
4:02 is the time that it gives teachers back in the day.
4:04 By 2030, we're going to need to hire 30 million teachers,
4:08 because we've got teachers retiring
4:09 and people not going into the field.
4:11 So we're going to have a teacher shortage that's coming,
4:12 along with a high burnout rate.
4:15 So I think AI can help us
4:16 with some of the repetitive processes
4:18 that we do over, and over, and over again.
4:20 Westin: For all it's promised today,
4:22 it turns out that the use of AI for education
4:25 isn't all that new.
4:27 While large language models
4:28 are only just making their debut in schools,
4:31 other forms of AI have a long history.
4:34 How long has either artificial intelligence,
4:37 or maybe we should call it "machine learning," going back,
4:39 how long has it been used in the classroom?
4:41 - You'd be surprised how AI and education,
4:45 uses of it in education,
4:46 have a lot of common sort of origins.
4:49 And so I come from an academic background
4:52 where my advisor's advisor's advisor
4:55 back in 1980s was creating
4:59 what we now call as cognitive tutors
5:01 that were being used in Pittsburgh classrooms.
5:04 And people have studied...
5:05 there's this work done in the 1990s on ethnography,
5:10 basically, of how teachers are using AI in the classroom.
5:13 What does introduction of AI in the classroom
5:14 change in the social structures?
5:17 Student-student interaction or student-AI...
5:19 student-teacher interaction?
5:21 Westin: It's only in the last few years
5:23 that AI has developed
5:25 to the point where companies like Kira
5:26 can put it to use in the classroom.
5:29 According to surveys by RAND,
5:31 as of the 2023-24 school year,
5:33 a quarter of all US teachers
5:35 were already using AI to teach students,
5:38 and Pasinetti says the key is the conversation.
5:41 -We use AI as a catch-all term
5:43 for a lot of different technologies.
5:45 In reality, what's happened with the most recent wave of AI,
5:49 with LLMs in particular, is a more discursive medium.
5:53 So AI now, unlike even three, four years ago,
5:57 is able to have real conversations with students.
6:00 It's able to engage with students
6:02 on a level that feels more human in some respect.
6:05 We founded Kira four years ago, before AI was popular.
6:11 AI wasn't cool. It seemed premature.
6:13 And in many cases, it was entirely verboten.
6:16 There were a lot of districts that said,
6:18 "We can't really be talking about AI with parents
6:20 "because it gives them a lot of anxiety,
6:22 "or there's a lack of understanding
6:24 "about what AI can do."
6:25 And so the whole conversation would die at the very beginning.
6:31 So I would say there's been a radical shift
6:34 where that anxiety and resistance
6:37 has given way to curiosity.
6:40 Westin: And some of that curiosity
6:42 has turned into plain necessity,
6:44 as school districts across the country
6:46 struggle to deal with teaching students
6:48 in a multitude of languages.
6:50 -It's important to bring students' home language
6:52 into their classrooms.
6:54 A very promising benefit or opportunity for AI here
6:58 is to bridge the gap
7:00 between the language that the teacher speaks,
7:02 and the different languages
7:03 that the students in the classroom speak.
7:05 Thirty-two states in the United States
7:08 reported that their students speak over 200 languages.
7:12 And these 32 states
7:14 have a deficit of bilingual resource teachers.
7:17 So there are all these opportunities out there
7:19 in terms of what's already working in the classroom,
7:21 but there is limited human resources.
7:24 -The number of districts that can now teach languages
7:27 like Mandarin, or Arabic, or, you know, French and Spanish,
7:32 where historically it would have required
7:34 hiring a native speaker of that language,
7:38 and, as a result,
7:39 been much more difficult to achieve for a district.
7:42 The ability to do that now is immediate.
7:46 Districts can leverage AI tools to teach a new foreign subject,
7:51 or a foreign language as a subject
7:53 in a way that they couldn't have in the past.
7:55 Westin: For all its advantages,
7:57 the widespread use of AI in the classroom
7:59 is not without its risks.
8:01 Risks we've all heard about with other applications of AI.
8:04 Things like hallucinations and bias,
8:07 but also risks that are unique to education.
8:10 A Pew study from last year
8:12 showed that a quarter of teachers think AI tools
8:15 do more harm than good in the classroom.
8:18 Only 6% say AI is beneficial.
8:21 -Any piece of AI that is generating text anywhere
8:24 where it is student facing,
8:26 you have to be careful about the implications
8:29 hallucinations have
8:30 in this specific context of education.
8:32 The goal in the classroom
8:34 is for us to help students learn.
8:36 Hallucination is absolutely detrimental
8:39 to students learning.
8:41 Any sort of misinformation, it could...
8:44 I would also talk about potential biases,
8:47 potential, you know, toxic content,
8:50 the kinds of... any kind of text
8:54 that the AI is generating that is harmful
8:57 for students learning and well-being.
8:59 Westin: But perhaps a greater risk
9:01 with AI applied to the classroom
9:03 is the risk of expecting too much of it.
9:06 -AI is a tool like anything else.
9:08 It has very, very real limitations.
9:11 The fact that it's discursive
9:13 and it resembles an interaction with a human,
9:17 in some isolated cases,
9:19 makes it feel a lot more advanced
9:22 than technologies we've interacted with in the past.
9:24 But as of today,
9:26 AI's boundaries are still limited.
9:28 And it's very easy to overstate. There's a lot of hype.
9:31 It's very easy to feed into a narrative of fear.
9:35 But the reality is AI is still answering
9:38 fairly discreet questions being asked by students,
9:41 and supporting teachers
9:42 with the answering of those questions.
9:44 I think students interacting with computers one on one
9:48 is definitely something that schools, district teachers,
9:53 parents, students themselves need to be careful of.
9:57 It can take students
9:59 out of interpersonal relationships.
10:01 It can dull students' abilities to interact with their peers.
10:05 So AI is by no means a panacea.
10:08 -Over 90% of innovation in AI for education fails.
10:12 I believe that it is because they do not consider what's...
10:16 the kind of practices
10:17 that's happening in the classroom
10:19 and it's built in this black box.
10:20 Second is also, often in computer science,
10:23 I have a background in computer science,
10:26 often we make oversimplifying assumptions
10:28 about real world context.
10:29 So we are sort of assuming
10:31 that these factors do not play a role,
10:34 and we build a system for this ideal context.
10:37 But classroom, real world classrooms,
10:39 are far from ideal.
10:40 Students have very different needs.
10:42 There's a lot of learner variability.
10:44 They come in with a lot of preconceptions,
10:46 that often is hard for us to catch right away.
10:49 Westin: So with all the hype, and a fair amount of failure,
10:53 how can an investor sort out which AI education applications
10:57 have the most promise?
10:59 -The AI hype certainly has led to a lot of funding
11:02 for what I would call as
11:04 systems that are generated to put a lipstick on the pig.
11:10 Is that the phrase? [Laughs]
11:12 If you haven't even thought about, fundamentally,
11:15 how are you going to improve education,
11:17 and you're only coming from the point of view of,
11:19 "We have this AI tool
11:20 "and we're going to find something to apply to."
11:22 It's not going to work.
11:24 It's yet another fancy tool, but the underlying things
11:27 that have been broken in education
11:28 continue to be broken.
11:30 Westin: Whether it's investing, or teaching, or learning,
11:34 everyone agrees that, in the end,
11:36 it all comes back to the teachers themselves.
11:39 -I think that's the real promise of AI.
11:41 It allows teachers to do what they do best.
11:43 When you think of your favorite teacher,
11:45 you don't think of, you know,
11:47 the teacher who taught derivatives the best.
11:49 You think of the teacher who encouraged you,
11:52 who made you feel smart, who made you feel capable,
11:55 who made you feel like you could learn anything
11:57 and accomplish anything.
11:59 And I think that's where teachers really shine.
12:02 That's where AI is really a democratizing power,
12:04 I would say, in education.
12:06 -There is something about that human intent.
12:07 There is something about a human being
12:09 caring about a child, about a student,
12:12 which I don't think AI is able to do that.
12:15 -My dad just a few years ago,
12:17 he changed a headlight in the car.
12:18 And I'm like,
12:20 "Why don't you take it somewhere and get it done?"
12:21 He said, "I watched a YouTube video
12:22 "and it taught me how to do it."
12:24 So I think that we have some dispensations
12:26 that are happening in education right now,
12:28 where some shifts are happening.
12:30 Teachers aren't being minimized,
12:32 but our roles are changing a little bit
12:33 to where we're guiding students to learning,
12:35 and we're able to personalize learning more
12:37 because we have the time to be able to do that.
