Latest / Integrate This! Ed Tech Podcast / AI, Ethics & the Future of Education | Rebecca Bulstma

Transcript

AI, Ethics & the Future of Education | Rebecca Bulstma · 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. Welcome to the Integrate
0:31
This EdTech podcast. I'm David Berner. As always,
0:34
join my co -host, Jeremy Nicola. And today we're
0:37
very excited to announce a guest that we have
0:39
joining us, AI ethics researcher and keynote
0:43
speaker, Rebecca Boltzmann. We actually were
0:46
passing ships in a recent summer institute that
0:51
took place in Oetono, Minnesota. And so I...
0:55
wanted to reach out to Rebecca. We kind of follow
0:57
each other on LinkedIn and I see a lot of the
1:00
work that you're doing. And so I'll just give
1:03
you a moment to introduce yourself and some of
1:05
the things that you do. But we're very pleased
1:08
to have you join us. Thank you so much. I'm honestly
1:12
thrilled to be here. Such a professional setup.
1:15
I feel borderline famous being on this fancy
1:18
podcast. So thank you for having me. Really all
1:21
I I do, as I talk about generative AI, kind of
1:24
live in this messy space between I use it all
1:27
the time, every new product that comes out I
1:30
experiment with. I can tell you about all the
1:31
features of every large language model and I
1:34
use them all. And then I research artificial
1:37
intelligence ethics at the University of Edinburgh
1:39
in Scotland and really just look at all the ways
1:42
this can go horribly wrong. And then I just try
1:45
and exist in the middle and I talk about things
1:47
all along that pendulum and try and make sense
1:51
of them. in public and try and help other people
1:54
who are trying to make sense of it too. Yeah.
1:57
And one of the things I was going to mention
1:59
too is that, so you're not currently in Canada,
2:04
right? Or are you? I am in Canada. And funny
2:08
story, I had a class in Edinburgh this morning
2:12
at 2 a .m. my time. I have to be there like in
2:16
person, air quotes. And so I had a class at 2
2:19
a .m. and now it's 8 a .m. where I am. So it's
2:23
like absolutely me. Yeah. And so I got to ask,
2:27
what are things like? Is it very similar? Because,
2:31
you know, like I mentioned, you were, you come
2:34
to the States and you've done keynotes here and
2:36
everything. And what is the comparison between
2:40
like what's going on in Canada around these conversations
2:43
and the U .S.? Are you seeing like differences
2:47
in that? I actually do the majority of my work
2:51
in the US. I don't know that there's the same
2:55
level of things happening in Canada yet. We don't
2:58
have the equivalent of what I know some states
3:00
have like Ohio statewide legislation around AI
3:05
and education. EdTech isn't as big of a thing
3:08
up in Canada as it is in the US. Right now, people
3:13
are just left to their own figuring it out. Individual
3:16
schools are doing PD up here. I would say nobody
3:21
in like my immediate life knows much about generative
3:24
AI, cares much about generative AI. And it's
3:26
not a conversation that's happening a ton, at
3:28
least in the spaces that I'm a part of. Okay.
3:32
Yeah. Well, maybe, and maybe that'll continue
3:34
to expand as, as you move forward, but yeah.
3:38
Yeah. Um, so when, when we think about kind of
3:40
like your background here, right? Um, your background
3:43
isn't necessarily coming from a K -12 space.
3:47
Um, but yet you're, you're definitely, I think
3:51
out there at least not just in the technology
3:54
side of things, the AI side of things, but also
3:57
I think on the communication side of things,
3:59
right? Um, you know, can you describe how you
4:02
can take your background and experience and,
4:05
you know, give it that educational, that ed tech
4:08
type spin. Yeah, it's kind of an interesting,
4:11
it's an interesting backstory. And it's an interesting
4:14
like conglomerate of all these different fields.
4:16
But my background is in communications and public
4:18
relations. But actually, for the last eight years,
4:21
I worked in a K to 12 district running there
4:24
at director of comms. And so I worked directly
4:27
with the superintendent and the executive leadership
4:30
team in all the meetings, helping make all the
4:32
decisions. And so I really had a really good
4:35
firsthand look and understanding of how things
4:38
work on the back end, how leadership works, how
4:40
they make decisions about educational technology,
4:43
and really saw kind of the value of like public
4:48
engagement and making things easy for people
4:51
to understand. As a communicator, my whole life,
4:53
I've just tried to take things that are complex
4:56
and distill them down for an audience and make
4:57
them easy to understand. When Generative AI came
5:00
along, I was trying to understand it right when
5:03
ChatGPT came out. I couldn't find anyone who
5:06
could explain it to me. The only people who understood
5:08
it were data science guys who would just start
5:12
talking about transformer architecture. and deep
5:16
learning and neural networks. And I couldn't
5:18
understand it. And I couldn't find anyone to
5:20
explain it to me. So I started taking these complex
5:24
concepts around AI and how it worked, reading
5:27
about them, and then just trying to make them
5:29
easy to understand and put them in my own words
5:31
and then share them on my LinkedIn to maybe help
5:35
other people understand it. I started doing that
5:37
at my district, too. and seeing what landed and
5:40
what didn't. And then from there, I started speaking
5:43
at like K -12 communicators conferences. And
5:46
then I think superintendents were probably like,
5:48
hey, do you know anyone who knows about AI? And
5:50
their comms people who are usually like the right
5:52
hand people were like, yeah, it just kind of
5:54
ended up leading altogether. And the more I knew,
5:57
the more freaked out I got, which is how I ended
5:59
up on the research end of things. So that's kind
6:01
of the high level of the story. And one of the
6:06
things too that you emphasize is the human centric
6:09
piece to this. I think like, um, even in our,
6:13
you know, with our, some of our work and the,
6:16
our Minnesota department of education, putting
6:19
out some, they put out some guiding principles
6:21
and one of the main ones there was, uh, human
6:24
centered. So, you know, focusing on that in your,
6:27
in your work and the things that you're doing,
6:30
uh, can you talk a bit about what that means?
6:34
to you like, uh, the human centric concept and
6:37
how that differs from just using AI tools because
6:41
this, these are conversations that we've had
6:43
a lot where, um, it happened so quickly that
6:47
there are all these tools that are already available
6:50
for teachers. Um, but what does that mean to
6:54
you when someone says like human centric AI?
6:58
I see a couple of different directions with that.
7:00
I think we need to think about it like that human
7:03
in the loop, human -centric, because generative
7:07
AI, like large language models, they hallucinate,
7:10
they make mistakes, they don't get things right,
7:12
which means we need to have the human eyes on
7:15
it before things go out in the world to make
7:17
sure that we're not putting something out that
7:20
reflects like a really biased or outdated stereotype.
7:24
We know that it was trained on the Internet,
7:26
so It was trained on a data set, the history
7:30
of humanity that doesn't necessarily reflect
7:33
how we think today. It would maybe give you outputs
7:37
that reflected very traditional families, or
7:39
gender roles, or cultural stereotypes. So human
7:42
in the loop and human centric make sure that
7:47
there's actual context on the AI stuff that might
7:50
be wrong. On the other side, I think the human
7:53
centric pieces important just from like a future
7:57
ethical point of view of making sure we don't
7:59
make ourselves irrelevant. Like we have to figure
8:01
out how to work with this technology and incorporate
8:04
it into what we're already doing so that we don't
8:06
risk just putting ourselves out of work or making
8:11
ourselves obsolete. So I think if we learn to
8:13
work with it, with humans at the center and using
8:16
it to enhance and augment us, that kind of human
8:19
centric matters on that side, but then also for
8:22
accuracy and and making sure that the outputs
8:25
make sense and aren't causing harm. Both sides
8:29
have to do with harm, but that's a couple of
8:31
the reasons I think that human -centric, human
8:33
-in -the -loop is so important. You know, there's
8:36
this video that I show in this one AI presentation
8:39
that I give, and I think it's touching on kind
8:41
of what you're saying here. And it is not necessarily
8:46
AI directly, right? But it's... Should we be
8:49
turning over to technology things that are, should
8:53
be empathetically human? Um, and in the video,
8:57
it's these two kids, um, that, uh, you know,
9:01
they're, I think their mom works in the, the,
9:03
the local nursing home and they have these little,
9:06
um, they're like cat robots, um, that purr and
9:10
stuff when you pet them. Right. And so they went
9:13
around and they raised some money and they were
9:15
able to buy two of these cat robots and they
9:18
bring them into the nursing home and the residents
9:20
there just love them because they can pet the
9:24
cat and it looks like it's interacting with them
9:26
and it's, it's kind of helping making them feel,
9:29
you know, um, I guess just more comfortable.
9:35
I don't, I'm not exactly sure, but Um, but then,
9:38
you know, and, and I present this video and I
9:40
share it and everyone just thinks it's so cute
9:42
and lovely that these cats robots, uh, are, are
9:46
making these residents feel better. And I'm just
9:48
thinking, yes, that's great. These kids are super
9:50
innovative and. You know, hats off to them for
9:53
showing like design thinking and the process
9:56
of trying to solve a problem, you know, but at
9:59
the same time, isn't that more naturally a role
10:02
that a human could play? in offering that comfort
10:06
and support to those people in need. And I see
10:09
that as, then I present that as a moment and
10:11
people are like stopping and thinking like, oh,
10:14
just because AI can do something for us doesn't
10:17
mean it always should. Yeah, that's a great point.
10:21
And actually I'm just building a keynote right
10:23
now and I forgot that that was in Jurassic Park.
10:26
I don't know if you guys remember, there was
10:28
actually a quote in Jurassic Park where they
10:31
- Like the first one. Yeah. Yeah. I would quickly
10:34
say like just because we can do something doesn't
10:37
mean that we should, but I have actually mixed
10:39
feelings on this because I think about it a lot.
10:41
And I think about the, a lot of the harm that
10:44
came about from the pandemic and people being
10:46
lonely and people ending up with even more mental
10:50
health issues than normal in a system that can't
10:54
necessarily support everybody. There's wait lists
10:57
for therapists. people who need care and can't
11:01
get it. And so part of me wonders, is it better
11:03
than nothing if people feel like they can get
11:06
some sort of emotional connection, even if it
11:08
is artificial? Is it better than nothing until
11:13
you can get a human? Does it demonstrate to us
11:17
how necessarily important that human piece is
11:21
and how we'll never really fully replace humans?
11:24
Because we would probably all rather talk to
11:26
a real human expert You have to make sure that
11:30
this tool, if people are using it out of desperation,
11:33
isn't causing harm, which we're seeing some people
11:36
who are using it as a therapist, it leading them
11:39
way down the wrong path. So I think there's a
11:42
lot of things to be thinking about there. And
11:45
then also something else I've been thinking about
11:47
is if you have this close relationship, let's
11:50
say as a therapist or as a friend with a large
11:52
language model, which isn't as weird as it sounds,
11:54
like I really get into conversations with about
11:58
different topics. Lots of people do with their
11:59
large language models now. It's actually the
12:01
number one use case is advice and therapy for
12:05
large language models. But as we kind of go down
12:08
that path, we become less tolerant of like, say,
12:11
your best friend or your sister who has bad days
12:13
or off days or struggles or isn't always available
12:17
to you. Do we just automatically turn to what's
12:19
easy, even if it's artificial, instead of what's
12:22
meaningful and human and what actually builds
12:25
connection? So that's Again, one of those tensions
12:28
that we see, right? Right. Kind of the, you know,
12:32
when it's better than the best human available
12:34
to you at that time kind of concept. And what's
12:37
best, right? Like is it better and best because
12:41
it's always available and endless? Or if you
12:45
can't have real meaningful dialogue with it,
12:48
which a lot of the research tells us it's not
12:50
actual meaningful dialogue unless it's two way
12:53
and both people are getting something out of
12:55
it, then How helpful is it and is it better than
12:58
nothing? Sorry, you're talking to an ethicist.
13:00
I see every side of everything and it's kind
13:02
of abstract. We're trying to get you to commit
13:05
to one particular thing you hear. Come on. I
13:09
don't think I've ever done that in my whole life.
13:10
Yeah, yeah. Well, I think you've kind of started
13:14
to touch on some of these here as well, but like.
13:17
maybe there's some room to expand upon. Where
13:19
do you see some of the biggest opportunities
13:23
in education technology for the integration of
13:27
AI? So one of the things I think about a lot
13:32
is in business and in AI, we do what's called
13:36
foresight and forecasting. You've heard of business
13:38
forecasting where you kind of predict what's
13:40
going to happen. And a lot of those frameworks
13:42
have been applied to AI. And one of the visuals
13:46
that I like is a futures cone. And you kind of
13:49
picture it like, oops, sorry, a cheerleader,
13:51
you know, one of those cheerleader cones that
13:54
they yell through when they're trying to get
13:56
the crowd hyped up. But it shows that from any
13:58
point in time, there are endless numbers of possibilities
14:01
that are plausible, that could potentially happen,
14:04
that are possible, that are things maybe we wouldn't
14:08
like, things that would be optimal. And so I
14:11
think you have to think through every possible,
14:14
like it could go here, it could go here, it could
14:16
go here. And knowing that it will probably land
14:19
somewhere in the middle and I think about things
14:23
like that a lot. I think we get to choose a lot
14:26
of what happens and where those end up, which
14:28
I like, but I don't think there's a lot of active
14:31
choosing happening right now. I think that everything's
14:33
just kind of happening so fast that it's happening
14:37
to us and we're not making conscious choices
14:39
about it. So the future of education will really
14:43
depend on the decisions that we're making about
14:45
it right now that lawmakers are making about
14:47
what's allowed and not allowed, how damaging
14:49
it is to students, either banning the technology,
14:54
right? You're not preparing it for a workforce
14:56
where they're going to be using it forever or
14:59
not training educators how to use it properly.
15:02
There's a lot of data security risks, but the
15:04
future of education I think is going to look
15:06
very, very different than the way we understand
15:08
it now and the way that it's been for the last
15:10
200 years, which I don't think is a bad thing.
15:12
I know there's a lot of models on both sides
15:14
from ultra traditional to now there's like the
15:18
Alpha school that basically has these learning
15:20
guides and they have kids working with AI to
15:24
deliver them the banking model of education,
15:28
which is, I am going to make deposits into your
15:30
knowledge bank and it is your job to accept them
15:32
and put them into your brain. And then the rest
15:35
of the time they get to explore things that they're
15:37
actually interested in. And I like a lot of things
15:40
about that, but I also think that that human
15:44
teacher element is super important. that's been
15:47
demonstrated through research, that relationality
15:49
piece has got to be critical. So I hope it ends
15:51
up somewhere in the middle in a way that benefits
15:54
more kids than I think the traditional education
15:56
is benefiting right now. So teachers aren't afraid
16:01
or they don't need to be as afraid of losing
16:03
their jobs to AI because that human part is still
16:07
necessary. But they're going to need to accept
16:10
that the way they do their job is changing. I
16:13
think to a point, I think what matters about
16:16
teaching is going to stay the same. If you're
16:18
a teacher because you love kids and you love
16:20
helping them learn and grow and see endless possibilities
16:23
for their futures and think bigger and differently
16:25
about things, that's not going to change. If
16:30
you love standing in front of a class and lecturing
16:33
and assigning essays, you might not love teaching
16:37
in five to 10 years. It might not look like that.
16:41
teachers and the educators and all professionals
16:43
that need to worry about AI taking their jobs
16:45
are the ones who refuse to acknowledge AI, decide
16:49
they're not going to learn it, decide it doesn't
16:51
matter, and decide that it's bad and it's evil
16:56
and it's a fad. I think if you're thinking along
16:59
the lines of any of those things, you might have
17:01
to worry about your job in five to 10 years because
17:05
you're not going to be able to deliver education
17:07
in the way that it's heading. in the same way
17:09
that will be effective for students. If you're
17:10
hell -bent on always assigning the same essay
17:13
you've assigned for 15 years in this exact way,
17:16
you're going to be very disappointed when kids
17:18
are using AI to generate it because every single
17:20
other teacher permits it in their class and kids
17:24
won't conform to the old way of thinking in education.
17:28
It's just not going to work. Yeah, it's a great
17:31
point too because there is some elements where,
17:34
like you're kind of saying that teachers have
17:37
to be willing to adjust to some extent because
17:40
of the reality that their students are using
17:43
these. And I like what you said there about like
17:46
if you're doing the same thing that you've done
17:48
for the last 15 years, it's time to maybe rethink
17:51
different approaches of what that could look
17:54
like and not just trudging down the same road,
17:58
yielding zero or negative results. And that kind
18:04
of touches on you know, the future a bit with
18:08
what's kind of coming and what teachers should
18:12
be thinking about. If you were to think about
18:16
the current risks that are happening right now
18:19
with like ethical considerations, what would
18:22
be your top two to three things? Like if administration,
18:27
if they're trying to tackle AI use in their school
18:30
with their teachers and you know or just their
18:34
staff in general and with students what are what
18:36
in your mind are the top two to three things
18:39
they need to ask themselves first with what is
18:42
currently going on this year? I think the idea
18:47
of like risks and AI ethics gets super overwhelming
18:51
because there are so many and they're all scattered
18:54
like swirling all over the place and you just
18:56
have to kind of reach up into the cloud and pull
18:57
one down and there's so many. And so I've started
19:00
thinking about it in terms of like three buckets,
19:03
right? Like we have the big bucket that is things
19:06
that you should probably be aware of, but you
19:08
can't change. Like we can't change how these
19:11
models were trained. They give biased output.
19:14
We can't change right now directly that they're
19:17
harming the environment. These are big ethical
19:19
issues, right? Miss and disinformation crisis.
19:21
These are big ethical issues. We can't control.
19:24
We should probably be aware of them. We're going
19:25
to put those in that bucket. There's a bucket
19:28
that's as an individual, your ethical AI use
19:31
of these tools as an individual, what you can
19:33
control, right? Checking your outputs. how you
19:38
think about these tools, what you let them do
19:40
for you, how transparent you are in using them.
19:42
And then in the middle, there's the ethical considerations
19:45
for organizations and things that we maybe influence,
19:50
but we don't directly own, but are still within
19:52
our control. So those issues for like a school
19:56
leader would be things like making sure, well,
20:00
I think a big one that's been top of mind. for
20:02
me lately is creating opportunities for your
20:04
staff to learn and understand, making it safe
20:07
for them to fail, incentivizing innovation, right?
20:11
Saying, you know what, we're going to learn this
20:14
together about generative AI. Every Friday for
20:17
an hour, we're going to be time to do this. We're
20:19
actually going to have a contest or some sort
20:22
of way that we can share the great tools and
20:24
the ways that we're coming up with this and the
20:26
people who share the most often or you know incentivizing
20:29
people and creating a culture of sharing and
20:32
openness that gives people permission to fail
20:34
I think is really underrated because a lot of
20:38
people are scared because a lot of us grew up
20:39
in cultures where you have to be right and there's
20:43
no real opportunity or room for failure you feel
20:47
all this pressure and so people are just paralyzed
20:50
and they don't even want to start so creating
20:51
safe spaces like that for conversation I think
20:55
will be a big ethical direction that leaders
20:59
can actually control and set. You can't control
21:02
if the state is going to put in curriculum for
21:04
generative AI or what your funding for that looks
21:07
like. But what you can control is helping to
21:10
empower people however you decide to do it, especially
21:13
your teachers. I think that's the most important
21:15
conversation to have right now. Yes, kids are
21:17
going to be using it. But until the educators
21:20
and the leaders understand it, they can't teach
21:22
kids how to use it. They can't police it properly.
21:25
That needs to be the top priority, is helping
21:28
educators and leaders figure this out, experiment
21:31
with it, feel safe, add, fail, and having some
21:35
flexible guidelines or guiding principles in
21:38
place with a few kind of hardcore -ish policies
21:42
about what to do and not do with student data,
21:44
which already exist. But I think that's probably
21:47
the next best step forward. Do you think that
21:51
because of the the nature of how this was released
21:54
and the AI tools being so much like EdTech tools
21:58
of the past. Do you think that's why a lot of
22:02
people, a lot of teachers were thinking, oh,
22:04
just block it. You think it's because they were
22:07
equating that to EdTech tools of the past? And
22:11
this is very much, this is very different. Possibly.
22:15
Dr. Sarah Eaton is a researcher, like a professor
22:18
at the University of Calgary and she talks about
22:20
this kind of cycle within education where a new
22:23
technology emerges and everybody's excited and
22:26
scared by it and then they adopt it, they figure
22:30
out how to use it, kids and teachers put in guardrails,
22:35
kids figure out how to cheat and then a new technology
22:37
comes in and it's just this cycle. But I think
22:40
we've seen the exact same thing happen every
22:42
time, right? With calculators, with the internet,
22:46
probably with encyclopedias back in the day.
22:48
Oh no, everyone can have an encyclopedia at home.
22:51
Now they are going to cheat. So I think it's
22:55
just been how it's always been. But all of those
22:57
things, we were able to, they couldn't do your
23:00
work for you necessarily, right? And we developed
23:03
tools to kind of catch them. This feels different,
23:07
at least to me. because it actually can substitute
23:12
for thinking. We can cognitively offload as teachers,
23:17
as professionals, as students to these actual
23:20
critical processes. So with calculators, you
23:24
still had to know what numbers to put in, in
23:26
what order, right? With this, it's harder to
23:30
show your thinking. So I think... This is different
23:32
and we probably have reacted the way we've always
23:34
reacted. It's human nature. We don't like change.
23:36
We want things to stay as they are because we
23:38
think how things are is the best way. And you're
23:42
right. The fact that this drops available to
23:45
society all on one day without an instructor
23:48
manual for free is different than anything we've
23:51
ever seen. I lived in the country when the internet
23:53
came out. And so we didn't get it for a long
23:55
time. And when we did, it was dial up. And, you
23:58
know, lots of people had the internet before
24:00
I did. So it rolled out. We've never seen something
24:04
happen this quick on this scale that is this
24:07
powerful and this much of a threat to education
24:10
as we know it, in my opinion. Well, there's some
24:13
things that you mentioned in there that I feel
24:15
like I want to unpack a little bit here. And
24:17
you brought up so many great points. I'm not
24:20
sure where I want to start with it, but I think
24:23
one of them here. is you bring up cognitive offloading
24:27
and it is definitely something David and I have
24:29
talked quite a bit about. And I think it's almost
24:33
human nature in a way to try to find that easy
24:35
button. And I think it's easy for students to
24:38
do that. But I think it's also can be easy for
24:40
teachers to be able to do that, too. Oh, you
24:43
know, I want to revamp this lesson plan because
24:46
I just feel like it's time to give it a fresh
24:48
start. I'm just going to turn that all over to
24:50
AI and I'm just going to take it and use it,
24:53
even though we all know that wouldn't be best
24:55
practice. But yet, yet again, we know it's out
24:59
there and we know that it's happening. I think
25:01
it brings in this idea, too, that you have these
25:05
EdTech companies, you know, just to name a few,
25:08
maybe like a school AI, a magic school, CurePod,
25:12
and a whole bunch of others that I think are
25:14
out there for really good reasons. And I think
25:17
they are trying to be on that cutting edge of,
25:22
you know, hey, we want you to use these tools,
25:25
but we want you to use them. In a safe manner
25:27
with students, we want you to use them in an
25:30
ethical way. Um, but yet at the same time, they're
25:34
giving us the easy button to push. So why would
25:37
we just not push it? Right. Um, but at the same
25:41
time, you know, you talk about being an ethicist
25:43
and you can see both sides of it at the same
25:45
time. So yes, I see the easy button side, but
25:47
I also see all of the ways that AI could be used
25:52
to enhance critical thinking, to take an idea
25:56
that you have and take it to that next level.
26:00
And I think definitely there are times where,
26:04
and I'm gonna pick on English teachers here for
26:07
just a moment or two, but you've assigned that
26:11
same essay like you talked about for the last
26:13
15 years. Well. Yeah, AI can help you write a
26:16
better one, and students can cheat that way,
26:19
but students have, you know, forever have had
26:22
older siblings, and they can go find that same
26:25
essay their older sibling wrote and use that
26:27
as a way to cheat. So I don't see it as a process
26:31
of eliminating cheating, but can we find ways
26:34
to create better assessments that are going to
26:39
make students have to prove that they still had
26:42
to think their way through this problem? That's
26:44
the number one question, really, because when
26:47
you peel back all the layers, it's not necessarily,
26:51
or it shouldn't be, about the essay. It should
26:53
be what you want students to gain during that
26:55
process of writing the essay, those skills you
26:57
want them to develop. And you need to be able
26:59
to know that they understand what they're talking
27:01
about. And like full disclosure, like I am...
27:04
not a teacher. I educate people about AI. I am
27:09
a student, so I use AI in a lot of those really,
27:11
really critical ways that you've mentioned to
27:13
help me think differently about research or make
27:15
connections between random things or test the
27:19
limits of my understanding because I know how
27:22
to leverage the tool in that way, but not very
27:24
many people do. I have mixed feelings about ed
27:27
tech because anything that draws money away from
27:29
education that could potentially stay in education,
27:34
preferable for me. But yeah, your question about
27:38
assessments is good. One of the first things
27:40
I encourage when I'm training schools or teachers
27:43
to do is to give their assessment to AI and tell
27:47
them all the ways that a student could use generative
27:50
AI to cheat on the assignment and help them come
27:52
up with different ways to assess the core skills
27:57
that they're looking to assess that might be
27:59
AI -proof. I like to think this is kind of off
28:06
beat, but I really think that we are going to
28:09
probably evolve more towards some sort of an
28:12
oral society. Again, I think we've over -corrected
28:14
so far to texting and email. Now that large language
28:20
models can imitate that so well, I wonder if
28:23
the only way I'm going to know that my student
28:25
understands the concepts critically is if they
28:30
explain it to me and support with evidence and
28:32
use rhetoric and Socratic styles and engage with
28:35
each other through that reflective learning process
28:38
verbally, because you know that if it's coming
28:41
out of their mouth, it's coming out of their
28:42
brain and it's not coming from anywhere else.
28:45
So I wonder if we're going to see more of that
28:48
in the future. I could just see Socrates coming
28:51
back in that Bill and Ted's time machine phone
28:55
booth and saying how great it would be if everyone
28:58
just verbalized everything because wasn't he
29:00
the one that was like didn't thought putting
29:02
in writing? You like a way of taking your ideas
29:05
away from you so You know old things always become
29:09
new again things come back around and I just
29:11
I have to wonder and you know teachers automatically
29:14
go to well How am I gonna I don't have time to
29:16
listen to every you know, but if you don't have
29:18
all of the other things to worry about, let's
29:21
say, and your only job is to sit and engage in
29:24
dialogue and listening with students. And that's
29:28
all you had to do as a teacher and help redirect
29:30
and help them understand. Wouldn't that be fun?
29:33
Wouldn't that be amazing? Wouldn't that face
29:34
to face all the time if that was your big job?
29:37
I think that would be amazing. I think most teachers
29:39
would think that was amazing. I don't think anyone
29:40
loves lesson planning or writing, marking papers
29:44
or doing all that administrative stuff. So maybe
29:47
Maybe that's where we end up and maybe people
29:49
like it better. Yeah. And early on too, I know
29:52
some of our conversations with teachers were
29:55
when, when it was, uh, you know, first released
29:58
and we had all the conversations about the plagiarism
30:01
and kids turning in their paper, these papers
30:03
that weren't theirs. We talked a lot about like,
30:06
well, how can you adjust or extend your assessment
30:10
so that it, it actually challenges the student
30:13
to What is it that you know about this topic?
30:16
So like if they turn in a paper, um, do you,
30:19
is there follow up questions that you, you know,
30:22
hand them maybe on a sheet of paper that they
30:23
have to write out? Um, or like you said, is there,
30:27
is it like an oral assessment that follows the
30:29
work that was done? Um, and I think it does,
30:33
uh, it's a great point that it does, it does,
30:36
uh, give us opportunities to talk to students
30:39
more as opposed to just like your papers do turn
30:42
it in. I'll review it. give it back, you know,
30:45
like it kind of maybe will eliminate that and
30:48
have more conversations around what students
30:50
actually know and the student or the teacher
30:53
can engage them in that more. So it's really
30:55
interesting. And then that relationship piece
30:58
becomes more important too, right? And kids are
31:01
learning how to interact with each other and
31:03
with adults and articulate their ideas. And maybe
31:06
they're not scared to call and make doctor's
31:07
appointments or, you know, make an actual phone
31:11
call. Because they're so used to being able to
31:13
draft a text and have this rehearsal method on
31:17
everything they do that, you know, that real
31:19
life interaction is so scary. And I can say this
31:21
with authority because I have four kids in college.
31:24
So I've seen firsthand, you know, this kind of
31:27
transition to just hiding behind a screen. And
31:29
I just don't know if it's going to work anymore,
31:31
at least not in education. Yeah. If he, so can
31:35
you talk a little bit about, there's been a,
31:37
a huge. Number of frameworks and things that
31:41
have come out. What do you, do you see like,
31:45
uh, what's your full opinion on prompting frameworks
31:49
and things of that nature? Cause Jeremy and I
31:52
have kind of gone back and forth in our own conversations
31:54
about our preferred frameworks that we use or
31:59
train teachers to use, or like even Jeremy has
32:04
brought it before, like moving away from the
32:06
frameworks and that is a popular approach too.
32:10
When you're looking at all these or you know
32:12
the things that come across your desk, what's
32:15
your opinion on the frameworks and is there one
32:18
that is best or what's the best approach? When
32:22
I teach people about how to talk to AI or effective
32:26
ways to interact with AI, I give them kind of
32:28
four approaches and really it's going to work
32:30
differently for everybody. It's like It's the
32:33
reason they make cars in different colors, right?
32:35
Like you don't have cars in different colors,
32:37
they all do the same thing, but everyone likes
32:39
something a little bit different. I do think
32:41
there, especially when people are starting out,
32:44
some people really like a framework, right? Like
32:46
they just really are like, give me the recipe
32:48
so I at least feel like I know what I'm doing.
32:51
And probably that's what you guys did to start.
32:53
That's probably what I did to start. And then
32:54
from there, you feel more comfortable kind of
32:56
going off script. So the framework that I talk
32:59
about the most or that I recommend is actually
33:01
from Google. It's very standard. It's so easy
33:05
to come up with an acronym that does basically
33:07
the same thing. But really, you just need something
33:10
that gives it a task. Tell it what to do. Give
33:14
it a persona. That's the most critical part,
33:16
at least right now. Make it an expert in pedagogical
33:20
reasoning, the best in the world, award -winning,
33:24
lasagna chef, whatever it is. and then tell what
33:27
format you want the output in. I can't even pull
33:30
the other one off the top of my head task. Persona,
33:34
format, whatever the other Google one is, they're
33:38
all basically about the same. I have another
33:40
one for a teacher specifically that's clear,
33:43
give it lots of context, tell it what length
33:45
you want the response in, give it a bunch of
33:48
examples to choose from. give it the action of
33:50
what you want it to do, and then give it a roll.
33:52
So if you want to check all those boxes, that's
33:55
number one. That's a good way to do it. A second
33:57
one is just vibe prompting, where you just seriously
34:01
just put on your voice thing or just type and
34:04
say, listen, I have to do this thing. I don't
34:05
know where to start. How can you help me? Yeah,
34:07
but I tried that and this didn't work. And you
34:08
just make it a conversation back and forth like
34:10
you're talking to a friend, just going off the
34:12
vibes, which works too. Christopher Penn, who
34:15
is like a mentor of mine when it comes to AI,
34:18
actually has a great prompt that forces the AI
34:22
to ask you questions to get what you want. So
34:25
if you just say, I want to do this thing, I need
34:27
you to ask me questions one at a time until you
34:29
have enough information to build it for me or
34:32
to write a really great prompt. So that AI led
34:35
prompting. And then finally, there are some amazing
34:38
GPTs right within chat GPT. I have a favorite,
34:42
it's called prompt maker, where you just say,
34:45
I need a prompt that will do this. And it will
34:48
write you a fully detailed optimized prompt that
34:52
follows all the frameworks that has tons of context.
34:55
And all you have to do is copy and paste that
34:56
prompt and throw it in. So really you can use
34:58
a framework, you can go off vibes, you can do
35:01
AI led. prompting where it helps you figure out
35:04
step -by -step what to do where you can use a
35:05
GPT to write the prompt for you. I usually use
35:07
a GPT to write a prompt for me just because it
35:12
does a really, really good job and it knows how
35:14
to optimize for itself, if that makes sense.
35:20
Yeah, absolutely. And I've kind of gone around
35:23
in circles on that kind of idea too. turning
35:26
more and more to making it more of that conversation.
35:31
And I think that's been super helpful for a lot
35:33
of things that I've done. And I think the only
35:37
thing for me with that part of it is turning
35:39
the microphone on and listening to myself talk
35:42
to myself. I get kind of self -conscious, especially
35:45
if there's other people around. box on full volume
35:53
with no headphones or just have private conversations
35:55
walking on the subway without caring who hears.
35:59
Yeah, that's not me. But I'm trying to, you know,
36:03
I've got goals. I've got things that I'm trying
36:05
to work. OK, let's 50 or 35, 36 minutes here.
36:17
I lost where we are, David. Here's, here's why
36:21
we record and then we can cut little parts out.
36:25
Works for me. I was muted. Sorry. Um, yeah. So
36:27
the next one is the insights from the keynote
36:29
stage. Um, okay. Perfect. And you can, if you,
36:33
if we want to shift, I know a lot of your work,
36:36
you've been doing AI, but you can either do it
36:39
generally speaking, just the, cause it's the
36:43
question is you're constantly taking the pulse
36:45
of the education community. With the the keynotes
36:49
and things that you're doing and it's about fears
36:51
and misconceptions about AI But I didn't know
36:54
if if I mean you guys wanted to take it like
36:57
brought more broadly either way whatever you
36:59
want Okay, so I'll frame this question here somehow
37:03
and and then feel free So, Rebecca, as someone
37:08
who gets the opportunity to travel, to be a speaker,
37:12
a presenter, a keynoter at many different gatherings
37:16
of educators within the education community,
37:20
or even as a host of your own podcast, you're
37:26
taking the pulse of education wherever you go
37:28
and you're interacting with those people. What
37:31
do you think are some of the most common fears,
37:35
most common misconceptions that you hear from
37:38
people in your audience, and what do you do to
37:41
try to address that? There's definitely no shortage
37:44
of fear and misconceptions. That's kind of the
37:47
nature of this technology that came out so fast
37:49
and is moving so fast that people don't understand.
37:52
I think the biggest kind of sentiment that I
37:55
hear is that just people are super overwhelmed.
37:57
They don't know where to start. It's just this
38:00
information saturation. They're starting to be
38:03
like, oh, this matters. This is a big deal. Where
38:05
do I start? I'm already so busy. I'm overwhelmed.
38:11
My administration or whoever has told me that
38:14
this is cheating, so now I feel a certain way
38:16
about it going in. or I'm not allowed to use
38:20
it, but I feel like I should be learning it,
38:23
or people just feel really uncertain and overwhelmed.
38:25
That's a really common sentiment. They know it
38:28
matters. They just don't know where to start
38:30
necessarily. Or they just don't understand what's
38:33
in it for me. Like, yeah, okay, cool. Everyone's
38:35
talking about this. It's bright and shiny, but
38:37
I just don't think it's for me. I don't think
38:39
it can help me. It doesn't seem useful. I don't
38:41
need something that will write me funny birthday
38:43
poems. I think the best way to kind of address
38:48
some of those things is I do think there is one,
38:51
there's got to be like a hook, right? Like there's
38:53
always something that there probably was for
38:56
you guys too that made you be like, oh, that
38:58
could actually be really helpful for me. And
39:01
once you figure that out, you started digging
39:02
in deeper again to figure out how much that could
39:05
help you. It could help you in this way now.
39:08
And so I think really the only way to kind of
39:12
get comfortable with using generative AIs to
39:15
just kind of start playing with it a little bit.
39:17
And if you need a place to start, the good news
39:19
is all of the main labs, like there's no shortage
39:23
of free resources. And I know there's like, here's
39:26
a free course, or a course on 20 ways to prompt
39:29
AI, or here's a prompt library of a thousand
39:31
prompts, or really just go to OpenAI, Anthropics,
39:37
Claude, Google Gemini, they all have AI Basic
39:41
101 free classes that you can use to help you
39:46
get started, or just get a free chat GPT account.
39:51
Don't sign up with your school account, keep
39:52
it totally separate, keep it personal, personal,
39:56
work, work for a lot of reasons. But then just
39:59
start experimenting with whatever you're working
40:01
on right then or whatever you're passionate about.
40:03
Say you're the president of your local a motion
40:07
club or your model car club. Just go and start
40:11
having a conversation with Chad GPT, asking questions
40:14
about it or what you should do or planning an
40:16
event, see where it fails, see where you know
40:19
more and you can tell that it's wrong and just
40:21
start dipping your toe in a little bit. Or if
40:24
you're at a school, ask some of the kids about
40:27
it. Like have a kid in your life show it to you
40:30
and teach it to you so you can understand how
40:32
it works from. their point of view and their
40:35
lens because really a good practice is having
40:39
learning from kids right like having that sort
40:41
of be an interactive learning process but also
40:44
it's a good opportunity to get into their minds
40:46
about it because they understand it they're using
40:48
it like they get it but you just have to kind
40:52
of understand it a little bit better and be vulnerable
40:54
which is hard for a lot of teachers and this
40:57
is an interesting space that i'm exploring right
40:59
now because The whole transformative learning
41:02
process starts with this disorienting dilemma.
41:06
It's this moment in time where suddenly nothing
41:07
is the same anymore and you're struggling and
41:10
you're in this messy in -between space where
41:12
you thought you knew everything and now suddenly
41:14
you realize you don't know and it's an unsettling
41:17
place to be. A lot of us as adults haven't been
41:20
there in a really long time and I think it's
41:24
a good opportunity to remember that this is how
41:25
students feel all the time. and it's okay to
41:28
sit in the unknown for a bit and to not know
41:31
everything and to be open to being a learner
41:33
and taking the time to experiment and explore
41:36
with how this can help you specifically. Even
41:39
like putting on your own oxygen mask before you
41:41
can even help students with it. You just need
41:43
to invest some time in understanding it for yourself.
41:48
I think that's a really good point. I think if
41:50
you want true meaningful learning to take place,
41:55
that learner has to experience some sort of discomfort
41:57
and be in that area where they do feel the chaos
42:03
going on around them. And now they have to try
42:06
to figure it out. And I think that's where I
42:10
think, like you said, I think many educators
42:12
are feeling at this time of AI is that they haven't
42:17
lived in that uncomfortable space. for a period
42:20
of time, they aren't necessarily being empathetic
42:23
with the students that they're with and that
42:25
this is kind of what they're feeling, you know,
42:27
six, seven hours a day as they travel to all
42:29
of these different classes that they're being
42:32
asked to take. So that I really that that connected
42:36
with me. It resonated. And one of the things,
42:41
too, that we're seeing more and more is the kind
42:45
of. now that the tools are out there and teachers
42:48
are starting to use them, there becomes a question
42:51
about assessments. What is your thoughts on some
42:54
of the things that have come out or that you've
42:58
seen? In some of our conversations, and we had
43:02
a chance to interview Leon Furze, who is a researcher
43:05
out of Australia, and there are some researchers
43:09
and folks that are developing assessment scales
43:12
or frameworks for assessing students. But then
43:16
we also hear like, okay, we're not quite at the
43:19
student part yet. Um, and so while it's great
43:23
that we have folks that are working towards some
43:25
of that, um, and I think maybe you had mentioned
43:29
this too, is that our, the teachers aren't, haven't
43:32
quite wrapped their heads around use of these
43:35
tools. So, um, have, have you had a chance to
43:39
talk with anybody about assessments or what that
43:42
could look like when. a student is using AI because
43:46
most of the framework so far, the AI is integrated
43:50
at, it's like depth. So there's levels to the
43:56
assessment with AI use. What are some of your
44:00
thoughts there? Have you had a chance to talk
44:02
with some folks in that space? I have strong
44:04
opinions. So I actually, I am dead set on every
44:10
fiber of my being against AI detection tools.
44:14
And I don't think that AI should ever be used
44:17
for assessment, period, for a lot of reasons.
44:20
There's a lot of ethical issues connected to
44:22
that. There are a lot of glorious fails connected
44:25
to, for example, in the UK a few years ago, they
44:28
decided they were going to use AI to grade all
44:29
of their, you know, the equivalent of like a
44:33
state level or diploma exam in Canada, something
44:36
like a standardized test that decides whether
44:38
you get into university or not. And it went horribly
44:42
wrong. They realized that after the fact, after
44:47
it graded everybody, that part of the model's
44:51
own learning had been that kids from private
44:53
schools typically get 12 % higher than kids from
44:57
public schools. And that was a pattern it learned
44:59
on its own. So it applied that pattern to every
45:01
paper that it graded. and they had to go back,
45:05
remark them all the next year. They had a huge
45:07
surge of kids getting into college who didn't
45:11
get into college the first time because the AI
45:14
failed. The truth is we don't understand about
45:18
how these tools are working in the background,
45:21
making decisions, why they're giving US 69 and
45:25
US 73. We can't ask them questions because we
45:28
can't believe what they say. We know they're
45:30
embedded with bias. things that we might not
45:33
even understand, right? It might be on a kid's
45:34
name. It might be on a story that they told and
45:36
we can't push it. A parent comes in and says,
45:39
why did my kid get a 74 on this? You can't tell
45:42
them. So I just don't think there's a level of
45:44
transparency there that is, and that's important.
45:47
I think there are real questions around consent.
45:50
A kid doesn't consent to having their work put
45:52
into an AI model for those purposes because the
45:55
reality is most ed tech tools. are training their
45:58
models on the work that are put into them, which
46:01
means that you're putting kids' work in there
46:04
without their consent, and it's becoming part
46:07
of a training data set. And a lot of these are
46:09
just notoriously unreliable, is the truth. The
46:15
AI detection tools, sorry. I'm just going to
46:17
make sure I'm on Do Not Disturb here. I'll start
46:19
in a second again here. A lot of these tools
46:25
are notoriously unreliable, especially the AI
46:27
detection tools. They unfairly flag English as
46:30
a second language speakers. They really can't
46:34
tell if you or I were to put in like our master's
46:37
dissertation from 15 years ago, it would be flagged
46:39
as AI generated. I don't think there's any way
46:42
around that. And I just don't think we should
46:43
be using them. I think just in the same way that
46:46
we're saying to kids, like, don't use AI, it's
46:49
cheating. Like we should be looking at that ourselves
46:51
and being like, The most important part of teaching
46:53
is this part that helps you get to know a student
46:56
better. It's important that this is right because
46:58
the stakes are high. So let's not outsource that.
47:00
Maybe that needs to be human led and human directed.
47:04
And do you think too, like with the, uh, like
47:06
things like the, when teachers are designing
47:09
assessments as well, um, is there a good way
47:12
to do that to, you know, allow students certain
47:17
access to things? Uh, obviously some of the tools
47:21
that we've seen, like a magic school AI or school
47:24
AI, they've been working on some of that with
47:27
their, like the student rooms and things where
47:29
you can issue these tools, but, um, there's no
47:33
sign in required and there's some anonymity to
47:36
it. Um, is, is there, when, when teachers are
47:40
designing assessments, what do you think as far
47:44
as that goes with, um, what should we be giving
47:48
students? And what should we refrain from at
47:51
this point? Oh, that tricky. I again, I don't
48:00
know what we should be endorsing as teachers
48:02
until we understand it properly and the implications
48:05
and the risks. So I have questions about that.
48:10
But again, someone needs to be teaching these
48:12
kids how to be using these tools responsibly.
48:13
So if you have, let's say a sandboxed rubric.
48:18
GPT within an ed tech tool that you have that
48:21
acts as you the teacher and gives the rubric
48:24
and kids can put in their assignments, probably
48:26
like you're alluding to and have it. But again,
48:28
part of me struggles because your school's paying
48:30
thousands of dollars for that tool. If we just
48:32
teach a kid how to use a free version of chat
48:34
GPT properly, they can just put in the rubric
48:37
and have it support them in that way. And then
48:41
they're learning how to use the tool and schools
48:44
maybe have enough money to buy another teacher
48:46
instead of spending all of their money on bright
48:49
shiny ed tech tools. That's all opinion. And
48:53
one of the things too, that we've seen more and
48:55
more is there's some folks that are talking about
48:57
like localized models of these. So a district
49:00
would have their kind of their own in -house,
49:03
it would be like siloed and whatever interactions
49:06
happen with that would just remain there. Um,
49:10
because it would be a localized version of the
49:11
GPT or large language model. And that could be
49:15
a good option. Like I run local models myself.
49:18
Again, something to consider from an ethicist
49:21
point of view is making sure that kids know everything
49:26
you put in there. The district can see and we
49:28
might, you know, that you might train your district
49:32
model on it or it might be traced back to you.
49:35
I remember like how jarring it was for me to
49:38
realize that all of these kids' school accounts.
49:40
And it makes sense in hindsight through Google.
49:43
The administration could manage and flag keywords
49:47
that kids were searching for their safety, but
49:49
also see everything that was going on in these
49:51
kids' lives, this type of surveillance. And it
49:54
just made me a little uncomfortable. And so we
49:55
just would need to make sure that there was transparency
49:58
around some of that. There's just a lot of ethical
50:01
questions around even having your own model.
50:03
Is it better than outsourcing it to an outside
50:05
place? Absolutely. But you'd have to make sure
50:08
it was very, very secure and that you were transparent
50:11
about how it was used and what was happening
50:13
that way. So you talk about like some of those
50:19
those, you know, like training kids how to use
50:21
chat GPT appropriately being effective. Like,
50:24
what other sorts of skills do you think would
50:26
be important? I'm on a switch maybe to it could
50:29
be teachers as teachers are learning how to use
50:31
it. But even as we transition into a tool for
50:34
students as well. What are some of those critical
50:36
skills then that we should be focusing on that
50:39
are agnostic of a particular tool, but are just
50:43
critical for using generative AI? Honestly, I
50:47
think it's more about the process. I think that
50:49
co -creation process that maybe students and
50:51
teachers should be learning alongside each other
50:53
and co -creating assessments that would work
50:57
and that students give feedback on what works
50:59
and what doesn't. I think just the basics, like
51:03
what you're talking about, like how to effectively
51:05
chat with these tools, where the line is between
51:08
this is how you engage with them to further thinking
51:11
and here's where it becomes like it's not helping
51:14
you at all. Maybe discussions around that cognitive
51:17
offloading or going back to the why this is important
51:20
or how you can use this to make you think better
51:24
and smarter, maybe having kids critically analyze
51:28
AI outputs. or experiment with generating images
51:32
and examining where the bias and the stereotypes
51:36
are. So I think the biggest kind of cross -platformal
51:40
whatever skill is teaching kids how to think
51:43
critically about AI outputs and their AI youth.
51:47
So how to do it effectively so that they're complimentary
51:51
offloading and not, you know, repetitively. cognitive
51:56
offloading. I think one of my favorite stories
51:59
about this is actually that London cab drivers
52:03
and ambulance drivers never die from Alzheimer's,
52:07
which is fascinating to me in terms of cognitive
52:11
offloading because they have to memorize and
52:14
not rely on GPS entire maps of the cities. And
52:16
every day they have to think through the best
52:18
way to get from here to there without a GPS.
52:20
And they have to be actively thinking about those
52:23
things every single day. And the research is
52:25
actually fascinating because they're thinking
52:28
all the time and it doesn't lead to the cognitive
52:31
decline. So how do we complimentary cognitive
52:34
offload to AI in a way that complements our thinking
52:38
instead of a way that competes with our thinking?
52:42
I think that's the most critical skill and how
52:43
to challenge and think critically about the outputs
52:48
that AI creates. I think those are the two big
52:50
skills. So as you start to look forward here,
52:54
um, and I know this is really difficult to do,
52:57
but we like to ask this question and you know,
53:00
if you can answer best, best you can answer,
53:03
um, what is, I'm going to give you two dates
53:06
here. What does AI look like in September, 2026
53:11
in your opinion? And what does it look like in
53:14
September, 2030? I think in 2026, I think it
53:18
will be very similar to what we have now. I think,
53:20
I don't think we're gonna see these orders of
53:23
magnitude changes. I think, you know, if I looked
53:27
at it two years ago, the basics are kind of like,
53:30
it's way better, but you interact with it, you
53:32
get a response. I think that part is. going to
53:37
be the same. And I think there'll still be people
53:40
who don't think it matters and thinks it's a
53:42
fad. But I think they'll be a little too late
53:45
by that time. So I think it looks the same. I
53:47
think just more people will be figuring out how
53:49
to use it. As more people figure out how to use
53:51
it effectively, we'll come up with better ideas
53:53
and better use cases. But people are barely using
53:56
some of the amazing ways that you can use AI
53:59
now, like the Notebook LM thing. I can give that
54:02
a research paper and then have the hosts fight
54:05
about it like it's celebrity gossip and you know
54:08
like all of the different ways that you can interrogate
54:09
your thinking i think i don't think we'll see
54:12
and i don't know anything massively different
54:14
i do wonder by 2030 i just think wearables are
54:19
going to be a bigger deal than they are now we're
54:21
already seeing the ai glasses i have a pair and
54:25
you know those are in schools there it's not
54:28
like the most amazing amazing technology in the
54:30
world but it's you know, behind the scenes, it's
54:33
getting pretty good. And what does it mean once,
54:35
you know, everybody's wearing their technology
54:38
or they don't carry phones anymore, right? Like
54:41
it all becomes wearables. And I think, you know,
54:44
we've heard about Johnny Ives and Sam Altman
54:46
working on something new that's not a phone,
54:49
that's, we don't know what it is, but they say
54:51
it'll be on your desk with your computer and
54:53
your phone and this thing. So we don't know what
54:55
that is, but if it's a big deal, it might. change
54:58
if kids have those in schools, right? So what
55:01
does that look like and how can kids be learning
55:03
with those things? We'll probably ban them right
55:05
away, but, you know, how do we make sense of
55:10
those things? You know, we're so far behind.
55:13
We just finally banned cell phones in schools.
55:16
And so it'll be interesting to see. In terms
55:20
of AI, though, I think that we're getting as
55:23
far as we can with the the interfaces, like the
55:26
computer and the phone interfaces. I think we'll
55:29
be using voice a lot more. I think we'll have
55:31
more wearables and interactions, almost like
55:35
the kind of sci -fi thing you think about where
55:37
maybe I don't have a three monitor setup. Maybe
55:39
I do have whatever, if Apple finally gets it
55:42
together, their headset is or whatever. And I
55:44
just have my full office and everything with
55:47
me. And then what does that mean for students?
55:49
If you don't need a computer in every classroom
55:52
for every student and things happen this way
55:54
and it's a lot more virtual. How do we, that's
55:58
why we need to start thinking about these very
55:59
basic rudimentary things now, like large language
56:02
models that give you maybe right answers so that
56:05
we can stay caught up when things get released.
56:06
So by 2030, I think we'll have more wearables.
56:08
I don't think it'll be like science fiction,
56:10
but I think we'll be using AI in a lot different
56:13
ways than we are right now. Yeah, great. Thank
56:18
you. Um, so What we've kind of been ending some
56:21
of our episodes here with is we put the guest
56:24
on the spot here with some rapid fire kind of
56:28
lightning round type questions. And so David
56:32
and I will kind of alternate here. Feel free
56:35
to, if you want a one word answer, that's great.
56:37
If you want to give a little bit of explanation
56:39
to your answer, it's kind of whatever you choose,
56:43
right? So I guess I'll get to go with the first
56:47
one. And this one is kind of right up my alley
56:49
because Actually, you talked about what was that
56:51
hook that got you into using AI. And for me,
56:55
it was cooking. So my first question to kick
57:01
it off then is what's the best food in Canada?
57:05
Oh, you know, I don't know if Canada is known
57:08
for their food. The maple syrup is pretty fire,
57:11
I guess. Poutine at our cock crows. We have poutine.
57:14
So it's the good Costco fries with big chunks
57:17
of cheese and gravy That's that's a fan favorite,
57:20
but all of my favorite foods are at different
57:22
cities in the u .s Okay, so the next question
57:27
is this is a linkedin reference, uh, if you've
57:29
maybe I saw it I don't know if everybody follows
57:32
Rebecca, but you should definitely should because
57:35
there's a lot of great stuff there on her profile,
57:37
but also you recently posted about getting work
57:41
done in your PJs, and there was a picture of
57:43
you. So can you in fact get work done in your
57:47
PJs? I can confirm I am actually wearing pajamas
57:51
right now on the bottom. So I basically live
57:54
in my pajamas. I am business up top, all comfy
57:57
on the bottom. All right, at least we're going
58:00
to be putting that to bed. You mentioned you
58:03
need a mom of four. Yep. What's one piece of
58:06
parenting advice that you would give to parents
58:08
about their children's interactions with technology
58:11
and or AI? Ask them about it. Ask them to teach
58:15
you about it. Ask them to tell you how they're
58:18
using it. Have conversations about it. Ask them
58:21
how they feel about like if it's cheating. Find
58:23
out what their professors are saying about it
58:24
because I'm in a unique position where like I
58:28
know when their teachers are telling them information
58:30
that is wrong. and it's super annoying, right?
58:33
So there'll be, a kid will come home and say,
58:34
well, I'm not allowed to use AI, it's cheating.
58:36
And if they put it in an AI detector and it comes
58:39
back, then I'm kicked out of the class. And I
58:41
was, I'm like, bet, bet. I will fight to the
58:46
death against that AI detector for you, or that's
58:49
actually not true, or they can't do that, or
58:52
AI is not cheating if you use it like this. And
58:55
so talk to your kids about it because honestly,
58:57
there's a lot of harm being done. I have one
58:59
kid in college who lives in this constant state
59:03
of heightened anxiety that he's going to be unfairly
59:05
flagged by an AI detector and kicked out of his
59:07
program. even though he's too scared to use AI
59:10
because he was so brainwashed about it in high
59:12
school. Talk to your kids about it and have them
59:15
teach you about it and what their teachers are
59:17
saying about it. If you don't know anything,
59:20
have them teach you. But if you do know something,
59:22
then help challenge their own critical thinking
59:25
skills about it in good ways and in bad ways.
59:30
You'll learn a lot in the process and also it'll
59:32
make your kids feel safe to come to you and ask
59:34
you questions about AI too. All right. Uh, if
59:39
you had to teach a large language model, um,
59:42
AI, a human emotion, which one would you choose
59:45
and why? Honestly, like the timing on this probably
59:50
trucks because over the last couple of weeks,
59:53
I've lost a lot of faith in humanity. I'm not
59:55
going to lie. Like I, I, uh, I've been questioning
59:59
whether we even want to teach AI to be human
1:00:01
because there's just like a lot of really, you
1:00:04
know, things happening in the world where you're
1:00:06
just not seeing the best of humans and people.
1:00:09
If it were possible, you know, obviously, compassion,
1:00:15
you know, empathy, to be able to see it's not
1:00:21
even possible to be able to feel the things that
1:00:23
humans feel because it can demonstrate and appropriate
1:00:26
empathy so well. But I think if I could teach
1:00:30
humans too. To also have this thing, it would
1:00:32
be compassion and empathy. Let's just give it
1:00:34
to everybody. It doesn't matter if it's robotic
1:00:38
or numerical priority. Yeah. Yeah. So you live
1:00:42
and work on a farm. What is one simple low tech
1:00:47
tool that you think will never ever be replaced
1:00:50
by AI? Great question. Honestly, like. I still
1:00:58
to this day cannot think properly without a pen
1:01:01
in my hand. I just think there's something to
1:01:05
actually pen and paper sometimes. I think there'll
1:01:09
always be a need for that. That's a great question
1:01:13
though. I'm trying to think of a good kind of
1:01:15
low tech tool because I am kind of a high tech
1:01:18
kind of gal and the world is kind of going that
1:01:20
way a lot. Yeah, I've probably gone high tech
1:01:25
in most areas of my life. I even have like a
1:01:27
coffee mug that stays hot all day. So that's
1:01:32
not even like a uniquely - That's essential.
1:01:34
That is, yeah. Yeah. So, but that's a great question
1:01:37
because that's going to be what I'm going to
1:01:38
be thinking about all day. And I'll probably
1:01:41
come up with a really good answer probably later
1:01:45
tonight, but you know, maybe a good book or something
1:01:47
for - actually like holding a book and turning
1:01:50
pages in your hands that, yes, you can do high
1:01:53
tech, but nothing quite replaces that experience
1:01:56
in my mind. Well, if you think about it, be sure
1:01:59
to let us know. I will. And Rebecca, we've just
1:02:03
really appreciated you taking the time to spend
1:02:05
with us. And we thought it was just a great,
1:02:08
insightful conversation. And I hope there's a
1:02:10
lot of takeaways from it that our listeners will
1:02:13
get. I appreciate it. Thanks so much for having
1:02:15
me on and happy to chat anytime. Maybe we'll
1:02:18
revisit some of those predictions I made in 2026
1:02:21
and again in 2030 and we'll see where it all
1:02:24
ended up. We'll make sure to hold you to them.
1:02:27
Thanks, Rebecca. Thank you. Thanks for listening
1:02:30
to Integrate This. Want to take your learning
1:02:33
further? Head over to our YouTube channel for
1:02:36
quick tutorials, deep dives, and more ed tech
1:02:39
tips you can use right away. Don't forget to
1:02:42
subscribe and catch new podcast episodes every
1:02:45
Thursday on Apple, Spotify or wherever you get
1:02:49
your podcasts. Until next time, keep exploring,
1:02:53
keep innovating and keep integrating this.