Latest / Integrate This! Ed Tech Podcast / AI for Education: Integrating AI into Learning with Amanda Bickerstaff

Transcript

AI for Education: Integrating AI into Learning with Amanda Bickerstaff · Integrate This! Ed Tech Podcast

0:00
Welcome to Integrate This, your go -to EdTech
0:03
podcast where curiosity meets classroom innovation.
0:07
Join your hosts, David Berner and Jeremy Mikla,
0:09
as we explore the rapidly evolving world of educational
0:12
technology. From game -changing tools to insightful
0:15
conversations with today's top EdTech voices,
0:18
we're here to help educators stay informed, inspired,
0:21
and ready to innovate. So tune in, level up,
0:25
and let's integrate this. and we're really excited
0:29
to have her on our program. Amanda has just kind
0:35
of like an amazing background, but yet at the
0:38
same time, she's really doing some very great
0:40
things when it comes to AI literacy and AI used
0:45
by teachers and students and really gives you
0:48
kind of a great perspective. When it comes to
0:53
the things that we are doing or not doing, should
0:55
be doing maybe things we want to avoid. And we're
0:59
just really excited to have here with us. And
1:02
you know, Amanda, thank you for for being here.
1:05
Well, I have to say that was the most fun intro
1:08
to a podcast I've ever been on. Like, why wasn't
1:10
I invited to go play video games, guys? That
1:13
that actually came from ISTE in Denver, right,
1:17
David? Yeah, we went to like one of those after
1:21
hours type things. And it was at I forget the
1:24
name of the place, but it was like they had all
1:27
of these old school arcade games and we were
1:29
just kind of like going around and plugging the
1:32
tokens into there. I think I beat you at Miss
1:34
Pac -Man, but you said Jeremy set a high score
1:37
on the machine. I will say I do have a Tetris
1:46
lamp behind me because I am like a really strange
1:50
skill as I am. very, very, very good at Tetris.
1:54
To the point when I was 16, I actually broke
1:58
a desktop game. And on the flip phone, since
2:02
I'm old, everybody, the one on the flip phone,
2:05
I was faster than the game. So I could play for
2:07
three or four hours in one game. So that's the
2:11
only thing. But everything else, you would have
2:13
shellacked me and anything else. But I will say,
2:16
I couldn't beat you at Tetris. That sounds awesome.
2:20
So other than being, you know, like a top -notch
2:23
Tetris player, you know, Amanda, can you describe
2:25
a little bit for us kind of your background?
2:28
I know you're originally a classroom biology
2:32
teacher, but how did you go from being biology
2:35
teacher to founder of AI for education? Well,
2:39
I mean, I think that, you know, best laid class,
2:41
I didn't even want to be a teacher. I will just,
2:44
it wasn't in my... kind of thought process about
2:48
what I wanted to be. I've always been very externally
2:52
focused. I've always been a big, like, helper.
2:54
I'm kind of that, you know, that person. And
2:57
when I was in college, I wanted to, like, if
3:00
you asked me what I wanted to be when I was little,
3:01
I wanted to be a doctor. And so that was a way
3:04
that I could help people every day. And I went
3:05
to college and, like, was diagnosed with a form
3:09
of arthritis that was pretty debilitating when
3:11
I was 20. And it was very clear very quickly
3:14
that I could not be a doctor. I couldn't go through
3:17
residency and all the things. And so I had to
3:20
recalibrate very quickly and think about what
3:23
could I do of service to others? And I'd always
3:26
enjoyed tutoring and everything. So that's what
3:28
actually brought me to be a teacher. So I was
3:30
an alternative route to certification teacher.
3:33
Shout out to all of those wonderful people. That
3:36
gets thrown into the deep end. So I was 22 years
3:38
old. I moved to New York City from Atlanta. And
3:42
I started teaching six weeks later, you know,
3:45
in an incredibly high complexity school, like
3:47
one that doesn't exist anymore. And so that was
3:50
my job. But I also knew that I, like the same
3:52
way that I couldn't be a resident. I wasn't going
3:55
to be able to teach in a high -pressure environment
3:57
for hundreds of years. So I was actually a teacher
3:59
for a shorter time than I think people know.
4:02
I was there for three years, and then I started
4:04
working in education. And I think I was still
4:08
questioning if education was right for me, even
4:12
when I was thinking about my next steps. But
4:14
I managed to get into a PhD program that I've
4:18
joked about in many places that I did not finish.
4:23
it kind of forced me to think through what I
4:26
really wanted to do. And what was, I did my PhD
4:29
the wrong way, which is one of the reasons why
4:31
I didn't finish it, is I almost worked, I worked
4:35
like 60 hours in different jobs, whether it was
4:38
research assistance or program evaluation or
4:40
curriculum writing, that it kind of just showed
4:44
me that I was much more interested in outcomes
4:48
and impact than I was in research and like writing.
4:52
And so I wish I had figured it out, though. I
4:54
will say, guys, it took me like four years to
4:56
figure that out. I think I was a good kid, a
4:59
good student. I was going to finish without,
5:01
even though I was actually not enjoying it, nor
5:03
did I see my place in academia really once I
5:06
got started. So what brought me to where I am
5:10
today is that it's really hard to make a transition
5:13
from education. And I think it always has been.
5:16
How do you get people to look at you? as like
5:19
your skills. But luckily, I'd built a lot of
5:22
new skills through that PSG program with all
5:24
the work I was doing. And I was able to start
5:25
working in education technology. And, you know,
5:29
really around pressure learning around STEM,
5:33
like not, not generally, I everybody there was
5:35
never an AI moment in any of this. But that led
5:39
me to like kind of continue to grow. pretty quickly.
5:43
So I moved through different jobs to the next
5:47
thing where the before I started education, I
5:49
was the CEO of an education technology company
5:52
in Australia that work with 700 schools. So this
5:56
is a you know, it was not a small organization
5:58
and really learned how to think about education
6:02
technology from the perspective of I was leading
6:05
a team building technology for schools. And everybody
6:09
If you ever get mad at all the ed tech not working,
6:11
I totally understand why. We can have a whole
6:15
podcast on that. But one of the things that is
6:17
one of the reasons why it doesn't work is that
6:19
it's very hard to bill for schools. And we all
6:22
have different needs and tech stacks. And so
6:25
it's very hard to do. So I learned just a lot
6:28
very quickly in the hard way, especially during
6:31
COVID. But when I, so when I left Australia,
6:35
I have talked about this before, but I was really
6:37
burnt out. So I took six months abroad because
6:40
I knew if I, I'm such a like a worker bee that
6:42
if I came back to New York City, I would need
6:44
to pay for my like apartment. When I knew if
6:47
I was gonna have to pay for my apartment, I would
6:48
have to get a job. And I just genuinely couldn't
6:51
do it. I couldn't, I couldn't do that. So I traveled
6:53
abroad. So I wasn't even thinking about AI at
6:56
all when it, when Chai 60 was released. So into
6:59
2022, I'm in Japan, having ramen, going to temples.
7:05
I wasn't thinking about chashupe tea. But then
7:07
when I came back to the US and I was thinking
7:10
through my next steps, trying a bunch of different
7:13
stuff. But the first time I used chashupe tea,
7:15
this is a story that I've told before. It's true.
7:18
There's no kind of crystal or rose -colored glasses.
7:23
It didn't take me more than one prompt to realize
7:26
that someone needs to do to be thinking and talking
7:29
about how this is going to impact education,
7:32
what it meant to have practical applications
7:35
of generative AI and teaching and essentially
7:37
learning, and that know what? Very few people
7:41
in this world can pick up Chach VT and understand
7:45
that while anybody, there's no barrier to entry.
7:48
If you can text or type, you can use generative
7:51
AI. But to use it well, requires a pretty significant
7:55
amount of rethinking of how we interact with
7:58
technology. So while it was never in my head,
8:02
I could still not explain blockchain, everybody.
8:04
I still don't know. For some reason, this made
8:08
sense to me in such an intuitive, deep way. And
8:11
what's really cool about it, and I know this
8:13
is one of the reasons why we love what we do,
8:15
is I've never, even starting, never more than
8:19
a couple of weeks away from working with a school.
8:21
So everything that we have done since June, 2023,
8:26
we have been working with practitioners. Never
8:29
have we ever come up with anything that was not
8:32
tested in real time with the people that need
8:34
it the most. That's just amazing to hear you
8:38
talk about this whole story. And what I heard
8:43
you talk about, I think that stuck out to me
8:45
something. that you mentioned is that you've
8:47
always been kind of focused on the outcomes of
8:49
it. And I could see where your brain is going
8:52
then when you first sit down and you try chat
8:54
GPT and you're putting in that first prompt,
8:57
right? It's all about what is it going to give
8:59
back to you? And then how can you take that and
9:03
use that to make whatever it was that you were
9:04
doing at that time to make it better? You know,
9:07
and I think that's just something that as educators,
9:10
we naturally then want to find that way to make
9:13
it better for our students. And that's kind of
9:16
how I've approached EdTech is like, it needs
9:20
to be something that. Well, well, number one,
9:23
you've got to start with you knowing what your
9:25
learning goal is going to be. Right. And then
9:27
trying to choose what the best tools are going
9:30
to be to meet that. But it has to be something
9:32
that can either accelerate a student's learning,
9:35
accommodate a student's learning or engage them
9:39
in such a way that they've never been able to
9:41
to do in a different way. And so that's what
9:44
I kind of feel like. I'm not trying to put words
9:46
in your mouth, but I kind of feel like that that's
9:48
kind of been a driver for what you've been able
9:52
to do with your company. And maybe can you talk
9:57
about where's AI for education now and where
10:00
is it going? OK, so I'm just going to do some
10:05
inside baseball. It was very hard for David and
10:09
Jeremy to get me on this because we've been trying
10:11
to do this for like three months. So we are growing.
10:14
We are. when we started, first of all, when we
10:18
started, it was me. Like this is, it was me and
10:21
a laptop. You know, I was literally learning
10:23
in front of people. It's very funny. I have never
10:26
really tested a new feature until in front of
10:29
people. Cause I feel like that's kind of genuine
10:30
of like trying it out together. Like the first
10:33
time I used GPT -5 was in front of a group of
10:35
teachers. So that's always been a part of the
10:38
journey. So we have, like I said, our first school
10:40
was in June of 2023. Then I would say that the
10:45
business started in August of 2023. So this is
10:48
our third school year. And we just did a pretty
10:52
big analysis. So we've worked with over 250 institutions,
10:58
many of them multiple times since August, 2023.
11:02
That is K -12, higher education, associations,
11:05
nonprofits, and both international and national.
11:09
Our course, which these lovely men came and hang
11:12
out with us, is now at 30 countries. Participants
11:17
are 30 countries with 600 people across 275 institutions
11:24
just since January. So our reach of the work
11:28
that is more structured work is in close to 600
11:35
institutions already. But a lot of what we do,
11:38
and I think something that sets us apart, is
11:40
that we actually started with a free resource
11:44
kind of focus. And so we have more free resources
11:47
than probably any organization out there. And
11:51
I have had for a long time, even though we're
11:54
like this big versus some of the organizations.
11:56
And that reach is somewhere in the millions,
11:59
especially now. So we have about 750 ,000 people,
12:06
more than a million site visits every year to
12:09
our site. So that means that while our kind of
12:13
direct support is around those hundreds, our
12:16
larger support is closer to the thousands of
12:19
institutions. And one of the things that's super
12:23
nice is you're I mean, I know you're talking
12:25
about the reach of that and Jeremy and I being
12:28
a part of that. And one of the things that I
12:31
did you hear all this lovely You know, I just
12:33
have to do that out there and repeat that. I've
12:36
got to meet them in person. You guys have come
12:38
to the, I like, you know, I, we are so lucky.
12:40
Can I just say before David, sometimes education
12:44
gets a bad rap, like teachers or leaders or organizations.
12:49
We have top to bottom had the best experience.
12:51
We have. So many amazing experiences with educators.
12:56
And you know what? Like we need to start that
12:58
rhetoric. That's so negative. So you are lovely.
13:00
And I can I will even do like if you want to
13:03
do like one of those little like Amanda, because
13:05
that says that David and Jeremy are lovely. I'm
13:07
going to put it in my LinkedIn profile. Put that
13:11
on there. I'm going to copy it and just show
13:13
it to my wife. But yeah, so the other thing too
13:18
though is like the amount of resources that you
13:20
have put out there because it's given I think
13:22
even Jeremy and I have talked about some of the
13:25
resources that we we gained from AI for for education
13:31
and the thing the thing about it is that it starts
13:34
to move things in the right direction especially
13:37
like when you're talking about AI literacy and
13:39
Having in a framework for that and what that
13:42
looks like and what all goes into that Can you
13:44
talk a little bit about the AI literacy? Well,
13:47
I guess a couple of things, because one is defining
13:50
it, right? Because that was a really difficult
13:52
piece for people to start to wrap their heads
13:54
around. And many organizations were trying to
13:57
say, like, well, what is AI literacy? And then
14:00
what pieces should go into that? So can you talk
14:03
a little bit about for listeners, like, what
14:05
is AI literacy in your opinion? And then what
14:08
are the main pieces to that? And it's interesting
14:11
because it's the one thing we haven't published
14:14
outside of our work with organizations and our
14:17
participants. And the reason why we have it is
14:19
there's a lot of noise in the space. And so we
14:21
don't want to come out with something until we
14:23
feel like it is meaningful for the environment.
14:27
But one of the things that was really nice is
14:29
that we started doing this very early. So I feel
14:32
like if you're going to like, if I'm going to
14:34
add something to my background, it's going to
14:35
be like a pillow that's like embroidered and
14:37
says AI literacy with like 15 exclamation points.
14:40
Because it feels like the thing I've been talking
14:42
about the most since I started. And I don't think,
14:45
and David, it's a really good point, because
14:46
I don't think I knew. what I meant when I started
14:50
outside of I just felt in my bones. It had to
14:53
be practical. It had to be safe. It had to be
14:57
ethical and it needed to be something where you
14:59
either saw of that kind of AI literacy and like,
15:03
like you actually saw someone do it and then
15:05
you learned how to, it had to be something about
15:08
learning about and learning with. And so even
15:10
though it wasn't a formal piece at the very beginning,
15:13
that's never been far. In fact, I think the thing
15:15
that's so wild about this moment in time is that
15:18
our first, when I was just me, our first flagship
15:21
we ever did is almost exactly the same in flow
15:25
as today. And so the idea of being able to see
15:29
that hundreds of times has been really helpful.
15:32
So what we believe AI literacy is, is that AI
15:35
literacy is a skills and knowledge and mindsets.
15:37
And I just want to underline mindsets because
15:39
that's so important here, of being able to use
15:42
AI in three ways, safely, ethically, and effectively.
15:45
And we call it our C framework just to make it
15:47
easy. And the reason why I want to underline
15:51
mindsets is that One of the big things that people
15:55
pushed back on me having a private library when
15:57
it was just me and a laptop was, oh, prompting
16:00
is going to go away. And if you think about prompting,
16:02
that means the ways in which you interact with
16:04
the bot. Oh, it's not going to be important.
16:06
The bot will do it for you in three years. But
16:08
all I could ever think was, it matters right
16:11
now. Why would we not focus on right now? Why
16:13
does it have to be five years in the future?
16:16
But what I have seen is that while the bots have
16:18
become more sophisticated, prompting techniques
16:21
still matter. But it's almost more the mindsets
16:23
of how you are the conductor and evaluator of
16:27
an AI's outputs. And that's why it works so beautifully
16:30
with the C framework, because that means that
16:32
you are thoughtful about what you use and how
16:35
you use it, and then what you give the bot. Am
16:38
I giving the bot all of my information about
16:41
myself? Am I using it in ways that could be deeply
16:46
unhealthy for me potentially where I'm cognitively
16:49
offloading or I'm over relying on these tools
16:52
to do the human part of life, so to speak, which
16:55
we see right now happening with those that are
16:57
having mental health crises using generative
16:59
AI as a companion. The ethical bias, you cannot
17:05
talk about AI without bias. It is impossible.
17:08
These systems are built and trained on human
17:10
data. And they are also probabilistic, which
17:13
means that if you see something in a training
17:14
data set a lot, you will more likely see it on
17:16
the outcomes. And actually, can I tell a little
17:19
story from yesterday? So when David and Jeremy
17:22
are in our trainer cohort, we talk about this
17:26
Stanford study about how these AI systems, if
17:30
they're asked about students in classrooms, it
17:33
kind of will spurt out all of this, like how
17:36
we've been thinking about kids for a long time.
17:38
And the example is that the star student in English
17:41
class is a girl and the struggling student is
17:44
a minority boy. So one of the leaders yesterday,
17:48
we were down at Mount Hershey School, did this
17:50
on his own twice. And it was Louis and Miguel.
17:53
were the two kids that needed help, and Maria.
17:56
It was Maria twice in two different context windows.
18:00
And it was like, oh, but Maria was so nice. She
18:03
smiled a lot. And she's a tutor. And so to watch
18:08
that in real time, that that's what we think
18:10
about ethics. It can also be things like climate
18:13
and things around deep fakes and misinformation.
18:17
And if you understand that these systems are
18:20
not thinking and that they are going to be designed
18:23
by people that have their own blind spots and
18:24
then also the way that they work cause ethical
18:27
issues all the time, that then what we can focus
18:31
on is the affect effective. Like what does this
18:33
mean for me? Like I need. I need assistive technology,
18:37
even though I don't have a personal IEP, but
18:39
I know that I struggle with a blank page. Or
18:41
I have always wanted to do this crazy thing,
18:44
and now I can be augmented. I can have a collaborator,
18:47
a problem of practice support. And so that's
18:50
where that effective comes in of like, I understand
18:53
the what and how and why of it. then I could
18:56
start supercharging what I do. Because the thing
18:58
is, no time in history do the majority of humans
19:02
have access to amazing technology, primarily
19:06
for free. And that is something that, if we do
19:09
it the right way, could actually start to really
19:12
change the ways in which people gain, instead
19:16
of them losing or being monetized, that we can
19:20
gain from these systems even more than ever before.
19:24
The big things, too, that was super helpful for
19:27
me was like the whole iceberg concept of all
19:31
the different things that were underneath what
19:34
the initial catalyst was for everybody getting
19:37
concerned, which was the cheating thing. Right.
19:39
And so for me, like when I I do. we do our own
19:43
trainings on AI, right, with our organization.
19:45
And one of the things that was very helpful is
19:48
seeing that so that I can begin to construct
19:52
and put in examples for teachers of what that
19:54
looks like, like what is cognitive offloading
19:57
look like? What should you be concerned about
19:59
with like the deep fakes or giving the examples
20:02
of the companions like the chat bots, kids are
20:06
leaning into those more as a friend or as a significant
20:09
other in some cases as well. So I know for for
20:13
my own trainings and things, that's been a very
20:16
helpful thing to look at. Because then it starts
20:19
addressing so much more than just the cheating
20:22
piece. So man, I feel like if you asked me why
20:26
I started to first of all, I don't know how this
20:29
has happened. I now have 70 ,000 LinkedIn followers.
20:31
And that sounds like a huge brag, but it's wild
20:34
to me. I am not a social media person. I hadn't
20:38
posted on LinkedIn for a year before I started
20:40
posting in April of 2023. But the reason why
20:44
I started posting is I was already so annoyed,
20:47
so annoyed at how all anyone can say was that
20:51
AI was for cheating. I couldn't believe that
20:56
we could not have some sort of nuanced conversation.
21:00
I know, I was naive, everybody, but in my head,
21:02
I was like, there's so much more. And so the
21:05
iceberg metaphor, I have a new metaphor, guys,
21:08
that I'm trying to work out. It took me a little
21:09
bit of time to figure out what we were doing,
21:12
but the iceberg is this idea of like, it's truly
21:16
an iceberg. I think sometimes when people say
21:17
icebergs, they're like, Oh, but it just means
21:19
there's a lot below the water. But what I actually
21:22
mean is that if we go the way we're going where
21:25
the only thing we're talking about in schools
21:26
is stopping kids from cheating, then we will
21:29
hit the iceberg because everything below it is
21:32
going to have an impact potentially more quickly.
21:36
That's like the issue with academic integrity
21:38
that is different today than it was in the past
21:41
is it has never been easier before not to cheat,
21:44
but to cognitively offload. And the thing is,
21:48
to go and cognitively upload everything would
21:50
have required an enormous amount of effort for
21:52
a young person before, whether that was notes
21:54
and essay mills, and it would cost a lot of money
21:57
and effort. But now with this easy button for
22:00
every traditional assessment, that if we're not
22:04
talking about cognitive offloading, of course
22:06
we want kids to have honesty and integrity. But
22:09
what is school for? Is school for a kid? submitting
22:14
a thing at the end of a semester and getting
22:17
a grade? Or is it about the actual learning path?
22:20
And like, why have we prioritized the destination
22:23
so much that there's almost like this incentive
22:26
to cheat in some ways and shapes or forms? So
22:29
I think that I get so like, just clearly you
22:32
can hear it in my voice, like we got us, we got
22:35
to trust kids, we got to talk to them and we
22:36
had to change things. But secondly, we've been
22:38
talking about artificial companionship for more
22:40
than a year. I think we were, and I'm seeing
22:43
before the Character AI thing that came out,
22:46
because I was in Iowa, and a leader who we'd
22:50
worked with the day before was with a training
22:52
with us, came up to me and said, Amanda, I really
22:55
appreciate the fact that you were talking about,
22:57
we weren't calling it digital wellbeing then,
22:59
but we were talking about the greater piece,
23:01
because my 16 -year -old neurodivergent daughter
23:04
is spending six hours a day on Character AI.
23:08
And she was like, I don't know what to do. because
23:11
my daughter isn't talking to anyone else. So
23:14
there's like a positive to this, but I know it's
23:17
unhealthy. And so since then, and a second grader
23:21
having Snapchat, he thought AI was out and he
23:25
was using it every day without any knowledge
23:28
it wasn't real, was why we started talking about
23:30
this literally before anything major happened
23:34
in the world. But I think that it's so scary
23:38
to me that we cannot seem to like, get out of
23:41
our own way here. Our job as educators is not
23:44
to catch kids cheating. And some of us, I think,
23:48
feel that way. And I get it. And if you're one
23:49
of those people, I understand why. I was a high
23:52
school teacher as well. And you want kids to
23:55
learn. But we have to reframe how we think about
23:58
this moment in time, or we're going to miss it.
24:00
We're going to miss this opportunity to not have
24:03
a lost generation, similar to COVID, but also
24:06
to have important conversations. That can actually
24:09
make things better. Sorry, so box off. Right,
24:14
so I see it almost to steal another one of your
24:17
terms that you mentioned a little bit earlier,
24:19
but perhaps what we need is a shift in mindset
24:21
from teachers and reframe the conversation to
24:26
not be about the cognitive offloading and pushing
24:29
the easy button. And by the way, I've started
24:32
to kind of think of it too, as we also need to
24:34
do the same thing with creativity. We don't want
24:37
to. offload all creativity to AI to do for us
24:41
either. So we then need this mind shift that's
24:44
going to help us flip the script on that, right?
24:46
That's going to lead us to have more critical
24:49
interactions with AI that's actually going to
24:52
elevate and enhance our thinking and our creativity.
24:57
So my question is, you know, like, what do you
25:00
think we need to do to make that flip? Is it
25:02
just going to is it is it just continue to remind
25:07
teachers that there's more that you can do with
25:09
this, that it's it's not just trying to catch
25:11
teachers or or is is there another approach we
25:14
should be looking at? Well, I think the first
25:16
thing is my new metaphor, guys, is that like,
25:20
I have to figure out how to stop beautiful. So
25:22
don't at me if it's not great. Or if you have
25:24
a great idea, tell me but my new metaphor is
25:26
that like if you had just like, this is perspective
25:30
that like AI is this wrecking ball for traditional
25:33
like it's coming in, it's causing cracks and
25:36
it's like hitting like, ooh, traditional assessments
25:38
and like student interests and whatever. But
25:40
the thing is, is that it's actually not causing
25:43
cracks. The cracks have been there for decades.
25:47
What it's doing is it's toppling, it is forcing,
25:50
it is pushing over what's been wrong. It's been
25:53
exposing the cracks and in some cases breaking
25:56
the system. And so I think that that's part,
25:59
so what I would say is that it's not up to teachers
26:02
If we ask it to be teachers and we don't do the
26:04
work from the ecosystem and the strategic work
26:07
at a school system, federal, like state, then
26:11
it doesn't matter. Because no teacher in any
26:14
classroom is going to be able to come against
26:16
a broken system. And we know that, right? We've
26:19
all been teachers. And the most innovative people
26:24
still have to come back to a world in which their
26:27
kids have to pass a test. And that has really
26:32
caused an enormous like kind of almost like tightening
26:36
of opportunity, which is why I kind of get inspired
26:39
by like elementary school as a model moving forward,
26:42
because I think that that's where we focus on
26:44
durable skills more. That's where there's more
26:46
freedom. That's where there's more time to like
26:48
learn foundational skills. That's where there's
26:50
more creativity and that there are things there
26:53
that we could do. But like, let's just be real,
26:55
everybody, the system itself. has not worked
26:59
for most kids for a really long time. And if
27:02
we think that AI has caused that, no, AI has
27:06
exposed it the same way that COVID did. But the
27:09
difference between COVID and AI is that AI is
27:11
only going to get better, and it's going to be
27:15
perceived right now, but then a requirement for
27:19
a job. It's a perceived idea right now. I'm gonna
27:22
say that because most people don't even know
27:24
what AI literacy means. I was just in front of
27:26
a board of trustees and they were asking me,
27:29
right? The same way that we talked about just
27:30
a moment ago. But I think that that is really
27:33
like meaningful for me. But I wanna give an example
27:35
from yesterday. I'm gonna go back to yesterday
27:37
where we were doing a... We're really lucky.
27:41
We actually are starting to do longitudinal work
27:43
around strategy and larger work. And so we were
27:47
doing a vision and guiding principles day. So
27:49
we started with AI literacy, because you guys
27:51
know me. So we did AI literacy in the morning.
27:54
And then what we did is we did these creative
27:57
tensions, where we had everyone stand up. And
28:00
it was teachers, leaders. Cross -functional and
28:03
they had to go to like if they were not poles
28:06
They were like two big statements like what is
28:09
the barrier for educational change on one side?
28:11
It was infrastructure and like like risk aversion
28:15
and the other was it was mindsets and Like like
28:18
ship like like actual changes and people were
28:20
like really interesting because for that school
28:22
It was mindsets like they didn't even the infrastructure
28:25
attack. No, okay It's gonna be mindsets because
28:27
that's gonna lead to the infrastructure changes
28:29
And then what we did is we then came up with
28:31
some stakeholder mapping with empathy maps with
28:34
students and teachers. And then what we did is
28:37
some vision statements and guide principles.
28:38
But what I did is I said, hi, everyone, no one's
28:41
allowed to use AI for these parts. But what I
28:45
did is that every time we came to, the work had
28:48
been done, I took pictures of the stakeholder
28:51
maps, empathy maps, put them into Claude. And
28:54
what it did is it synthesized it, and now instead
28:57
of spending the 45 minutes going around the room
28:58
talking about our posters, we talked about the
29:01
problem. And then when we did the vision statements,
29:04
it came together and found the differences and
29:06
the similarities, and that led to another conversation.
29:08
And by the end of the day, not only had we come
29:10
out with guiding principles and vision in one
29:12
day, but also they had learned and modeled where
29:17
AI can augment, not replace their own work. And
29:21
to watch adults have to sit down and free write
29:23
for eight minutes was exactly what would happen
29:25
in a classroom. But then when the outcome was
29:28
so much better, because they didn't have to do
29:30
that other work, and they could talk about the
29:32
meaningful part, I cannot tell you the amount
29:35
of light bulbs. The fact that we were able to
29:37
do both more work than possible by using AI strategically,
29:40
but then also model this meaningful practice,
29:44
that's what should be happening in staff rooms
29:46
and classrooms, that it's not. this thing that
29:50
takes away learning, but it's a thing that allows
29:52
us to focus on when done correctly with AI literacy,
29:56
the part that matters the most. So I see it.
29:59
I see it. You've got, you know, I was imagining
30:02
two buildings and I was trying to picture your
30:05
wrecking ball. metaphor there too. And on one
30:08
side is the infrastructure, the other side is
30:10
the mindsets. And then I just see like someone
30:13
standing over there on the mindset side and wanting
30:16
to once they've kind of come to that realization,
30:19
push that sucker towards the other side and have
30:22
it start exposing more of those cracks or maybe
30:25
even starting to knock it down. And I think the
30:27
only thing that that image in my head would need
30:29
is like Miley Cyrus. I'm trying not to have a
30:34
weird Miley Cyrus moment, guys. But I do agree,
30:38
because the thing is, is that what we see, and
30:39
this is why it gets exciting to us, because when
30:42
I started AI for Education, 0 % of organizations
30:47
had enough AI literacy to have these conversations.
30:51
And we're now in year three of, I would say,
30:54
of course, the 2022 and 2023 school year. was
30:59
like partially impacted, but not really. It was
31:02
there, but it wasn't really. So let's say this
31:03
is the third, third, you know, the third big
31:05
piece is that like, like we can have these conversations,
31:10
but you know what it does is like the people
31:11
in the room were like, we need, you know, what
31:13
was so cool and a relatively conservative place.
31:17
We need intentional risk taking. We need our
31:21
students to feel like they can, they can take
31:23
risks, our teachers and our leaders to create
31:25
space for that. Oh my, like that something sounds
31:30
so little, but think about like what that means.
31:33
It means actually starting to change the way
31:36
we do things in a way that's supported. And I
31:39
think that that's what happens. Like you talk
31:41
about the people themselves start to push over
31:43
the walls because they see they've known the
31:45
cracks are there, but they've never had, they've
31:47
never had the support or the reason, right? Cause
31:51
it's hard change is hard, especially in education.
31:54
But I don't know. I feel really galvanized from
31:57
that work yesterday. And I wish I could do it.
31:59
It's like the train, the trainer, our flagship.
32:01
There are things that we do, our guideline development,
32:04
that really there's a moment where I see it.
32:06
I see the future of what we can do together.
32:09
And yesterday was one of those moments. What's
32:12
your opinions on some of the hang ups on that?
32:17
Like is it just like where you say you have districts
32:20
that I think In some cases, like teachers just
32:24
don't know what to do. Or maybe leadership doesn't
32:28
feel comfortable quite yet pushing forward with
32:32
that. And so there's like a leg that's happening
32:35
there in your own conversations. What do you
32:38
think is the biggest impediment? Yeah, I think
32:42
the first thing David is that we're not quite
32:44
a place where everyone agrees this is important.
32:47
So we're still not at a place where the majority
32:50
of schools and districts see this as a priority
32:54
that would shift other priorities. And so we,
32:57
I would call it a canary in the sense of like,
33:01
we were so early in the work that we do that
33:03
we've kind of seen and hopefully moved the conversation,
33:07
right? And so what I would say is that we're
33:09
still not quite to a place where the majority
33:14
of leaders believe this is worth pushing into
33:18
whether that is because they believe it's they
33:20
listen to the noise that it's a hype cycle or
33:23
it's not making that much of change right now
33:24
or honestly like we just need kids come to school
33:27
like you know we have chronic absenteeism or
33:30
you know we already know that you know income
33:33
level of individuals in schools makes a difference
33:35
on access to training and these tools right so
33:39
you know those schools that have high complexity
33:41
this is not seen as a priority because there
33:43
are like 9 000 priorities that are all like meaningful
33:46
and immediately meaningful, right? Instead of
33:49
a meaningful potentially in a year or two years.
33:51
So I think that's one. Number two is I think
33:54
that there is the iceberg metaphor is apt here
33:57
because there's so much negative rhetoric. Oh
34:00
my gosh, we are like, people have internalized
34:04
this. And like you have people that heard two
34:08
years ago that making 10 images was like driving
34:10
four miles. And that was true then. It is a thousand
34:14
percent not true today. There are questions about
34:19
it's only gonna be for academic integrity breaches
34:22
or shows that as a teacher I don't care and I'm
34:25
not trying hard enough. Like I hear teachers
34:27
and leaders saying it feels like cheating when
34:29
I talk about it. And so we've created this like,
34:33
and I don't know, it's not intentional, but like
34:36
I'm gonna call out the Atlantic article that
34:39
just came out where the young woman, a senior
34:42
at a high school in New York City says, it's
34:45
all bad. She's like, Jedi is just ruining education.
34:49
But the thing is that I know about New York City
34:52
is that there's no policies or a literacy in
34:54
place. So that is happening in a school where
34:57
there's no support. And how can we be surprised?
35:00
So I think that what is difficult right now is
35:04
that we haven't reached the moment where this
35:10
becomes non -negotiable, and the loudest voices
35:17
are the most negative voices. No, that just makes
35:22
sense with, like, I think so much of it, too,
35:25
with how quickly everything has moved. There's
35:27
a lot of districts that are still just trying
35:29
to wrap their head around the informative piece,
35:31
the introductory piece to AI, and not so much
35:35
into even the applications of it yet. And they
35:38
also, one of the pieces that sticks out, is it's
35:40
moving too quickly. How could we ever catch up?
35:43
Or how could we ever keep up? And I always have
35:45
to tell people, I feel like a therapist. A, it's
35:47
not too late. It might be in a year. We started
35:50
changing that. I'm not nearly as nice as I have
35:52
in the past. And also, if our flagship is 80
35:57
% the same as it was two and a half years ago,
36:03
then we have, like, gender of A has changed,
36:06
but GBT -5. on the whole works exactly like GbD
36:09
3 .5 in terms of UX and UI. It still hallucinates.
36:13
It's still sycophantic. It still is creative.
36:17
It's still random. It's still like, it has new
36:20
features. You can create animate versions of
36:23
yourself now and you can do research and there's
36:26
more reasoning capabilities. You can do more
36:28
complex work, but there's just like these, I
36:32
don't know what the word is and I need to take
36:34
time to actually look up if there's a word for
36:35
this. The thing that we're coming up against
36:38
is that people hear a thing once that sits with
36:43
them, and they can't move past it. Climate is
36:46
at the individual level, not the data, not the
36:49
tech company level. That's where the climate
36:51
impact is. But there's no movement, or AI is
36:56
moving too quickly. No one can do anything. How
36:59
do we change? How do we keep these? these anchor
37:04
statements that are not true from becoming these
37:08
organizing principles going forward. And it's
37:10
something that we struggle with a lot is like,
37:12
like, like, even though it was true two years
37:14
ago, it's not true today. And like, we have to
37:16
make move past the same thing with prompt engineering
37:18
not mattering. Guess what, guys, prompt engineering
37:21
still matters, even though we're in GPT five,
37:23
like, if anything, it's, it's more important
37:26
in the sense of like, understanding how to use
37:28
the tools. We underestimate to like, if somebody
37:31
trainings that I've done, we really underestimate.
37:34
There's a lot of teachers that don't know where
37:36
to start. And the prompt engineering, the prompt
37:38
frameworks are huge for that because they're
37:41
like, well, I don't know what to do with this.
37:43
So giving them some direction is helpful. I've
37:46
had plenty of conversations around that. But
37:49
also there's this new rhetoric that all teachers
37:52
are using generative AI because they all have
37:54
magic school or they are, you know, all teachers,
37:56
all students are using AI. Guess what? I believe
37:59
very strongly that out of the nine percent of
38:01
the humanity that's using chat routine monthly
38:03
that the majority of them do not know how to
38:05
use it. So is there some training you think that
38:10
needs to be involved as part of the AI literacy
38:13
and it's just a foundational understanding of
38:17
how generative AI works? And would that help
38:21
lead to more effective use down the road? Oh
38:23
my gosh, yes. Better UX and UI of the tools,
38:27
where they don't give you the, don't get me started,
38:30
but if you go to Chef team right now and you
38:32
look at the suggested prompts, they are the worst
38:35
prompts you could ever use. They're like lesson
38:38
plan. That is the worst way to prompt an AI system.
38:43
And that is what those companies are telling
38:45
you to do. And so I think that there is that
38:48
part. But one of the things that we've seen,
38:50
and you guys know it, right? Especially after
38:51
doing the work with us. The best thing about
38:53
AI literacy, if you do it well, it takes 90 minutes
38:56
to half a day. Doesn't mean that that's AI fluency,
39:00
but AI literacy of understanding the limitations,
39:03
the capabilities, getting your feet wet, trying
39:05
it out, creating more ethical use. We do that
39:09
constantly. You guys have seen it. 90 minutes
39:11
to half day. That's it. We can get. people that
39:15
like have not don't even know how to cut and
39:16
paste well to be like oh my god who you mean
39:21
where almost with frustration sometimes a teacher
39:23
will be like you mean that i have been handwriting
39:25
rubrics for 30 years like you know that kind
39:28
of like moment of like where has this been all
39:30
my life but i do think that there's a misconception
39:33
on that too ai fluency takes continued embedded
39:38
thoughtful training and use but ai literacy You've
39:44
seen it. We can just do this. You know, you guys
39:46
do it. We do it like it's not crazy and it's
39:49
something that can have a huge benefit. Yeah,
39:53
can we talk a little bit too just about the student
39:56
side like the I know and you might you might
39:58
say this would be like we're just not quite there
40:01
yet and probably. Is it makes sense given what
40:05
we were just talking about with even teachers
40:07
not understanding fully how to use these but.
40:09
What shifts have you seen in schools gravitating
40:13
towards student use? Because obviously a lot
40:16
of the AI tools that we have that are coming
40:19
out, a lot of the companies have now built in
40:23
some student elements to that while trying to
40:25
maintain some student anonymity in those programs.
40:30
What's your opinion on the student use? What
40:35
I would say is, if you take that kind of canary
40:38
in the coal mine, we were really pushing AI literacy
40:41
for adults two years ago. And we were working
40:44
primarily with early adopters then. We have been
40:48
trying to encourage AI literacy for young people
40:52
for about a year and a half now. We didn't think
40:55
we had to. Honestly, there were other incumbents
40:57
that had done this much longer than us that we
40:58
thought had that nailed. And we just see that
41:02
not being the case. And so we have changed our
41:05
mission to be an AI literacy for a million educators
41:08
and students as of September. And I will say
41:11
that we're starting to convince organizations
41:15
to be really intentional about student literacy.
41:18
So that's very simple. They were prompt -a -thon
41:20
next week with students. We've had students and
41:23
teachers work together. We have student -facing
41:26
courses going. And in fact, this will be our
41:29
first announcement, but we have a free student
41:31
course, similar to our teacher course, launching
41:34
in October. It'll be four hours with certificate.
41:37
It is C framework aligned. It is about, like
41:41
they actually build their own code of ethics
41:43
around it, but they also think about, they're
41:45
all learning gaps. They think about creativity.
41:48
They think about studying. They think about all
41:51
of these like really important applications with
41:55
this really strong lens towards don't replace,
41:59
augment, assist, collaborate. And so what I think
42:02
is interesting though is that there are a couple
42:06
of big barriers to getting this getting AI literacy
42:11
for students really going. One is that there's
42:14
a there's this need to get teachers going first.
42:15
And so if you teacher PD days really being like,
42:20
you know, very so like precious and hard to get
42:24
means that that can be pushed back. Second thing
42:27
is there's this really negative like misconception
42:31
that because kids are using AI, they know how
42:34
to use it. So the similar thing of like digital
42:36
literacy, that digital data equals digital literacy.
42:40
And y 'all, that is just not true. We would not
42:42
be so pissed at our young people in classrooms.
42:45
I'm sorry I'm using the wrong language. But if
42:48
students are using AI well and correctly, we
42:50
would never know. But what they're doing is they're
42:53
using it in the worst possible way that is frustrating
42:56
on so many levels. It shows a complete lack of
43:00
understanding of the technology, of critical
43:02
evaluation, of understanding how to use the tools.
43:05
And so we're not frustrated with students using
43:06
AI. We're frustrated with students using AI badly.
43:09
And so I think that those two barriers are significant,
43:13
which is why, similarly to our free course being
43:16
like kind of this huge thing two years ago, in
43:18
fact, I think we were one of the first free educator
43:21
courses on AM literacy at all after TraxBT is
43:26
that we're hoping that by leading from that place
43:28
of like, students need this and we need to teach,
43:32
give them the space to understand it, that others
43:34
will come around and there'll be more consensus.
43:36
and that students can advocate. If they find
43:39
this, they can advocate for AI literacy themselves.
43:42
But I do think that the last thing I'll say is
43:46
classroom use is that the crazy thing about this
43:48
is that things like School AI and Magic School,
43:51
which have chatbots for students, are primarily
43:54
used in elementary school. And the place in which
43:57
there is the least amount of AI literacy is elementary
44:00
school. So students are actively using GenAI
44:04
tools. without AI literacy as young as a second
44:07
to kindergarten. Yeah, and those conversations
44:12
are actually really interesting too because I
44:15
think as you move further along the grade levels
44:19
and you have where this becomes more of a factor
44:22
in student use, then some of those things, those
44:26
ethical concerns that we talked about earlier
44:28
start to creep in more and more depending on
44:30
where that's falling on the grade continuum.
44:35
Yeah, I will say I think that my new soapbox
44:41
is student allergy. My new Amanda screaming into
44:46
the void and hoping that people listen. But we're
44:49
really hopeful for the course. I think actually,
44:52
honestly, with you two, it's a little more technical
44:56
because we can go more technical with students
44:58
because it's more like. make we have a little
45:00
bit more time. I think you guys would dig it.
45:02
It's actually, we're having, we're talking about
45:04
prompt injections. We're talking, we're like,
45:07
like it's really like, I don't know. I really
45:09
like, like it's, it's, it's not easy to be a
45:11
team of like eight people and be like, we're
45:13
going to create a, you know, this huge free course
45:15
with everything else we're doing, but it's so
45:18
energizing to like have fun with this and like
45:21
new prompts for the library. And it's like deeply
45:24
like ethical and thoughtful. I don't know. I'm
45:27
excited about it. So, you know, what would, you
45:30
know, you mentioned like the least amount of
45:32
AI literacy being given towards elementary students.
45:35
So let's just say like third, fourth grade, you
45:38
know, what do you think an effective AI literacy
45:41
integration would be into one of those classroom
45:46
spaces? I think the most important thing is to
45:49
start to have young people recognize that, like
45:53
how these bots work and don't work, that they
45:55
are not real. that they are designed to feel
45:58
real. A third or fourth grader, when talking
46:00
to Alexa, can understand the limitations. And
46:04
you have on -ramps, right? A lot of young people
46:06
using these tools without knowing it for a long
46:09
time. Snapchat is being used by 10 -year -olds,
46:11
and they have an AI system since two and a half
46:13
years ago. So I think that for us, it's still
46:16
the same. It's not like you don't go deeply into
46:20
ethics, but you can go into Watch it not label
46:23
a map correctly. I don't know if you saw our
46:25
LinkedIn post recently. But you guys, we did
46:27
it live when we were together. Like, it's wild.
46:30
It's so bad. Y 'all, yesterday, it gave me, like,
46:32
we asked Chachiti to label a map as an example
46:35
of hallucination or accuracy. And yesterday,
46:36
we did it. And it gave me, like, a plot again.
46:41
And then it put the plot into a kind of, sort
46:46
of, like, state line map. And when I asked it
46:50
to read its output, it said eight out of 10.
46:53
And I'm like, this isn't even a map. And it's
46:55
like, oh, JKJK, yeah, you're right. But doing
47:00
that with a fourth grader, you think they're
47:01
going to go and trust this moving forward? Absolutely
47:04
not. Showing them if something can create, rewrite,
47:08
or design, that it's going to be generative AI.
47:11
And you have to use it differently than other
47:12
systems, third or fourth grade. Talking about
47:15
are you over relying? Are you going to chat chibi
47:19
tea or whatever about your own personal life?
47:23
Instead of your friend or me why is it and and
47:26
is it because it's safe does it feel safer? Is
47:29
it because it's more positive? Is it more judge
47:31
like less judgmental? But why like but if you
47:33
know that these tools are not there's nothing
47:35
on the other side and in fact it can actually
47:37
tell you things that are harmful How do we change
47:40
our behavior a third or fourth grader can understand
47:43
that? Then in fifth and sixth and seventh and
47:46
eighth and ninth and 10th, 11th and 12th, you
47:48
just spiral that curriculum, you spiral that
47:50
knowledge and you add, you reinforce, you go
47:54
deeper into the technology and how to use it,
47:56
you start to apply it, you start to gradually
47:58
some responsibility. But third and fourth graders,
48:01
why wouldn't we do that? First of all, I'm gonna
48:03
just give you one more example. You guys know
48:06
I love. This is what happens when you literally
48:09
do the work. I have like 9 ,000 anecdotes, guys.
48:12
In fact, we could do an entire three hours with
48:14
the man of telling you anecdotes. But we do our
48:17
same flagship with young people as adults. And
48:20
so we do bias and image generation. And we do
48:23
a CEO, which is always kind of a nice one because
48:25
it's not. It's funny because it's so bad, guys.
48:28
It's so bad. You've seen it. But what we did
48:30
is. We gave people, we showed them Canva and
48:34
we were at a guideline development in Illinois
48:35
and it was amazing. There were five eighth graders.
48:39
And I come over when the eighth graders, when
48:41
we were exploring Canva, I was trying her best
48:44
to create an unsexist CEO picture with Canva.
48:48
And what she did is it actually created an image
48:51
of a white CEO in the middle of two adoring women
48:56
staring at him like this. And on the top it says,
48:58
super unsexist CEO. literally written out. And
49:04
I, first of all, she did this on her own. She
49:06
was not asked to do this, but talk about that
49:09
seminal piece of learning in that moment. First
49:12
of all, about bias, but about control. Like she
49:16
can't, we don't control and misgenerate us right
49:18
now, the same way we control, kind of control
49:20
large language model outputs like CHSBT. But
49:23
I think that that doesn't, an eighth grader did
49:25
this on her own. You can do that in sixth grade,
49:27
fifth grade, fourth grade. And you know what?
49:29
The best thing about AI literacy today is that
49:32
it's interesting and weird and funny. It won't
49:35
be the same way in five years, and it's definitely
49:38
more engaging than comparing two websites or
49:40
looking for sources. Yeah, I think without a
49:43
doubt, and I've done very similar ones. In fact,
49:45
you kind of reminded me of one that I created
49:49
this image to illustrate kind of somewhat that
49:51
point. But, you know, give me a roomful of teachers
49:55
going through a professional development training.
49:59
And so invariably, all of the, I would say 90
50:03
% of the teachers that they showed in there of
50:05
maybe 25 or so, they were white women, probably
50:11
around 50, and were wearing glasses. And that,
50:17
so, I mean - At least there are women in it.
50:21
think of the data that this is being trained
50:22
on where are they getting that information about
50:24
what a teacher looks like and you know flicker
50:27
whether it's flicker images or they're getting
50:29
threads from reddit or wherever but like I and
50:32
then I got to the point where I had to I had
50:35
to prompt this about five different times to
50:37
get the the results that I wanted to see and
50:41
so I think that that also could be another image
50:44
or another activity with that same group of third
50:48
fourth fifth graders that you're talking about
50:50
where they could be designed to create a diverse
50:52
group, you know? Well, the thing is a family
50:55
body type, like body image. If you do a family
50:58
on a picnic like we did in the course, it's only
51:01
watermelon and like fruit. It's only healthy
51:05
snacks and everyone is really thin. And it's
51:08
mostly like, you know, not to go too deep into
51:11
this, but like it's, it's, there's so many deeper
51:14
like biases. And it is the thing is that Why
51:20
not dig into the places of uneasiness, meaning
51:25
like the things that don't work, because how
51:27
much do you, like when have we ever taught young
51:29
people or ourselves to be critical users of technology?
51:35
We have been the frogs in the pot of boiling
51:37
water where it's free, it's available, it's designed
51:42
to be addictive, whether it's social media, devices,
51:44
et cetera. And we have... over time, given more,
51:48
like who even reads terms of service anymore,
51:50
right? Like, or ever that why not disrupt the
51:54
system and say, hey, like tech is tech. It wants
51:57
to, you know what the goal of tech is for you
51:59
to buy it. And you know, let's be critical users.
52:03
And you know what? I'm going to make a conscious
52:06
decision to have this, to use this in a way that's
52:09
meaningful to me. But I'm going to do it in a
52:11
safe and responsible way. And you know what?
52:13
Maybe sometimes I don't make 100 images, even
52:15
though my friend did. Because you know what?
52:17
That could be environmentally negative. Not nearly
52:20
as environmentally negative as building a new
52:21
model. But you can make those decisions. And
52:26
I think that that's where the learning happens
52:31
at points of friction, not at points of ease.
52:36
And I think that when We have points of friction.
52:40
That's the things kids remember forever. Okay.
52:44
So moving forward, Amanda, the one thing that
52:48
we are curious about is how can, as this is all
52:53
shifting and going on, what would be your recommendation
52:56
for how teachers can stay grounded and what their
53:02
responsibility or the main goal of educating
53:04
students is, as all this is shifting around them,
53:07
what would be your recommendation during all
53:10
this? I think the first thing is it's incredibly
53:14
important to build your own AI literacy. I mean,
53:16
I just think that AI literacy is a foundation.
53:21
You won't even know the right questions to ask
53:24
until you understand what this is or isn't. And
53:26
I think that this is something that is incredibly
53:29
important. And that could be something that is
53:33
our free course. It could be coming to one of
53:35
your sessions. It could be coming to one of our
53:37
webinars or other courses on Google. And I know
53:42
Cloud has, Anthropic has some. There's a lot
53:44
of really good opportunities. But I think that
53:48
AI literacy as your foundation. But I'm going
53:50
to say this is the best time to partner with
53:54
each other and young people. And so that doesn't
53:58
mean go to young people and say, teach me how
53:59
to do this, but we're going to learn together.
54:01
Because again, they're probably going to be showing
54:04
you the worst possible version of this outside
54:06
of the same amount of sophisticated users as
54:09
adults. So the last thing I'm going to say, and
54:12
this is something that I cannot underscore enough,
54:16
transparency and modeling of ethical, safe, and
54:22
effective behaviors is the most important thing
54:24
we need to do today. If you feel like it's cheating
54:28
yourself, you should interrogate, is that because
54:32
it's something maybe you should have done, or
54:34
it's so important, meaning like this is something
54:36
that's a core part of teaching that I shouldn't
54:38
have offloaded? Or is it because of the negative
54:40
rhetoric because we're not talking about it?
54:43
And if it's a former, maybe the same way that
54:46
students have to think about what is actually
54:47
the most important use of this is an important
54:50
conversation for yourself. But if it's a latter,
54:52
which is more common, Then, you know what's the
54:55
best way to do this? Talk to your administration
54:57
about setting clear guidelines. Start to be transparent.
55:01
Show people what you do. Show kids what you do.
55:04
If you're using it to build instructional material
55:07
where kids won't just turn off complete because
55:09
they're like, oh, they use AI. Like, hey, guys,
55:11
we're going to use Canva AI right now to create
55:13
an interactive element because we're struggling.
55:16
I'm going to show you how to do it. We're going
55:17
to vibe code. I'm going to give it feedback.
55:19
You know what? I didn't do a great job of giving
55:22
it specificity, so it's not perfect. But now
55:24
it's good, and we're going to go do it. And then
55:26
we're going to reflect on, did this help? If
55:29
not, OK, then maybe we should have done something
55:31
differently. Or maybe I shouldn't. Could there
55:32
have been a human intervention that was better?
55:35
But think about how rich and amazing that is
55:37
and how much learning can happen instead of,
55:39
hey, everybody, this isn't working, so I'm going
55:41
to keep talking for 30 minutes. or I'm going
55:44
to have you do a worksheet or, you know, like
55:47
what we do when things go wrong, right? How are
55:50
we overcompensating and are redirecting? So I
55:53
think that those types of activities and that
55:56
transparency and that modeling of meaningful,
55:59
effective and safe use is such a gift right now.
56:03
And it's a way to create spaces in which we're
56:07
not hiding use. And I think that that's really,
56:10
really important. Because if a kid comes to you
56:11
and says, hey, Jeremy, I do not have a tutor
56:15
at home, and I am struggling with this concept,
56:18
and I know you're not available at 2 o 'clock
56:20
in the morning when I'm doing work, what should
56:22
I do? Can I use AI? OK, so Jeremy says, yes,
56:25
but what I want you to do is I want you to go
56:27
to Google's study mode because I'm a little bit
56:31
more confident. It's not going to be perfect,
56:33
but I think it'll help you here. And I want you
56:35
to then come to me the next day. We're going
56:37
to interrogate how it went. And you're going
56:39
to use these best practices. Instead of a kid
56:41
either avoiding that AI use because of the fear
56:44
of associating with cheating or using it in a
56:47
way that can potentially harm their learning
56:48
because they're just using ChatBT and ChatBT
56:51
is telling them the completely wrong answer because
56:53
it's not like it's not fit for purpose for the
56:56
thing they need. And I just think, what if it
56:59
could also be based upon the content that you've
57:01
already given to kids that is coming from your
57:04
own voice using something like Notebook LM? Notebook
57:06
LM, a gem, yeah, absolutely. Or if you're fortunate
57:11
enough to have a paid version of Conmigo, any
57:14
of those things would be great. Yeah, because
57:16
the thing is, these kids are doing it with Quizlet
57:19
already, but they're using their own notes that
57:21
probably have deficiencies in them. But if I
57:24
did it where I use Notebook LM and I created
57:27
a beautiful notebook with all, if you don't know
57:30
what Notebook LM does, it essentially can create
57:32
this, this, what they call the wall garden where
57:34
instead of the, the, the, um, Gemini model going
57:37
out into the world and potentially pulling in
57:39
things that weren't very like meaningful or correct,
57:42
instead you're able to put the chapter, like
57:45
if it's not copyrighted, but the chapter of the
57:46
book, your notes and your slide decks. and then
57:50
build a frame where students can ask questions
57:52
directly, get a study guide, flashcards, timelines,
57:57
briefings, then you're controlling the environment
58:00
and it takes like, guys, it takes like 10 minutes.
58:04
And you're limiting the possibility of hallucination
58:07
because it's really based upon just your stuff.
58:10
Nothing eliminates the possibility. Nothing eliminates
58:12
entirely, right? But it makes it less, but it
58:16
also gives you an opportunity to double check
58:19
that the work in a more direct manner going directly
58:22
to the source that you've given. It's still,
58:25
I will say if anybody's using Notebook LLM, first
58:27
of all, no generative value systems do not hallucinate
58:29
or make mistakes, but I will say that if you're
58:33
using audio overviews, that is a place that will
58:36
have the highest hallucination rate at Notebook
58:37
LLM. Good point. So just like I hate to be like
58:40
that person, but I am that person. You guys know.
58:44
But I do think that that is something even the
58:48
kids know. What if they find a realization? Wonderful.
58:50
That's why you always double check. Before we
58:56
get into our lightning round, that's our fun
58:58
questions for the most part. Real quick, is there
59:02
anything you want to promote or highlight that
59:04
you have coming up or that? viewers or listeners
59:08
should know? Well, we've got some webinars coming
59:11
up which will be fun. We've got a kind of a deeper
59:16
prompting one around reasoning models and deep
59:18
research. That's going to be internal. We're
59:21
going to do one on student well -being with the
59:22
Rhythm Project. We're going to do one with Anna
59:25
Mills on using AI for feedback. Not giving, like
59:28
this is, it's not grading. Just want to underline
59:30
that. We are also, new cohorts of our courses
59:34
are out. We have some starting next week in September,
59:37
but also November and January. It's a free student
59:39
course coming out in October. Also a futurist
59:42
piece that we've convened and done about like
59:45
what the future could be with AI and education.
59:47
So lots of good stuff. Also new updates for our
59:49
prompt library. So I'd say like, and also hopefully
59:53
if we can get our stuff together, an update and
59:55
website where it's easier to find all the resources
59:57
we have. So we're hoping that that's before the
59:59
end of the year as well. Great. And we'll be
1:00:02
sure to include some of the links and things
1:00:04
for that in our descriptions. But we're going
1:00:07
to jump into our lightning round before you go.
1:00:10
Jeremy, you want to kick it off? All right. So
1:00:13
we're just kind of to random questions. They
1:00:15
can just be, they might be AI related. They might
1:00:17
not. It might be a mix of everything here. Okay.
1:00:20
And just short blurb answers. Okay. I'm ready.
1:00:26
What book are you currently reading or what's
1:00:28
a book that's kind of shaping your thinking recently?
1:00:32
I read mostly fiction. So I'm going to go with,
1:00:35
I'm reading the new It's a book about vampires.
1:00:39
And then, of course, I can't remember the name
1:00:41
of the book. But it was by the person that wrote
1:00:43
Adding the Roo. It's pretty good. I'm also rereading
1:00:46
Invisible Man by Ralph Ellison. By the way, it's
1:00:52
on my shelf. That's why I pointed, not just to
1:00:54
be weird. OK, yeah. Next to the tepid stuff.
1:00:57
What's the most overrated AI fear you hear from
1:01:00
educators? That it'll take my jobs. Will AI literacy
1:01:10
eventually be a required course in schools or
1:01:13
required content? I hope so. How soon? Not soon.
1:01:19
I mean, I would say PISA's test is 2029. I think
1:01:24
we have a long way to go. I think that things
1:01:26
like California pushing on AI literacy as a graduation
1:01:29
crime, it can help, but we have a, guys, we have
1:01:31
a path. What's your favorite way to recharge
1:01:37
after a long day of workshops or speaking agents
1:01:39
or talking to podcasts, hosts? I live in New
1:01:44
York City, so one of my favorite things to do
1:01:46
is put on my headphones with music and walk.
1:01:50
I love New York, walking in New York City. It's
1:01:54
my favorite thing to do. I think you kind of
1:01:58
answered what the next question was going to
1:01:59
be, but what's your favorite deli item in New
1:02:02
York? My favorite deli item, I'm very bread motivated,
1:02:09
so like bakeries. So like we gotta go bagel,
1:02:13
a good New York bagel. I know we don't have those
1:02:16
in Minnesota, but you can come out and hang out
1:02:18
with me. Sounds good. All right, and then this
1:02:21
is the last question. So if everything's shut
1:02:24
down, the grid goes down, you're completely offline
1:02:28
for a week, what would you do? Spot to go? anything
1:02:33
like that, what would you do in that scenario?
1:02:36
I would go hang out with my 92 -year -old grandmother
1:02:39
who is very grumpy but I love her so much and
1:02:43
we just hang out and I would read and I would
1:02:46
spend time with her. I would exercise and that's
1:02:50
what I would do. Awesome. All right well thanks
1:02:54
again Amanda for your time. We really appreciate
1:02:57
it and Make sure that you, if you're listening
1:03:00
or watching, go check out AI for Education. We'll
1:03:03
have the links and ways to connect with Amanda
1:03:06
below. So thanks everybody and thanks again,
1:03:08
Amanda. Thank you. Thanks for listening to Integrate
1:03:13
This. Want to take your learning further? Head
1:03:16
over to our YouTube channel for quick tutorials,
1:03:19
deep dives, and more ed tech tips you can use
1:03:22
right away. Don't forget to subscribe and catch
1:03:25
new podcast episodes every Thursday on Apple,
1:03:30
Spotify or wherever you get your podcasts. Until
1:03:33
next time, keep exploring, keep innovating and
1:03:36
keep integrating this.