1
00:00:03,703 --> 00:00:05,538
This is the Discovery Files Podcast
2
00:00:05,538 --> 00:00:06,306
from the U.S.
3
00:00:06,306 --> 00:00:09,309
National Science Foundation.
4
00:00:09,776 --> 00:00:11,544
Building on a foundation of support
5
00:00:11,544 --> 00:00:12,345
for fundamental
6
00:00:12,345 --> 00:00:14,881
AI research, scientific discovery,
7
00:00:14,881 --> 00:00:16,516
advanced computing infrastructure,
8
00:00:16,516 --> 00:00:17,384
and the development
9
00:00:17,384 --> 00:00:19,052
of the American Stem workforce,
10
00:00:19,052 --> 00:00:21,154
NSF is committed to accelerating
11
00:00:21,154 --> 00:00:23,556
AI driven innovation that strengthens US.
12
00:00:23,556 --> 00:00:24,858
Scientific leadership,
13
00:00:24,858 --> 00:00:26,593
expands access to world class
14
00:00:26,593 --> 00:00:28,495
research infrastructure, and prepares
15
00:00:28,495 --> 00:00:29,295
the next generation
16
00:00:29,295 --> 00:00:30,597
of American science,
17
00:00:30,597 --> 00:00:32,032
technology, engineering,
18
00:00:32,032 --> 00:00:33,466
and mathematics talent.
19
00:00:33,466 --> 00:00:34,834
We are joined today by Tiffany
20
00:00:34,834 --> 00:00:35,835
Barnes, Shiyan
21
00:00:35,835 --> 00:00:37,737
Jiang and Xiaoyi Tian,
22
00:00:37,737 --> 00:00:39,706
NSF-supported researchers
23
00:00:39,706 --> 00:00:40,740
whose Elementary
24
00:00:40,740 --> 00:00:41,674
AI project
25
00:00:41,674 --> 00:00:42,742
works to prepare students
26
00:00:42,742 --> 00:00:44,210
for an AI driven future.
27
00:00:44,210 --> 00:00:45,111
Professor Barnes
28
00:00:45,111 --> 00:00:46,513
and Jiang and Doctor Tian,
29
00:00:46,513 --> 00:00:47,881
thank you so much for joining me today.
30
00:00:47,881 --> 00:00:49,115
Thank you for having us.
31
00:00:49,115 --> 00:00:50,216
So I'd like to start
32
00:00:50,216 --> 00:00:52,018
with the big definition.
33
00:00:52,018 --> 00:00:53,153
Professor Barnes, can you tell us
34
00:00:53,153 --> 00:00:55,555
what is the Elementary AI project?
35
00:00:55,555 --> 00:00:56,156
Elementary
36
00:00:56,156 --> 00:00:59,692
AI is an NSF funded computer Science
37
00:00:59,692 --> 00:01:00,960
for all research practice
38
00:01:00,960 --> 00:01:02,629
partnership between North
39
00:01:02,629 --> 00:01:04,130
Carolina State University
40
00:01:04,130 --> 00:01:05,231
Computer Science Department
41
00:01:05,231 --> 00:01:06,299
and the Friday Institute
42
00:01:06,299 --> 00:01:08,535
and Montgomery County Schools.
43
00:01:08,535 --> 00:01:10,336
We partnered to bring AI
44
00:01:10,336 --> 00:01:11,438
and computational thinking
45
00:01:11,438 --> 00:01:13,073
to all elementary students
46
00:01:13,073 --> 00:01:15,975
in all six schools in the district
47
00:01:15,975 --> 00:01:16,976
by working in partnership
48
00:01:16,976 --> 00:01:18,778
with teachers to learn AI,
49
00:01:18,778 --> 00:01:19,813
but also integrate it
50
00:01:19,813 --> 00:01:21,414
into the existing curriculum.
51
00:01:21,414 --> 00:01:23,383
And our goal is to help students
52
00:01:23,383 --> 00:01:25,351
improve on their end of year scores
53
00:01:25,351 --> 00:01:27,120
and improve overall excitement
54
00:01:27,120 --> 00:01:28,555
about learning in the district.
55
00:01:28,555 --> 00:01:31,024
Is AI usually introduced that early?
56
00:01:31,024 --> 00:01:32,225
The general public probably doesn't
57
00:01:32,225 --> 00:01:34,561
really understand how early computer
58
00:01:34,561 --> 00:01:35,695
things are being introduced
59
00:01:35,695 --> 00:01:37,197
to kids at this point in time.
60
00:01:37,197 --> 00:01:38,498
AI is not introduced
61
00:01:38,498 --> 00:01:40,033
very often in elementary school,
62
00:01:40,033 --> 00:01:41,234
but there are a lot of ideas
63
00:01:41,234 --> 00:01:42,035
in AI
64
00:01:42,035 --> 00:01:42,802
and computational
65
00:01:42,802 --> 00:01:44,504
thinking that can be introduced
66
00:01:44,504 --> 00:01:45,939
when kids are very young.
67
00:01:45,939 --> 00:01:47,540
So we focus on concepts
68
00:01:47,540 --> 00:01:49,142
like computational thinking,
69
00:01:49,142 --> 00:01:51,144
which includes pattern recognition,
70
00:01:51,144 --> 00:01:53,113
abstraction, decomposition.
71
00:01:53,113 --> 00:01:54,080
Kids are already doing
72
00:01:54,080 --> 00:01:55,081
these things in school,
73
00:01:55,081 --> 00:01:56,216
and so we think it's
74
00:01:56,216 --> 00:01:57,584
a very powerful partnership
75
00:01:57,584 --> 00:01:59,953
to work with elementary school teachers
76
00:01:59,953 --> 00:02:01,554
while the kids are learning things,
77
00:02:01,554 --> 00:02:01,821
you know,
78
00:02:01,821 --> 00:02:02,522
when they learn,
79
00:02:02,522 --> 00:02:03,656
like what are shapes,
80
00:02:03,656 --> 00:02:05,358
even when they're really small.
81
00:02:05,358 --> 00:02:07,260
And that's a pattern recognition task.
82
00:02:07,260 --> 00:02:08,495
And AI researchers
83
00:02:08,495 --> 00:02:11,030
have always been inspired by human
84
00:02:11,030 --> 00:02:12,499
learning, and kids are learning,
85
00:02:12,499 --> 00:02:13,399
and they know that.
86
00:02:13,399 --> 00:02:15,468
And teachers can hook
87
00:02:15,468 --> 00:02:17,871
those ideas about AI learning
88
00:02:17,871 --> 00:02:19,772
to what kids are learning anytime.
89
00:02:19,772 --> 00:02:20,874
For a follow up question,
90
00:02:20,874 --> 00:02:22,475
a little bit of background.
91
00:02:22,475 --> 00:02:24,377
How did this project come together?
92
00:02:24,377 --> 00:02:25,812
It grew out of a decade long
93
00:02:25,812 --> 00:02:27,380
foundation of trust between North
94
00:02:27,380 --> 00:02:29,115
Carolina State University,
95
00:02:29,115 --> 00:02:30,183
the Friday Institute,
96
00:02:30,183 --> 00:02:31,951
which has been working across the state
97
00:02:31,951 --> 00:02:33,052
to help promote
98
00:02:33,052 --> 00:02:34,654
educational innovation
99
00:02:34,654 --> 00:02:36,923
and outcomes, and Montgomery
100
00:02:36,923 --> 00:02:37,390
County Schools,
101
00:02:37,390 --> 00:02:38,691
has been partnering a long time
102
00:02:38,691 --> 00:02:40,760
with NC State as well.
103
00:02:40,760 --> 00:02:44,397
So in 2015, the Friday Institute
104
00:02:44,430 --> 00:02:46,666
helped with 1 to 1 tech
105
00:02:46,666 --> 00:02:48,601
integration and learner agency
106
00:02:48,601 --> 00:02:49,302
initiatives.
107
00:02:49,302 --> 00:02:50,336
Carrie Robledo,
108
00:02:50,336 --> 00:02:52,705
who is in the district now as a teacher.
109
00:02:52,705 --> 00:02:54,674
She was at the Friday Institute
110
00:02:54,674 --> 00:02:56,342
and she was a digital learning coach.
111
00:02:56,342 --> 00:02:56,976
And Joanna
112
00:02:56,976 --> 00:02:59,913
Perkins is one of the directors for K-12
113
00:02:59,913 --> 00:03:01,114
curriculum in the district.
114
00:03:01,114 --> 00:03:03,316
And she partnered with us to co-design
115
00:03:03,316 --> 00:03:04,350
this partnership
116
00:03:04,350 --> 00:03:05,418
as they were introducing
117
00:03:05,418 --> 00:03:07,387
the new CKLA curriculum.
118
00:03:07,387 --> 00:03:07,787
We felt
119
00:03:07,787 --> 00:03:09,756
it would be really advantageous time
120
00:03:09,756 --> 00:03:11,024
to also bring in AI
121
00:03:11,024 --> 00:03:11,858
to help the teachers
122
00:03:11,858 --> 00:03:13,059
and the students get excited
123
00:03:13,059 --> 00:03:15,195
about this new age of AI.
124
00:03:15,195 --> 00:03:16,162
So, Professor Jiang,
125
00:03:16,162 --> 00:03:17,997
I'd like to ask you a question here.
126
00:03:17,997 --> 00:03:19,599
How have you
127
00:03:19,599 --> 00:03:20,867
or how has this project
128
00:03:20,867 --> 00:03:21,601
really worked
129
00:03:21,601 --> 00:03:22,835
to empower the teachers
130
00:03:22,835 --> 00:03:23,336
to become
131
00:03:23,336 --> 00:03:25,271
AI educators, to adopt
132
00:03:25,271 --> 00:03:27,307
AI concepts into their curriculum?
133
00:03:27,307 --> 00:03:29,075
One of the biggest misconception
134
00:03:29,075 --> 00:03:29,676
that teachers
135
00:03:29,676 --> 00:03:30,710
need to become
136
00:03:30,710 --> 00:03:33,713
like AI experts before they can teach AI.
137
00:03:33,746 --> 00:03:36,382
But our philosophy is very different.
138
00:03:36,382 --> 00:03:38,351
We help teachers like what Dr Barnes
139
00:03:38,351 --> 00:03:39,118
just mentioned.
140
00:03:39,118 --> 00:03:39,953
We help teachers
141
00:03:39,953 --> 00:03:42,055
recognize that many of the skills
142
00:03:42,055 --> 00:03:43,856
they already teach in classrooms
143
00:03:43,856 --> 00:03:45,225
such as identifying
144
00:03:45,225 --> 00:03:47,493
patterns, asking good questions,
145
00:03:47,493 --> 00:03:49,862
interpreting evidence, and discussing
146
00:03:49,862 --> 00:03:51,130
ethical issues
147
00:03:51,130 --> 00:03:53,333
that are very central to AI literacy.
148
00:03:53,333 --> 00:03:55,501
So we build on teachers
149
00:03:55,501 --> 00:03:57,136
existing experiences
150
00:03:57,136 --> 00:03:58,938
and also teachers, they receive
151
00:03:58,938 --> 00:04:02,041
ongoing coaching, proper development,
152
00:04:02,075 --> 00:04:03,309
collaborate with lead
153
00:04:03,309 --> 00:04:04,811
teachers, share lesson
154
00:04:04,811 --> 00:04:06,546
ideas, become leaders
155
00:04:06,546 --> 00:04:08,114
who support colleagues
156
00:04:08,114 --> 00:04:09,749
across the district.
157
00:04:09,749 --> 00:04:12,151
So the goal is to not to like it,
158
00:04:12,151 --> 00:04:14,621
simply like training individual teachers.
159
00:04:14,621 --> 00:04:15,455
Our goal is to
160
00:04:15,455 --> 00:04:17,223
build a sustainable community
161
00:04:17,223 --> 00:04:19,559
of AI educators for the district.
162
00:04:19,559 --> 00:04:21,094
I'm thinking about some of the challenges
163
00:04:21,094 --> 00:04:22,095
that might come in here
164
00:04:22,095 --> 00:04:24,230
in the broader district especially,
165
00:04:24,230 --> 00:04:26,266
and thinking about getting the tech
166
00:04:26,266 --> 00:04:27,300
to the teachers
167
00:04:27,300 --> 00:04:28,101
maybe, maybe
168
00:04:28,101 --> 00:04:30,136
getting school boards on to the project.
169
00:04:30,136 --> 00:04:31,537
Are there limitations
170
00:04:31,537 --> 00:04:32,939
to the curriculum involved?
171
00:04:32,939 --> 00:04:34,674
Can you talk about some of the challenges
172
00:04:34,674 --> 00:04:36,175
in getting this implemented?
173
00:04:36,175 --> 00:04:39,178
So in terms of curriculum limitations,
174
00:04:39,178 --> 00:04:41,814
we worked very closely with teachers
175
00:04:41,814 --> 00:04:44,183
to identify the natural connection
176
00:04:44,183 --> 00:04:46,853
with their existing curriculum, like Dr.
177
00:04:46,853 --> 00:04:48,388
Barnes mentioned about CKLA
178
00:04:48,388 --> 00:04:49,656
which is a new literacy
179
00:04:49,656 --> 00:04:50,857
curriculum that they
180
00:04:50,857 --> 00:04:53,026
were adopting this last year.
181
00:04:53,026 --> 00:04:56,829
So that's not a really major limitation.
182
00:04:56,829 --> 00:04:58,331
But the biggest challenge
183
00:04:58,331 --> 00:04:59,098
that I saw
184
00:04:59,098 --> 00:05:00,900
is that a helping people
185
00:05:00,900 --> 00:05:01,267
see that
186
00:05:01,267 --> 00:05:03,936
AI belongs in elementary education,
187
00:05:03,936 --> 00:05:05,171
and it can fit naturally
188
00:05:05,171 --> 00:05:06,906
into existing classrooms.
189
00:05:06,906 --> 00:05:08,708
Teachers may and initially
190
00:05:08,708 --> 00:05:10,276
before we came in,
191
00:05:10,276 --> 00:05:11,377
they may worry that
192
00:05:11,377 --> 00:05:12,979
AI is too technical
193
00:05:12,979 --> 00:05:15,748
or too advanced for young kids.
194
00:05:15,748 --> 00:05:16,549
With many kids,
195
00:05:16,549 --> 00:05:17,717
they don't even have
196
00:05:17,717 --> 00:05:18,217
that many
197
00:05:18,217 --> 00:05:18,985
digital literacy
198
00:05:18,985 --> 00:05:20,553
skills and others
199
00:05:20,553 --> 00:05:21,554
might concern
200
00:05:21,554 --> 00:05:22,388
that teaching
201
00:05:22,388 --> 00:05:24,857
AI means adding another subject
202
00:05:24,857 --> 00:05:25,692
to a already
203
00:05:25,692 --> 00:05:27,026
very full curriculum,
204
00:05:27,026 --> 00:05:28,761
on top of the important
205
00:05:28,761 --> 00:05:30,630
skills like math and EOA.
206
00:05:30,630 --> 00:05:32,465
So our approach addresses
207
00:05:32,465 --> 00:05:34,200
those challenges or concerns
208
00:05:34,200 --> 00:05:36,035
by integrating AI into lessons
209
00:05:36,035 --> 00:05:37,303
teachers already teaching.
210
00:05:37,303 --> 00:05:40,273
And we are not asking to replace
211
00:05:40,273 --> 00:05:41,307
reading or mass.
212
00:05:41,307 --> 00:05:42,742
We reach those subjects
213
00:05:42,742 --> 00:05:44,410
with AI literacy skills.
214
00:05:44,410 --> 00:05:45,878
So like the example,
215
00:05:45,878 --> 00:05:47,980
these can learn about pattern recognition
216
00:05:47,980 --> 00:05:49,949
while they study birds
217
00:05:49,949 --> 00:05:51,150
doing a science activity.
218
00:05:51,150 --> 00:05:52,251
Because birds might have
219
00:05:52,251 --> 00:05:53,453
a certain features
220
00:05:53,453 --> 00:05:55,221
that can be patterns and doesn't
221
00:05:55,221 --> 00:05:56,823
connect to AI concepts.
222
00:05:56,823 --> 00:05:58,291
And another challenge
223
00:05:58,291 --> 00:05:59,158
I will say
224
00:05:59,158 --> 00:06:01,160
is a long term sustainability,
225
00:06:01,160 --> 00:06:01,794
which is something
226
00:06:01,794 --> 00:06:03,296
we are working on very hard.
227
00:06:03,296 --> 00:06:03,629
That's why
228
00:06:03,629 --> 00:06:04,731
we work very closely
229
00:06:04,731 --> 00:06:06,833
with district leaders and teachers
230
00:06:06,833 --> 00:06:08,334
through what we call
231
00:06:08,334 --> 00:06:09,836
research practice, partnership.
232
00:06:09,836 --> 00:06:11,938
And through this kind of partnership.
233
00:06:11,938 --> 00:06:13,473
We hope that AI
234
00:06:13,473 --> 00:06:14,574
continues long
235
00:06:14,574 --> 00:06:16,576
after the research project ends.
236
00:06:16,576 --> 00:06:17,510
I'd like to add
237
00:06:17,510 --> 00:06:18,411
that the district
238
00:06:18,411 --> 00:06:21,314
has been a real partner in this project.
239
00:06:21,314 --> 00:06:22,982
So not only have the teachers
240
00:06:22,982 --> 00:06:23,916
been real partners
241
00:06:23,916 --> 00:06:26,419
bringing new AI to their students
242
00:06:26,419 --> 00:06:27,653
much of the school year,
243
00:06:27,653 --> 00:06:29,789
but the district administration
244
00:06:29,789 --> 00:06:31,491
has been working with us.
245
00:06:31,491 --> 00:06:31,791
You know,
246
00:06:31,791 --> 00:06:32,191
we meet
247
00:06:32,191 --> 00:06:33,259
at least every month,
248
00:06:33,259 --> 00:06:34,560
and sometimes
249
00:06:34,560 --> 00:06:34,894
we meet
250
00:06:34,894 --> 00:06:36,162
with the superintendent,
251
00:06:36,162 --> 00:06:37,463
and sometimes we meet
252
00:06:37,463 --> 00:06:39,432
with the curriculum specialist.
253
00:06:39,432 --> 00:06:41,901
And the fact that we have these embedded
254
00:06:41,901 --> 00:06:43,035
teachers in the districts mean
255
00:06:43,035 --> 00:06:43,936
that we can regularly
256
00:06:43,936 --> 00:06:45,171
find out what's going on.
257
00:06:45,171 --> 00:06:45,972
And then
258
00:06:45,972 --> 00:06:47,573
we also get to hear, like,
259
00:06:47,573 --> 00:06:48,674
what are the important problems
260
00:06:48,674 --> 00:06:50,042
that the district is trying to solve.
261
00:06:50,042 --> 00:06:53,413
And this is a strong type of research
262
00:06:53,413 --> 00:06:54,714
that NSF is funding
263
00:06:54,714 --> 00:06:56,349
these research practice partnerships.
264
00:06:56,349 --> 00:06:58,618
And it makes me feel wonderful
265
00:06:58,618 --> 00:06:59,719
to work on this kind of thing,
266
00:06:59,719 --> 00:07:01,387
because I don't feel
267
00:07:01,387 --> 00:07:01,788
that I'm
268
00:07:01,788 --> 00:07:04,390
just bringing an extra thing to schools,
269
00:07:04,390 --> 00:07:06,192
but I feel that we are actually helping
270
00:07:06,192 --> 00:07:06,893
solve problems
271
00:07:06,893 --> 00:07:08,661
that the district wants to solve,
272
00:07:08,661 --> 00:07:09,662
and the initiatives
273
00:07:09,662 --> 00:07:12,665
that they are interested in doing.
274
00:07:12,832 --> 00:07:15,301
This partnership enables us to be able
275
00:07:15,301 --> 00:07:16,602
to listen, to be able
276
00:07:16,602 --> 00:07:19,639
to brainstorm solutions to problems
277
00:07:19,639 --> 00:07:21,641
in the school district or just,
278
00:07:21,641 --> 00:07:22,909
you know, desires.
279
00:07:22,909 --> 00:07:24,343
I'm not saying there are problems.
280
00:07:24,343 --> 00:07:26,145
I'm saying that, you know,
281
00:07:26,145 --> 00:07:28,214
whatever those priorities are,
282
00:07:28,214 --> 00:07:29,782
since we're there as partners,
283
00:07:29,782 --> 00:07:31,784
we actually get to talk together about
284
00:07:31,784 --> 00:07:33,753
what do we want to be excited about?
285
00:07:33,753 --> 00:07:34,854
What do we want to help other people
286
00:07:34,854 --> 00:07:35,521
be excited about?
287
00:07:35,521 --> 00:07:37,023
What do we want to help accomplish?
288
00:07:37,023 --> 00:07:38,758
And I think that it's key
289
00:07:38,758 --> 00:07:39,358
not only that
290
00:07:39,358 --> 00:07:40,359
the teachers are involved,
291
00:07:40,359 --> 00:07:41,260
but also that
292
00:07:41,260 --> 00:07:43,429
the administration is on board.
293
00:07:43,429 --> 00:07:43,663
You know,
294
00:07:43,663 --> 00:07:45,298
they're thinking about buying computers
295
00:07:45,298 --> 00:07:46,098
for every kid,
296
00:07:46,098 --> 00:07:47,500
and there's new initiatives
297
00:07:47,500 --> 00:07:49,435
across the state about,
298
00:07:49,435 --> 00:07:50,203
you know, determining,
299
00:07:50,203 --> 00:07:51,304
like how much screen
300
00:07:51,304 --> 00:07:52,505
time should kids have?
301
00:07:52,505 --> 00:07:53,940
So we had discussions
302
00:07:53,940 --> 00:07:55,041
in some of our meetings,
303
00:07:55,041 --> 00:07:56,375
like, what do researchers
304
00:07:56,375 --> 00:07:56,943
think about that?
305
00:07:56,943 --> 00:07:57,677
What does research
306
00:07:57,677 --> 00:07:59,612
say about how much screen time
307
00:07:59,612 --> 00:08:00,613
children should have?
308
00:08:00,613 --> 00:08:03,349
And we all had discussions and agreed
309
00:08:03,349 --> 00:08:05,184
that screens can be useful,
310
00:08:05,184 --> 00:08:06,586
especially when you get to do something
311
00:08:06,586 --> 00:08:09,222
like chatbots, Toy Talk that Dr.
312
00:08:09,222 --> 00:08:10,189
Tian and Dr.
313
00:08:10,189 --> 00:08:12,058
Jiang were bringing to the project.
314
00:08:12,058 --> 00:08:13,059
But you know,
315
00:08:13,059 --> 00:08:14,694
when they're not doing things like that,
316
00:08:14,694 --> 00:08:15,661
if they're just doing something
317
00:08:15,661 --> 00:08:16,596
that could have been done
318
00:08:16,596 --> 00:08:17,530
by reading a book
319
00:08:17,530 --> 00:08:19,265
or doing a worksheet,
320
00:08:19,265 --> 00:08:21,200
or doing something kinesthetic,
321
00:08:21,200 --> 00:08:22,502
where they're moving around the room
322
00:08:22,502 --> 00:08:23,803
and trying to make an analogy
323
00:08:23,803 --> 00:08:25,838
between how their physical body works
324
00:08:25,838 --> 00:08:27,306
and how some concept
325
00:08:27,306 --> 00:08:28,307
that they're learning in math
326
00:08:28,307 --> 00:08:31,477
or reading actually can be related to.
327
00:08:31,511 --> 00:08:32,078
There's different
328
00:08:32,078 --> 00:08:33,513
needs for different schools,
329
00:08:33,513 --> 00:08:34,947
different kids, different teachers,
330
00:08:34,947 --> 00:08:36,682
and being able to work together
331
00:08:36,682 --> 00:08:37,550
in this practice
332
00:08:37,550 --> 00:08:40,019
partnership means that we can address
333
00:08:40,019 --> 00:08:41,387
those as they come up.
334
00:08:41,387 --> 00:08:43,322
Doctor Tian, I had a question
335
00:08:43,322 --> 00:08:44,457
that kind of came out of here
336
00:08:44,457 --> 00:08:45,658
that you wrote a paper about
337
00:08:45,658 --> 00:08:47,393
and thinking about accommodating
338
00:08:47,393 --> 00:08:48,561
different learning styles.
339
00:08:48,561 --> 00:08:49,562
Can you talk a little bit
340
00:08:49,562 --> 00:08:50,830
about the approach
341
00:08:50,830 --> 00:08:53,199
to maybe somebody like myself?
342
00:08:53,199 --> 00:08:54,233
I've learned better
343
00:08:54,233 --> 00:08:55,902
with hearing in school
344
00:08:55,902 --> 00:08:58,905
and reading it as opposed to being shown.
345
00:08:58,938 --> 00:08:59,839
Can you talk a little bit
346
00:08:59,839 --> 00:09:01,073
about these kind of concepts
347
00:09:01,073 --> 00:09:02,341
and how you approach them?
348
00:09:02,341 --> 00:09:05,077
When we are working with the schools,
349
00:09:05,077 --> 00:09:07,313
one of the biggest challenges
350
00:09:07,313 --> 00:09:08,581
that we have seen
351
00:09:08,581 --> 00:09:10,182
is how do we design
352
00:09:10,182 --> 00:09:12,652
AI learning experiences
353
00:09:12,652 --> 00:09:15,221
when students come into the classroom
354
00:09:15,221 --> 00:09:16,923
with very different needs,
355
00:09:16,923 --> 00:09:17,323
you know,
356
00:09:17,323 --> 00:09:20,426
they have very different reading levels.
357
00:09:20,459 --> 00:09:22,628
They have different confidence
358
00:09:22,628 --> 00:09:26,232
level or trust level with AI technologies
359
00:09:26,232 --> 00:09:27,800
and different ways
360
00:09:27,800 --> 00:09:29,302
that they prefer to learn.
361
00:09:29,302 --> 00:09:32,538
So we are really thinking about,
362
00:09:32,572 --> 00:09:32,838
you know,
363
00:09:32,838 --> 00:09:34,240
the kinds of scaffolds
364
00:09:34,240 --> 00:09:35,341
that we can provide,
365
00:09:35,341 --> 00:09:36,943
both in the lessons
366
00:09:36,943 --> 00:09:40,346
and in the technology itself.
367
00:09:40,513 --> 00:09:42,615
On the instructional side,
368
00:09:42,615 --> 00:09:45,585
we try to provide multiple ways
369
00:09:45,585 --> 00:09:48,120
for students to engage with the content.
370
00:09:48,120 --> 00:09:49,855
So, for example,
371
00:09:49,855 --> 00:09:51,958
instead of only asking students
372
00:09:51,958 --> 00:09:54,694
to read the story or read the book,
373
00:09:54,694 --> 00:09:57,797
we gave the option for them to read
374
00:09:57,797 --> 00:09:59,832
or listen to the materials.
375
00:09:59,832 --> 00:10:02,401
And we also designed some
376
00:10:02,401 --> 00:10:04,370
unplugged activities
377
00:10:04,370 --> 00:10:05,237
where students
378
00:10:05,237 --> 00:10:06,772
learn about computing concepts
379
00:10:06,772 --> 00:10:08,174
without using computers
380
00:10:08,174 --> 00:10:09,709
or be on the screen.
381
00:10:09,709 --> 00:10:12,712
So one activity that we ask students
382
00:10:12,812 --> 00:10:15,715
when we are teaching this AI concept, one
383
00:10:15,715 --> 00:10:18,384
important concept is to understand
384
00:10:18,384 --> 00:10:19,719
effective prompts.
385
00:10:19,719 --> 00:10:22,021
When you're talking with AI chatbots,
386
00:10:22,021 --> 00:10:22,588
how do you
387
00:10:22,588 --> 00:10:24,657
construct a good prompt
388
00:10:24,657 --> 00:10:27,259
so that you can get your desired answer?
389
00:10:27,259 --> 00:10:29,562
So we had an activity
390
00:10:29,562 --> 00:10:32,732
that asks students to compare different
391
00:10:32,798 --> 00:10:33,866
AI prompts
392
00:10:33,866 --> 00:10:36,435
so they would move to different sides
393
00:10:36,435 --> 00:10:37,436
of the classroom
394
00:10:37,436 --> 00:10:39,038
to vote, which prompt
395
00:10:39,038 --> 00:10:40,773
they think is more effective.
396
00:10:40,773 --> 00:10:42,942
And then they will discuss why.
397
00:10:42,942 --> 00:10:45,611
So that's a very active way to
398
00:10:45,611 --> 00:10:48,514
engage them physically and mentally,
399
00:10:48,514 --> 00:10:49,148
to help them
400
00:10:49,148 --> 00:10:52,118
think critically about AI, prompt design.
401
00:10:52,118 --> 00:10:55,054
And also on the technology side,
402
00:10:55,054 --> 00:10:56,188
since I'm more of a
403
00:10:56,188 --> 00:10:59,191
technology designer developer,
404
00:10:59,191 --> 00:11:01,827
we developed a system called
405
00:11:01,827 --> 00:11:05,498
Toy Talk, which is a platform for kids
406
00:11:05,531 --> 00:11:09,301
to create their own AI powered chatbots
407
00:11:09,335 --> 00:11:11,070
that kind of like their
408
00:11:11,070 --> 00:11:13,139
AI version of their favorite toys.
409
00:11:13,139 --> 00:11:15,441
And so they got to define
410
00:11:15,441 --> 00:11:17,076
the job of their toy,
411
00:11:17,076 --> 00:11:18,778
the personality,
412
00:11:18,778 --> 00:11:20,513
the communication tone,
413
00:11:20,513 --> 00:11:22,782
and any rules or guidelines
414
00:11:22,782 --> 00:11:24,150
their toys should follow
415
00:11:24,150 --> 00:11:26,318
when they are talking to the users.
416
00:11:26,318 --> 00:11:30,322
So in this platform, we provide them
417
00:11:30,322 --> 00:11:33,859
with multiple ways to express themselves.
418
00:11:33,859 --> 00:11:36,395
So some students enjoy typing
419
00:11:36,395 --> 00:11:37,730
and you know
420
00:11:37,730 --> 00:11:40,032
they can just type the content
421
00:11:40,032 --> 00:11:42,201
when they are defining their toys,
422
00:11:42,201 --> 00:11:43,602
while others,
423
00:11:43,602 --> 00:11:46,305
especially those lower grade students,
424
00:11:46,305 --> 00:11:48,140
they will be much more comfortable
425
00:11:48,140 --> 00:11:49,175
with speaking.
426
00:11:49,175 --> 00:11:53,379
So we had those speech to text functions
427
00:11:53,512 --> 00:11:54,146
that they can
428
00:11:54,146 --> 00:11:55,281
simply press a button
429
00:11:55,281 --> 00:11:56,515
and say whatever
430
00:11:56,515 --> 00:11:58,350
they want their toy to talk about,
431
00:11:58,350 --> 00:12:00,019
and they can also
432
00:12:00,019 --> 00:12:02,955
upload the picture of their favorite toy,
433
00:12:02,955 --> 00:12:05,257
have the toy speak back to them,
434
00:12:05,257 --> 00:12:09,228
and define the voice and or communication
435
00:12:09,228 --> 00:12:10,763
style of their toy,
436
00:12:10,763 --> 00:12:12,531
which makes this
437
00:12:12,531 --> 00:12:13,699
experience feel
438
00:12:13,699 --> 00:12:16,302
much more personal and engaging.
439
00:12:16,302 --> 00:12:18,671
And so we build scaffolds
440
00:12:18,671 --> 00:12:21,907
to help students in critically about AI.
441
00:12:22,208 --> 00:12:23,943
You know, for example,
442
00:12:23,943 --> 00:12:27,747
we try to visualize the A's confidence
443
00:12:27,747 --> 00:12:30,316
level in each of the responses
444
00:12:30,316 --> 00:12:32,518
so that the students can see,
445
00:12:32,518 --> 00:12:32,852
oh,
446
00:12:32,852 --> 00:12:33,819
how confident
447
00:12:33,819 --> 00:12:36,589
the system thinks their response is.
448
00:12:36,589 --> 00:12:39,024
And we also included
449
00:12:39,024 --> 00:12:40,626
buttons for fact check
450
00:12:40,626 --> 00:12:41,560
so that
451
00:12:41,560 --> 00:12:43,763
if the students wants to compare this
452
00:12:43,763 --> 00:12:45,297
AI generated information
453
00:12:45,297 --> 00:12:47,299
with some external resources,
454
00:12:47,299 --> 00:12:48,667
they can do that.
455
00:12:48,667 --> 00:12:50,636
We also encourage students
456
00:12:50,636 --> 00:12:53,739
to compare the responses across different
457
00:12:53,773 --> 00:12:56,575
AI models, to see if the different
458
00:12:56,575 --> 00:12:58,310
AI models can produce
459
00:12:58,310 --> 00:13:00,379
same or different answer
460
00:13:00,379 --> 00:13:02,014
for the same questions.
461
00:13:02,014 --> 00:13:03,682
And so these features,
462
00:13:03,682 --> 00:13:04,917
these scaffolds
463
00:13:04,917 --> 00:13:06,652
become natural opportunities
464
00:13:06,652 --> 00:13:07,520
for students
465
00:13:07,520 --> 00:13:10,723
to teach them about how to evaluate
466
00:13:10,756 --> 00:13:12,491
AI instead of simply
467
00:13:12,491 --> 00:13:14,360
accepting all the answers.
468
00:13:14,360 --> 00:13:15,427
And of course, you know,
469
00:13:15,427 --> 00:13:17,630
designing these supports also comes
470
00:13:17,630 --> 00:13:19,064
with challenges.
471
00:13:19,064 --> 00:13:20,132
For example,
472
00:13:20,132 --> 00:13:23,402
one of our students said that the speech
473
00:13:23,402 --> 00:13:26,739
to text feature wasn't always accurate
474
00:13:26,839 --> 00:13:28,440
in capturing everything
475
00:13:28,440 --> 00:13:30,376
the students wanted to say.
476
00:13:30,376 --> 00:13:34,180
So we had to iterate on the interface to
477
00:13:34,213 --> 00:13:35,614
make it more reliable
478
00:13:35,614 --> 00:13:36,916
and more age appropriate
479
00:13:36,916 --> 00:13:38,984
for these students to use.
480
00:13:38,984 --> 00:13:40,986
And that's been an important lesson
481
00:13:40,986 --> 00:13:41,954
that we learned
482
00:13:41,954 --> 00:13:44,590
that effectively learning, AI learning
483
00:13:44,590 --> 00:13:47,893
design isn't just about adding features,
484
00:13:47,893 --> 00:13:49,195
it's about learning
485
00:13:49,195 --> 00:13:50,095
from students
486
00:13:50,095 --> 00:13:53,332
needs and be able to quickly adjust
487
00:13:53,465 --> 00:13:54,667
and modify
488
00:13:54,667 --> 00:13:56,869
the features to support these students
489
00:13:56,869 --> 00:13:58,737
with different learning needs.
490
00:13:58,737 --> 00:13:59,772
You spoke a little bit there
491
00:13:59,772 --> 00:14:01,841
about how their experience
492
00:14:01,841 --> 00:14:03,909
might react to the idea
493
00:14:03,909 --> 00:14:05,344
of AI getting something wrong,
494
00:14:05,344 --> 00:14:06,579
and I wanted to ask you
495
00:14:06,579 --> 00:14:08,147
about how you’re
496
00:14:08,147 --> 00:14:10,349
incorporating or responding
497
00:14:10,349 --> 00:14:11,917
to AI hallucinations.
498
00:14:11,917 --> 00:14:13,786
Like we hear a lot about these things,
499
00:14:13,786 --> 00:14:15,354
kind of just pulling information
500
00:14:15,354 --> 00:14:17,189
that maybe isn't accurate at all.
501
00:14:17,189 --> 00:14:20,025
How are you accommodating that?
502
00:14:20,025 --> 00:14:21,694
So we treat,
503
00:14:21,694 --> 00:14:24,196
you know, hallucinations in our system
504
00:14:24,196 --> 00:14:26,031
actually as a learning moment
505
00:14:26,031 --> 00:14:28,300
for them to be able to,
506
00:14:28,300 --> 00:14:30,135
you know, find out that
507
00:14:30,135 --> 00:14:32,171
AI actually make mistakes.
508
00:14:32,171 --> 00:14:33,272
And we treat that
509
00:14:33,272 --> 00:14:34,907
as a learning opportunity
510
00:14:34,907 --> 00:14:36,709
for them to realize,
511
00:14:36,709 --> 00:14:40,012
how would I detect hallucinations?
512
00:14:40,179 --> 00:14:41,547
What kind of strategies
513
00:14:41,547 --> 00:14:43,015
would I use to
514
00:14:43,015 --> 00:14:45,050
find out those hallucinations?
515
00:14:45,050 --> 00:14:47,786
And once I find out this hallucination,
516
00:14:47,786 --> 00:14:50,122
what should I do to mitigate that?
517
00:14:50,122 --> 00:14:51,957
So that's actually
518
00:14:51,957 --> 00:14:54,693
one of my favorite moments to observe.
519
00:14:54,693 --> 00:14:56,462
We see that students,
520
00:14:56,462 --> 00:14:56,795
you know,
521
00:14:56,795 --> 00:14:58,931
when they see the hallucination
522
00:14:58,931 --> 00:15:00,232
on the platform,
523
00:15:00,232 --> 00:15:03,535
they usually were surprised
524
00:15:03,535 --> 00:15:05,337
because they were expecting
525
00:15:05,337 --> 00:15:06,939
AI to know everything
526
00:15:06,939 --> 00:15:08,540
and be correct on everything.
527
00:15:08,540 --> 00:15:11,110
So when they see the AI, chatbots
528
00:15:11,110 --> 00:15:12,077
say wrong things
529
00:15:12,077 --> 00:15:14,079
like a wrong fact about the story
530
00:15:14,079 --> 00:15:17,816
that they are telling, they were confused
531
00:15:17,816 --> 00:15:20,819
because the response sounds so confident,
532
00:15:20,819 --> 00:15:23,155
but the fact was wrong.
533
00:15:23,155 --> 00:15:24,857
But very quickly
534
00:15:24,857 --> 00:15:26,091
their surprise
535
00:15:26,091 --> 00:15:28,961
actually turned to curiosity.
536
00:15:28,961 --> 00:15:30,596
So instead of asking oh,
537
00:15:30,596 --> 00:15:33,098
why is this AI so bad,
538
00:15:33,098 --> 00:15:35,467
they started thinking about
539
00:15:35,467 --> 00:15:37,303
why would the AI say that?
540
00:15:37,303 --> 00:15:40,439
And how can I make the AI better?
541
00:15:40,572 --> 00:15:42,708
And since they are building their own
542
00:15:42,708 --> 00:15:43,943
AI chatbots
543
00:15:43,943 --> 00:15:44,910
and so they're also
544
00:15:44,910 --> 00:15:46,812
experimenting AI outputs
545
00:15:46,812 --> 00:15:50,316
by refining their AI prompts, right?
546
00:15:50,349 --> 00:15:53,786
What we see often was that they would
547
00:15:53,819 --> 00:15:56,655
go back and rewrite their prompts.
548
00:15:56,655 --> 00:15:59,291
Some of them ask the same questions again
549
00:15:59,291 --> 00:16:00,492
to just to
550
00:16:00,492 --> 00:16:02,895
see if the AI would produce
551
00:16:02,895 --> 00:16:04,463
consistent answers,
552
00:16:04,463 --> 00:16:06,398
and some would add
553
00:16:06,398 --> 00:16:08,801
more specific instructions,
554
00:16:08,801 --> 00:16:11,704
and some actually went to upload
555
00:16:11,704 --> 00:16:13,572
better knowledge sources
556
00:16:13,572 --> 00:16:15,708
from those trustworthy websites.
557
00:16:15,708 --> 00:16:18,677
And they also compare their response
558
00:16:18,677 --> 00:16:21,680
from different models and say, oh,
559
00:16:21,780 --> 00:16:23,749
this model actually performs
560
00:16:23,749 --> 00:16:24,950
better than the other model.
561
00:16:24,950 --> 00:16:27,953
Maybe I would use this model more so
562
00:16:28,253 --> 00:16:31,490
we don't see these mistakes as failures.
563
00:16:31,590 --> 00:16:32,958
They start seeing them
564
00:16:32,958 --> 00:16:33,559
as something
565
00:16:33,559 --> 00:16:36,362
that they can investigate and improve.
566
00:16:36,362 --> 00:16:38,263
That's the kind of mindset
567
00:16:38,263 --> 00:16:39,031
that we are trying
568
00:16:39,031 --> 00:16:41,133
to teach the kids about.
569
00:16:41,133 --> 00:16:44,603
Not blindly trust the AI for everything
570
00:16:44,603 --> 00:16:48,240
or completely rejecting it, but knowing
571
00:16:48,774 --> 00:16:50,242
how to question it,
572
00:16:50,242 --> 00:16:51,744
how to investigate it,
573
00:16:51,744 --> 00:16:53,679
and how to improve it.
574
00:16:53,679 --> 00:16:57,583
So what's amazing was that these students
575
00:16:57,583 --> 00:16:59,318
actually started
576
00:16:59,318 --> 00:17:03,022
developing a habit of verifying AI’s
577
00:17:03,055 --> 00:17:04,790
answer whenever they see them,
578
00:17:04,790 --> 00:17:08,293
so they would actually go into
579
00:17:08,360 --> 00:17:09,395
their textbook
580
00:17:09,395 --> 00:17:10,295
or the books
581
00:17:10,295 --> 00:17:11,797
that they are talking about
582
00:17:11,797 --> 00:17:14,133
to cross-check those facts,
583
00:17:14,133 --> 00:17:15,834
and they will go on
584
00:17:15,834 --> 00:17:17,536
to those trustworthy websites
585
00:17:17,536 --> 00:17:19,271
to verify their answer.
586
00:17:19,271 --> 00:17:21,874
So when they are doing these practices,
587
00:17:21,874 --> 00:17:23,842
we see that there are also developing
588
00:17:23,842 --> 00:17:25,344
literacy skills.
589
00:17:25,344 --> 00:17:28,414
And what's fantastic about this
590
00:17:28,414 --> 00:17:30,015
is that they are actually
591
00:17:30,015 --> 00:17:31,717
intrinsically motivated
592
00:17:31,717 --> 00:17:33,719
to go read the book,
593
00:17:33,719 --> 00:17:34,820
not the teachers
594
00:17:34,820 --> 00:17:36,355
telling them read the book.
595
00:17:36,355 --> 00:17:39,258
They are motivated to go there
596
00:17:39,258 --> 00:17:40,659
and verify the answer.
597
00:17:40,659 --> 00:17:43,228
And that practice has improved
598
00:17:43,228 --> 00:17:44,897
significantly their reading
599
00:17:44,897 --> 00:17:46,065
and writing skills.
600
00:17:46,065 --> 00:17:46,865
I'd like to mention
601
00:17:46,865 --> 00:17:48,400
two things related to that,
602
00:17:48,400 --> 00:17:49,201
if that's all right.
603
00:17:49,201 --> 00:17:50,069
One is that I'm
604
00:17:50,069 --> 00:17:51,537
a mom of a third grader
605
00:17:51,537 --> 00:17:52,838
who just finished third grade,
606
00:17:52,838 --> 00:17:53,906
and is moving on to fourth.
607
00:17:53,906 --> 00:17:56,542
And, you know, with this, these ages,
608
00:17:56,542 --> 00:17:58,477
they start to have end of grade tests.
609
00:17:58,477 --> 00:18:01,080
And every parent is familiar with this,
610
00:18:01,080 --> 00:18:03,348
and the kids have to practice
611
00:18:03,348 --> 00:18:04,583
getting ready for those tests.
612
00:18:04,583 --> 00:18:05,984
And some of the things they have to do
613
00:18:05,984 --> 00:18:07,853
are the very things that Doctor Tian
614
00:18:07,853 --> 00:18:08,921
was telling you
615
00:18:08,921 --> 00:18:10,722
that the kids were inspired to do
616
00:18:10,722 --> 00:18:12,491
when they were making their own AIs.
617
00:18:12,491 --> 00:18:15,160
So the things the students need to do
618
00:18:15,160 --> 00:18:15,961
for the end of grade
619
00:18:15,961 --> 00:18:18,197
tests are tell what evidence
620
00:18:18,197 --> 00:18:19,731
in the text was,
621
00:18:19,731 --> 00:18:21,633
you know, supporting this idea.
622
00:18:21,633 --> 00:18:23,469
So the kids will be asked, like,
623
00:18:23,469 --> 00:18:23,669
you know,
624
00:18:23,669 --> 00:18:25,237
what is the main idea of this story?
625
00:18:25,237 --> 00:18:26,305
What is the main challenge
626
00:18:26,305 --> 00:18:27,873
for the main character?
627
00:18:27,873 --> 00:18:28,841
And then they have to
628
00:18:28,841 --> 00:18:29,908
actually give that evidence.
629
00:18:29,908 --> 00:18:32,144
So that extra motivation,
630
00:18:32,144 --> 00:18:32,778
when the children
631
00:18:32,778 --> 00:18:34,546
got to make their own chat bots
632
00:18:34,546 --> 00:18:36,515
and we inject hallucinations
633
00:18:36,515 --> 00:18:37,349
because models keep
634
00:18:37,349 --> 00:18:38,584
getting better and better
635
00:18:38,584 --> 00:18:40,152
so that students can learn
636
00:18:40,152 --> 00:18:41,587
that there can be hallucinations,
637
00:18:41,587 --> 00:18:43,555
we actually make sure that they happen
638
00:18:43,555 --> 00:18:44,623
so that they can have
639
00:18:44,623 --> 00:18:46,158
that experience of knowing
640
00:18:46,158 --> 00:18:47,359
that they need to check that
641
00:18:47,359 --> 00:18:48,760
there is such a thing as evidence,
642
00:18:48,760 --> 00:18:51,196
and you can go back and check that.
643
00:18:51,196 --> 00:18:54,099
And this is all really also very cool
644
00:18:54,099 --> 00:18:55,467
because we didn't
645
00:18:55,467 --> 00:18:56,702
originally plan
646
00:18:56,702 --> 00:18:58,637
to make chatbots with this district.
647
00:18:58,637 --> 00:19:00,172
This was research that Doctor
648
00:19:00,172 --> 00:19:02,508
Tian and I were already doing
649
00:19:02,508 --> 00:19:04,143
to make some chatbots.
650
00:19:04,143 --> 00:19:05,177
And then as
651
00:19:05,177 --> 00:19:06,178
we were working with the teachers,
652
00:19:06,178 --> 00:19:06,678
we said, hey,
653
00:19:06,678 --> 00:19:07,813
we'd like to show you this tool
654
00:19:07,813 --> 00:19:08,614
we're working on,
655
00:19:08,614 --> 00:19:09,915
and the teachers loved it.
656
00:19:09,915 --> 00:19:11,517
And then together
657
00:19:11,517 --> 00:19:13,652
in a research practice partnership,
658
00:19:13,652 --> 00:19:15,420
we worked together to design
659
00:19:15,420 --> 00:19:17,756
this toy talk experience for students,
660
00:19:17,756 --> 00:19:19,558
and the teachers helped implement
661
00:19:19,558 --> 00:19:20,425
it on the ground,
662
00:19:20,425 --> 00:19:22,261
like every day during a summer program.
663
00:19:22,261 --> 00:19:23,662
And this was a really inspiring
664
00:19:23,662 --> 00:19:24,596
part of this project.
665
00:19:24,596 --> 00:19:26,865
And I love how you're getting at critical
666
00:19:26,865 --> 00:19:28,066
thinking around
667
00:19:28,066 --> 00:19:29,468
computer information
668
00:19:29,468 --> 00:19:31,136
for kids at such an early age
669
00:19:31,136 --> 00:19:33,172
as they are going to go on to live
670
00:19:33,172 --> 00:19:35,474
even more online than we have.
671
00:19:35,474 --> 00:19:36,275
Just being able
672
00:19:36,275 --> 00:19:37,109
to think about what
673
00:19:37,109 --> 00:19:38,677
they're seeing or interacting with
674
00:19:38,677 --> 00:19:40,112
is really tremendous.
675
00:19:40,112 --> 00:19:42,381
So I want to pivot slightly here
676
00:19:42,381 --> 00:19:44,283
to kind of the NSF end of things.
677
00:19:44,283 --> 00:19:45,617
And you got an NSF
678
00:19:45,617 --> 00:19:47,252
grant to launch this project.
679
00:19:47,252 --> 00:19:48,120
And I understand
680
00:19:48,120 --> 00:19:50,355
you got to supplement to participate
681
00:19:50,355 --> 00:19:51,790
in the AI challenge portion.
682
00:19:51,790 --> 00:19:53,158
So can we talk a little bit
683
00:19:53,158 --> 00:19:55,794
about how that has impacted the project?
684
00:19:55,794 --> 00:19:56,361
Thank you.
685
00:19:56,361 --> 00:19:58,030
We were really excited to have a chance
686
00:19:58,030 --> 00:19:58,897
to work with the district
687
00:19:58,897 --> 00:19:59,665
on the presidential
688
00:19:59,665 --> 00:20:00,399
AI challenge,
689
00:20:00,399 --> 00:20:01,066
because we already
690
00:20:01,066 --> 00:20:02,501
had a group of six lead
691
00:20:02,501 --> 00:20:04,203
teachers who were working with us
692
00:20:04,203 --> 00:20:05,337
to learn about AI.
693
00:20:05,337 --> 00:20:06,638
So when we found out
694
00:20:06,638 --> 00:20:07,673
about the possibility
695
00:20:07,673 --> 00:20:09,741
to get a supplement for participating,
696
00:20:09,741 --> 00:20:11,076
we encouraged all the teachers
697
00:20:11,076 --> 00:20:11,610
to continue
698
00:20:11,610 --> 00:20:12,277
working with us
699
00:20:12,277 --> 00:20:13,478
to think about how they wanted
700
00:20:13,478 --> 00:20:15,280
to bring AI into their classroom.
701
00:20:15,280 --> 00:20:16,915
And then we scaffolded
702
00:20:16,915 --> 00:20:18,850
workshops where we help the teachers
703
00:20:18,850 --> 00:20:20,519
go through the presidential challenge
704
00:20:20,519 --> 00:20:21,520
materials, learn
705
00:20:21,520 --> 00:20:23,555
what it was about how to apply,
706
00:20:23,555 --> 00:20:25,224
and we supported that process
707
00:20:25,224 --> 00:20:27,492
for the teachers who wanted to do that.
708
00:20:27,492 --> 00:20:28,360
So we had
709
00:20:28,360 --> 00:20:28,760
sort of
710
00:20:28,760 --> 00:20:30,462
professional development workshops
711
00:20:30,462 --> 00:20:32,097
where we worked together with them.
712
00:20:32,097 --> 00:20:33,732
They would come up with ideas
713
00:20:33,732 --> 00:20:35,434
and tell them to each other
714
00:20:35,434 --> 00:20:37,536
and collaborate in our research team,
715
00:20:37,536 --> 00:20:39,137
you know, facilitated the teachers
716
00:20:39,137 --> 00:20:40,472
in the district to do that.
717
00:20:40,472 --> 00:20:42,574
And then as they were getting ready,
718
00:20:42,574 --> 00:20:43,775
we had practice sessions
719
00:20:43,775 --> 00:20:45,110
where the teachers
720
00:20:45,110 --> 00:20:47,112
would present their ideas to each other.
721
00:20:47,112 --> 00:20:48,680
We all gave each other feedback,
722
00:20:48,680 --> 00:20:50,949
and then they implemented their projects
723
00:20:50,949 --> 00:20:51,984
in their classrooms,
724
00:20:51,984 --> 00:20:53,819
and then brought those back to
725
00:20:53,819 --> 00:20:54,519
our teacher team
726
00:20:54,519 --> 00:20:55,721
and the research team
727
00:20:55,721 --> 00:20:57,356
to tell us about what happened.
728
00:20:57,356 --> 00:21:00,158
And then we all got to reflect on
729
00:21:00,158 --> 00:21:01,360
what the students were learning
730
00:21:01,360 --> 00:21:02,995
and thinking about
731
00:21:02,995 --> 00:21:04,396
what could be done next.
732
00:21:04,396 --> 00:21:05,664
And that was a
733
00:21:05,664 --> 00:21:07,065
really rewarding experience.
734
00:21:07,065 --> 00:21:09,034
And those extra supplements
735
00:21:09,034 --> 00:21:09,601
for stipends
736
00:21:09,601 --> 00:21:10,102
for teachers
737
00:21:10,102 --> 00:21:11,837
enabled us to help
738
00:21:11,837 --> 00:21:13,672
the teacher spend that extra time
739
00:21:13,672 --> 00:21:14,473
that they have to spend
740
00:21:14,473 --> 00:21:15,440
outside of the classroom
741
00:21:15,440 --> 00:21:18,277
to make a new activity plan,
742
00:21:18,277 --> 00:21:19,578
get the materials together
743
00:21:19,578 --> 00:21:20,879
and bring it to the classroom,
744
00:21:20,879 --> 00:21:22,147
and then collect all of that,
745
00:21:22,147 --> 00:21:22,381
you know,
746
00:21:22,381 --> 00:21:23,348
and create videos
747
00:21:23,348 --> 00:21:25,117
and content for that Presidential
748
00:21:25,117 --> 00:21:25,951
AI challenge.
749
00:21:25,951 --> 00:21:27,152
So, Professor Barnes,
750
00:21:27,152 --> 00:21:28,020
I want to ask you
751
00:21:28,020 --> 00:21:30,756
about the impact of NSF support
752
00:21:30,756 --> 00:21:31,390
over your career.
753
00:21:31,390 --> 00:21:32,491
You've got a number of grants
754
00:21:32,491 --> 00:21:32,991
over the years.
755
00:21:32,991 --> 00:21:34,293
Can you talk a little bit about what
756
00:21:34,293 --> 00:21:35,794
difference that has made for you?
757
00:21:35,794 --> 00:21:38,530
I'm extremely grateful and lucky
758
00:21:38,530 --> 00:21:40,532
to have gotten the amount of support
759
00:21:40,532 --> 00:21:41,700
over $26
760
00:21:41,700 --> 00:21:44,736
million over my career, and mostly
761
00:21:44,736 --> 00:21:45,771
it has all been
762
00:21:45,771 --> 00:21:48,440
because I have a strong belief
763
00:21:48,440 --> 00:21:50,309
that education is
764
00:21:50,309 --> 00:21:51,543
where we should be investing,
765
00:21:51,543 --> 00:21:53,345
but also in our research way,
766
00:21:53,345 --> 00:21:54,546
and also by building tools.
767
00:21:54,546 --> 00:21:55,981
So I'm a computer scientist
768
00:21:55,981 --> 00:21:57,883
and an educator in my heart,
769
00:21:57,883 --> 00:21:59,117
and so I'm always building
770
00:21:59,117 --> 00:22:00,319
educational technologies
771
00:22:00,319 --> 00:22:03,522
and AI to try to learn from data,
772
00:22:03,522 --> 00:22:05,590
to build systems that help students,
773
00:22:05,590 --> 00:22:07,225
but also to bring teachers
774
00:22:07,225 --> 00:22:09,394
into this sort of world of technology
775
00:22:09,394 --> 00:22:11,797
and how we can integrate computer science
776
00:22:11,797 --> 00:22:13,265
with education,
777
00:22:13,265 --> 00:22:15,467
but also to bring everyday people
778
00:22:15,467 --> 00:22:16,068
into learning
779
00:22:16,068 --> 00:22:17,336
about computer science and AI,
780
00:22:17,336 --> 00:22:18,704
because increasingly
781
00:22:18,704 --> 00:22:19,237
our lives
782
00:22:19,237 --> 00:22:20,572
have become more and more digital,
783
00:22:20,572 --> 00:22:21,540
and so everyone
784
00:22:21,540 --> 00:22:22,941
actually needs to have access.
785
00:22:22,941 --> 00:22:24,176
And that's
786
00:22:24,176 --> 00:22:24,743
really where
787
00:22:24,743 --> 00:22:26,211
the National Science Foundation
788
00:22:26,211 --> 00:22:30,549
has given me a platform to involve people
789
00:22:30,549 --> 00:22:33,552
from of all ages, from little kids
790
00:22:33,552 --> 00:22:34,419
up through,
791
00:22:34,419 --> 00:22:34,720
you know,
792
00:22:34,720 --> 00:22:37,723
K-12 college students, graduate students
793
00:22:37,723 --> 00:22:39,591
and faculty like myself
794
00:22:39,591 --> 00:22:41,026
and also my collaborators
795
00:22:41,026 --> 00:22:42,361
who are here and others.
796
00:22:42,361 --> 00:22:43,995
So I think that, you know,
797
00:22:43,995 --> 00:22:45,997
there's kind of a unified vision
798
00:22:45,997 --> 00:22:47,532
that many of us who apply
799
00:22:47,532 --> 00:22:48,400
for our National Science
800
00:22:48,400 --> 00:22:49,468
Foundation funding have
801
00:22:49,468 --> 00:22:50,068
is that
802
00:22:50,068 --> 00:22:51,403
we want to leverage
803
00:22:51,403 --> 00:22:54,506
not only the funds to not only advance
804
00:22:54,506 --> 00:22:56,375
research and technology,
805
00:22:56,375 --> 00:22:57,943
but also to bring
806
00:22:57,943 --> 00:22:59,711
that impact to other people.
807
00:22:59,711 --> 00:23:01,079
And I think you would see
808
00:23:01,079 --> 00:23:01,580
if you looked
809
00:23:01,580 --> 00:23:02,881
at my trajectory
810
00:23:02,881 --> 00:23:03,415
of projects
811
00:23:03,415 --> 00:23:04,015
over the years,
812
00:23:04,015 --> 00:23:04,616
that actually
813
00:23:04,616 --> 00:23:06,585
a large portion of the projects
814
00:23:06,585 --> 00:23:07,386
that I do
815
00:23:07,386 --> 00:23:08,553
have, this component
816
00:23:08,553 --> 00:23:10,622
of reaching out to teachers
817
00:23:10,622 --> 00:23:12,591
or students or the community
818
00:23:12,591 --> 00:23:14,426
and expanding access.
819
00:23:14,426 --> 00:23:16,061
And I'm super privileged
820
00:23:16,061 --> 00:23:17,295
to be able to do that.
821
00:23:17,295 --> 00:23:18,930
For my last question today,
822
00:23:18,930 --> 00:23:21,633
I want to kind of follow up with maybe
823
00:23:21,633 --> 00:23:22,834
a round robin question
824
00:23:22,834 --> 00:23:24,403
here for all of you.
825
00:23:24,403 --> 00:23:25,937
How did you feel
826
00:23:25,937 --> 00:23:28,173
or what was your reaction to the
827
00:23:28,173 --> 00:23:28,907
AI Challenge win
828
00:23:28,907 --> 00:23:30,308
coming out of this project?
829
00:23:30,308 --> 00:23:31,343
Very proud.
830
00:23:31,343 --> 00:23:35,280
And we see that what we have implemented
831
00:23:35,280 --> 00:23:36,214
in the district
832
00:23:36,214 --> 00:23:38,283
and our amazing teachers
833
00:23:38,283 --> 00:23:41,787
is when this Presidential Challenge were
834
00:23:42,120 --> 00:23:43,822
genuinely proud
835
00:23:43,822 --> 00:23:46,825
of what they did, what they accomplished,
836
00:23:46,858 --> 00:23:48,627
and the kind of impact
837
00:23:48,627 --> 00:23:50,462
that they are bringing
838
00:23:50,462 --> 00:23:51,463
to their students,
839
00:23:51,463 --> 00:23:52,431
to their families
840
00:23:52,431 --> 00:23:55,267
and to the district and to the nation.
841
00:23:55,267 --> 00:23:58,069
I think that it's very inspiring
842
00:23:58,069 --> 00:23:59,471
for everybody,
843
00:23:59,471 --> 00:24:00,739
for all the teachers
844
00:24:00,739 --> 00:24:03,742
to learn about their experiences,
845
00:24:03,742 --> 00:24:05,143
you know, implementing this
846
00:24:05,143 --> 00:24:06,945
AI lesson in our classrooms
847
00:24:06,945 --> 00:24:08,980
and other teachers,
848
00:24:08,980 --> 00:24:10,515
they can do the same thing
849
00:24:10,515 --> 00:24:12,851
and in their own classrooms.
850
00:24:12,851 --> 00:24:13,618
We are planning
851
00:24:13,618 --> 00:24:16,121
to share more about this experience,
852
00:24:16,121 --> 00:24:17,856
to empower
853
00:24:17,856 --> 00:24:20,025
other teachers to do the same thing
854
00:24:20,025 --> 00:24:21,193
in their own classroom.
855
00:24:21,193 --> 00:24:24,196
Yeah, and I echo that we are super proud
856
00:24:24,429 --> 00:24:27,132
about accomplishment of the teachers,
857
00:24:27,132 --> 00:24:27,899
not just Carrie,
858
00:24:27,899 --> 00:24:29,801
but also the other teacher teams
859
00:24:29,801 --> 00:24:31,136
who participated.
860
00:24:31,136 --> 00:24:32,337
It's a lot of effort
861
00:24:32,337 --> 00:24:33,071
to think about, to design
862
00:24:33,071 --> 00:24:34,039
the lesson,
863
00:24:34,039 --> 00:24:35,907
implement it, and also submit
864
00:24:35,907 --> 00:24:37,142
for the competition.
865
00:24:37,142 --> 00:24:38,810
So there are other teams
866
00:24:38,810 --> 00:24:40,812
also participating in this challenge
867
00:24:40,812 --> 00:24:43,582
is a big motivation for the district
868
00:24:43,582 --> 00:24:44,316
and the teachers
869
00:24:44,316 --> 00:24:45,984
that can be leaders in this space,
870
00:24:45,984 --> 00:24:47,052
and I think that we
871
00:24:47,052 --> 00:24:48,587
should give more space for teachers
872
00:24:48,587 --> 00:24:50,055
to demonstrate their own expertise
873
00:24:50,055 --> 00:24:50,856
in this area.
874
00:24:50,856 --> 00:24:51,756
And the teachers
875
00:24:51,756 --> 00:24:52,858
impact the teachers
876
00:24:52,858 --> 00:24:54,459
so they can see the impact
877
00:24:54,459 --> 00:24:55,994
on the district and the nation.
878
00:24:55,994 --> 00:24:57,229
So very, very proud.
879
00:24:57,229 --> 00:24:58,797
I just want to add on that
880
00:24:58,797 --> 00:25:02,133
I was actually in Montgomery in the PD
881
00:25:02,234 --> 00:25:03,668
in one of their summer
882
00:25:03,668 --> 00:25:06,037
teacher conference yesterday,
883
00:25:06,037 --> 00:25:09,074
and we talk about the
884
00:25:09,140 --> 00:25:10,242
things that we learned
885
00:25:10,242 --> 00:25:12,377
from the AI design challenge.
886
00:25:12,377 --> 00:25:15,480
And all the teachers in that same room
887
00:25:15,480 --> 00:25:16,414
said that
888
00:25:16,414 --> 00:25:19,217
I want to participate for the next year.
889
00:25:19,217 --> 00:25:20,719
So that's the power
890
00:25:20,719 --> 00:25:22,020
that this Presidential
891
00:25:22,020 --> 00:25:22,988
AI Challenge
892
00:25:22,988 --> 00:25:24,756
that brings so much impact
893
00:25:24,756 --> 00:25:26,258
to other teachers.
894
00:25:26,258 --> 00:25:27,826
Just to contrast two teachers
895
00:25:27,826 --> 00:25:29,461
we had in the presidential challenge.
896
00:25:29,461 --> 00:25:29,995
So Carrie
897
00:25:29,995 --> 00:25:31,396
Robledo was at the Friday
898
00:25:31,396 --> 00:25:33,431
Institute for a decade
899
00:25:33,431 --> 00:25:35,634
at least, and was working with
900
00:25:35,634 --> 00:25:37,035
actually teachers across the state,
901
00:25:37,035 --> 00:25:38,570
like helping them do digital learning.
902
00:25:38,570 --> 00:25:39,871
And she was the winner
903
00:25:39,871 --> 00:25:41,473
of the Presidential AI Challenge
904
00:25:41,473 --> 00:25:43,508
and did her second grade activity
905
00:25:43,508 --> 00:25:44,709
about using teachable
906
00:25:44,709 --> 00:25:46,111
machines with insects.
907
00:25:46,111 --> 00:25:47,112
And her students
908
00:25:47,112 --> 00:25:48,380
like understanding that
909
00:25:48,380 --> 00:25:49,681
even though you can train an AI
910
00:25:49,681 --> 00:25:51,116
to recognize different kind of insects,
911
00:25:51,116 --> 00:25:52,017
it can make a mistake.
912
00:25:52,017 --> 00:25:53,251
And then they started to learn
913
00:25:53,251 --> 00:25:55,253
that the data they put in is actually
914
00:25:55,253 --> 00:25:56,855
what the AI is learning from.
915
00:25:56,855 --> 00:25:58,323
But we had another teacher
916
00:25:58,323 --> 00:25:59,791
who this year was her first year
917
00:25:59,791 --> 00:26:01,660
doing anything like this.
918
00:26:01,660 --> 00:26:02,594
She was amazing.
919
00:26:02,594 --> 00:26:04,095
She was a first grade teacher,
920
00:26:04,095 --> 00:26:05,864
and she went above
921
00:26:05,864 --> 00:26:07,299
and beyond and was preparing
922
00:26:07,299 --> 00:26:08,133
for her Presidential
923
00:26:08,133 --> 00:26:09,935
AI Challenge, you know, ahead of time.
924
00:26:09,935 --> 00:26:11,236
And she was amazing.
925
00:26:11,236 --> 00:26:13,438
And, you know, collecting her
926
00:26:13,438 --> 00:26:14,272
students artwork
927
00:26:14,272 --> 00:26:16,074
about their imaginings, about AI
928
00:26:16,074 --> 00:26:17,309
and what they learned.
929
00:26:17,309 --> 00:26:18,977
And so we're seeing growth
930
00:26:18,977 --> 00:26:21,012
not just from teachers who are already
931
00:26:21,012 --> 00:26:23,315
well prepared to do this kind of work,
932
00:26:23,315 --> 00:26:24,716
but teachers who are just coming in
933
00:26:24,716 --> 00:26:25,717
and getting very excited
934
00:26:25,717 --> 00:26:27,519
because this is a new age
935
00:26:27,519 --> 00:26:29,154
and being able to get the chance
936
00:26:29,154 --> 00:26:31,056
to be involved with our team,
937
00:26:31,056 --> 00:26:32,057
but with other teachers
938
00:26:32,057 --> 00:26:33,224
who are excited to learn
939
00:26:33,224 --> 00:26:34,626
and then being able
940
00:26:34,626 --> 00:26:36,161
to share their progress on
941
00:26:36,161 --> 00:26:37,062
a national stage
942
00:26:37,062 --> 00:26:38,630
has made a big difference.
943
00:26:38,630 --> 00:26:40,332
Special thanks to Tiffany Barnes,
944
00:26:40,332 --> 00:26:42,334
Shiyan Jiang and Xiaoyi Tian.
945
00:26:42,334 --> 00:26:43,635
For the Discovery Files, I'm
946
00:26:43,635 --> 00:26:44,502
Nate Pottker.
947
00:26:44,502 --> 00:26:45,337
Watch video versions
948
00:26:45,337 --> 00:26:47,572
of these conversations on our @NSFscience
949
00:26:47,572 --> 00:26:48,440
YouTube channel.
950
00:26:48,440 --> 00:26:49,107
Please subscribe
951
00:26:49,107 --> 00:26:50,375
wherever you get podcasts
952
00:26:50,375 --> 00:26:50,942
and if you like
953
00:26:50,942 --> 00:26:52,477
our program, share with a friend
954
00:26:52,477 --> 00:26:54,179
and consider leaving a review.
955
00:26:56,214 --> 00:26:57,148
Discover how the U.S.
956
00:26:57,148 --> 00:26:58,316
National Science Foundation
957
00:26:58,316 --> 00:27:00,986
is advancing research at NSF.gov.
00:00:03,703 --> 00:00:05,538
This is the Discovery Files Podcast
2
00:00:05,538 --> 00:00:06,306
from the U.S.
3
00:00:06,306 --> 00:00:09,309
National Science Foundation.
4
00:00:09,776 --> 00:00:11,544
Building on a foundation of support
5
00:00:11,544 --> 00:00:12,345
for fundamental
6
00:00:12,345 --> 00:00:14,881
AI research, scientific discovery,
7
00:00:14,881 --> 00:00:16,516
advanced computing infrastructure,
8
00:00:16,516 --> 00:00:17,384
and the development
9
00:00:17,384 --> 00:00:19,052
of the American Stem workforce,
10
00:00:19,052 --> 00:00:21,154
NSF is committed to accelerating
11
00:00:21,154 --> 00:00:23,556
AI driven innovation that strengthens US.
12
00:00:23,556 --> 00:00:24,858
Scientific leadership,
13
00:00:24,858 --> 00:00:26,593
expands access to world class
14
00:00:26,593 --> 00:00:28,495
research infrastructure, and prepares
15
00:00:28,495 --> 00:00:29,295
the next generation
16
00:00:29,295 --> 00:00:30,597
of American science,
17
00:00:30,597 --> 00:00:32,032
technology, engineering,
18
00:00:32,032 --> 00:00:33,466
and mathematics talent.
19
00:00:33,466 --> 00:00:34,834
We are joined today by Tiffany
20
00:00:34,834 --> 00:00:35,835
Barnes, Shiyan
21
00:00:35,835 --> 00:00:37,737
Jiang and Xiaoyi Tian,
22
00:00:37,737 --> 00:00:39,706
NSF-supported researchers
23
00:00:39,706 --> 00:00:40,740
whose Elementary
24
00:00:40,740 --> 00:00:41,674
AI project
25
00:00:41,674 --> 00:00:42,742
works to prepare students
26
00:00:42,742 --> 00:00:44,210
for an AI driven future.
27
00:00:44,210 --> 00:00:45,111
Professor Barnes
28
00:00:45,111 --> 00:00:46,513
and Jiang and Doctor Tian,
29
00:00:46,513 --> 00:00:47,881
thank you so much for joining me today.
30
00:00:47,881 --> 00:00:49,115
Thank you for having us.
31
00:00:49,115 --> 00:00:50,216
So I'd like to start
32
00:00:50,216 --> 00:00:52,018
with the big definition.
33
00:00:52,018 --> 00:00:53,153
Professor Barnes, can you tell us
34
00:00:53,153 --> 00:00:55,555
what is the Elementary AI project?
35
00:00:55,555 --> 00:00:56,156
Elementary
36
00:00:56,156 --> 00:00:59,692
AI is an NSF funded computer Science
37
00:00:59,692 --> 00:01:00,960
for all research practice
38
00:01:00,960 --> 00:01:02,629
partnership between North
39
00:01:02,629 --> 00:01:04,130
Carolina State University
40
00:01:04,130 --> 00:01:05,231
Computer Science Department
41
00:01:05,231 --> 00:01:06,299
and the Friday Institute
42
00:01:06,299 --> 00:01:08,535
and Montgomery County Schools.
43
00:01:08,535 --> 00:01:10,336
We partnered to bring AI
44
00:01:10,336 --> 00:01:11,438
and computational thinking
45
00:01:11,438 --> 00:01:13,073
to all elementary students
46
00:01:13,073 --> 00:01:15,975
in all six schools in the district
47
00:01:15,975 --> 00:01:16,976
by working in partnership
48
00:01:16,976 --> 00:01:18,778
with teachers to learn AI,
49
00:01:18,778 --> 00:01:19,813
but also integrate it
50
00:01:19,813 --> 00:01:21,414
into the existing curriculum.
51
00:01:21,414 --> 00:01:23,383
And our goal is to help students
52
00:01:23,383 --> 00:01:25,351
improve on their end of year scores
53
00:01:25,351 --> 00:01:27,120
and improve overall excitement
54
00:01:27,120 --> 00:01:28,555
about learning in the district.
55
00:01:28,555 --> 00:01:31,024
Is AI usually introduced that early?
56
00:01:31,024 --> 00:01:32,225
The general public probably doesn't
57
00:01:32,225 --> 00:01:34,561
really understand how early computer
58
00:01:34,561 --> 00:01:35,695
things are being introduced
59
00:01:35,695 --> 00:01:37,197
to kids at this point in time.
60
00:01:37,197 --> 00:01:38,498
AI is not introduced
61
00:01:38,498 --> 00:01:40,033
very often in elementary school,
62
00:01:40,033 --> 00:01:41,234
but there are a lot of ideas
63
00:01:41,234 --> 00:01:42,035
in AI
64
00:01:42,035 --> 00:01:42,802
and computational
65
00:01:42,802 --> 00:01:44,504
thinking that can be introduced
66
00:01:44,504 --> 00:01:45,939
when kids are very young.
67
00:01:45,939 --> 00:01:47,540
So we focus on concepts
68
00:01:47,540 --> 00:01:49,142
like computational thinking,
69
00:01:49,142 --> 00:01:51,144
which includes pattern recognition,
70
00:01:51,144 --> 00:01:53,113
abstraction, decomposition.
71
00:01:53,113 --> 00:01:54,080
Kids are already doing
72
00:01:54,080 --> 00:01:55,081
these things in school,
73
00:01:55,081 --> 00:01:56,216
and so we think it's
74
00:01:56,216 --> 00:01:57,584
a very powerful partnership
75
00:01:57,584 --> 00:01:59,953
to work with elementary school teachers
76
00:01:59,953 --> 00:02:01,554
while the kids are learning things,
77
00:02:01,554 --> 00:02:01,821
you know,
78
00:02:01,821 --> 00:02:02,522
when they learn,
79
00:02:02,522 --> 00:02:03,656
like what are shapes,
80
00:02:03,656 --> 00:02:05,358
even when they're really small.
81
00:02:05,358 --> 00:02:07,260
And that's a pattern recognition task.
82
00:02:07,260 --> 00:02:08,495
And AI researchers
83
00:02:08,495 --> 00:02:11,030
have always been inspired by human
84
00:02:11,030 --> 00:02:12,499
learning, and kids are learning,
85
00:02:12,499 --> 00:02:13,399
and they know that.
86
00:02:13,399 --> 00:02:15,468
And teachers can hook
87
00:02:15,468 --> 00:02:17,871
those ideas about AI learning
88
00:02:17,871 --> 00:02:19,772
to what kids are learning anytime.
89
00:02:19,772 --> 00:02:20,874
For a follow up question,
90
00:02:20,874 --> 00:02:22,475
a little bit of background.
91
00:02:22,475 --> 00:02:24,377
How did this project come together?
92
00:02:24,377 --> 00:02:25,812
It grew out of a decade long
93
00:02:25,812 --> 00:02:27,380
foundation of trust between North
94
00:02:27,380 --> 00:02:29,115
Carolina State University,
95
00:02:29,115 --> 00:02:30,183
the Friday Institute,
96
00:02:30,183 --> 00:02:31,951
which has been working across the state
97
00:02:31,951 --> 00:02:33,052
to help promote
98
00:02:33,052 --> 00:02:34,654
educational innovation
99
00:02:34,654 --> 00:02:36,923
and outcomes, and Montgomery
100
00:02:36,923 --> 00:02:37,390
County Schools,
101
00:02:37,390 --> 00:02:38,691
has been partnering a long time
102
00:02:38,691 --> 00:02:40,760
with NC State as well.
103
00:02:40,760 --> 00:02:44,397
So in 2015, the Friday Institute
104
00:02:44,430 --> 00:02:46,666
helped with 1 to 1 tech
105
00:02:46,666 --> 00:02:48,601
integration and learner agency
106
00:02:48,601 --> 00:02:49,302
initiatives.
107
00:02:49,302 --> 00:02:50,336
Carrie Robledo,
108
00:02:50,336 --> 00:02:52,705
who is in the district now as a teacher.
109
00:02:52,705 --> 00:02:54,674
She was at the Friday Institute
110
00:02:54,674 --> 00:02:56,342
and she was a digital learning coach.
111
00:02:56,342 --> 00:02:56,976
And Joanna
112
00:02:56,976 --> 00:02:59,913
Perkins is one of the directors for K-12
113
00:02:59,913 --> 00:03:01,114
curriculum in the district.
114
00:03:01,114 --> 00:03:03,316
And she partnered with us to co-design
115
00:03:03,316 --> 00:03:04,350
this partnership
116
00:03:04,350 --> 00:03:05,418
as they were introducing
117
00:03:05,418 --> 00:03:07,387
the new CKLA curriculum.
118
00:03:07,387 --> 00:03:07,787
We felt
119
00:03:07,787 --> 00:03:09,756
it would be really advantageous time
120
00:03:09,756 --> 00:03:11,024
to also bring in AI
121
00:03:11,024 --> 00:03:11,858
to help the teachers
122
00:03:11,858 --> 00:03:13,059
and the students get excited
123
00:03:13,059 --> 00:03:15,195
about this new age of AI.
124
00:03:15,195 --> 00:03:16,162
So, Professor Jiang,
125
00:03:16,162 --> 00:03:17,997
I'd like to ask you a question here.
126
00:03:17,997 --> 00:03:19,599
How have you
127
00:03:19,599 --> 00:03:20,867
or how has this project
128
00:03:20,867 --> 00:03:21,601
really worked
129
00:03:21,601 --> 00:03:22,835
to empower the teachers
130
00:03:22,835 --> 00:03:23,336
to become
131
00:03:23,336 --> 00:03:25,271
AI educators, to adopt
132
00:03:25,271 --> 00:03:27,307
AI concepts into their curriculum?
133
00:03:27,307 --> 00:03:29,075
One of the biggest misconception
134
00:03:29,075 --> 00:03:29,676
that teachers
135
00:03:29,676 --> 00:03:30,710
need to become
136
00:03:30,710 --> 00:03:33,713
like AI experts before they can teach AI.
137
00:03:33,746 --> 00:03:36,382
But our philosophy is very different.
138
00:03:36,382 --> 00:03:38,351
We help teachers like what Dr Barnes
139
00:03:38,351 --> 00:03:39,118
just mentioned.
140
00:03:39,118 --> 00:03:39,953
We help teachers
141
00:03:39,953 --> 00:03:42,055
recognize that many of the skills
142
00:03:42,055 --> 00:03:43,856
they already teach in classrooms
143
00:03:43,856 --> 00:03:45,225
such as identifying
144
00:03:45,225 --> 00:03:47,493
patterns, asking good questions,
145
00:03:47,493 --> 00:03:49,862
interpreting evidence, and discussing
146
00:03:49,862 --> 00:03:51,130
ethical issues
147
00:03:51,130 --> 00:03:53,333
that are very central to AI literacy.
148
00:03:53,333 --> 00:03:55,501
So we build on teachers
149
00:03:55,501 --> 00:03:57,136
existing experiences
150
00:03:57,136 --> 00:03:58,938
and also teachers, they receive
151
00:03:58,938 --> 00:04:02,041
ongoing coaching, proper development,
152
00:04:02,075 --> 00:04:03,309
collaborate with lead
153
00:04:03,309 --> 00:04:04,811
teachers, share lesson
154
00:04:04,811 --> 00:04:06,546
ideas, become leaders
155
00:04:06,546 --> 00:04:08,114
who support colleagues
156
00:04:08,114 --> 00:04:09,749
across the district.
157
00:04:09,749 --> 00:04:12,151
So the goal is to not to like it,
158
00:04:12,151 --> 00:04:14,621
simply like training individual teachers.
159
00:04:14,621 --> 00:04:15,455
Our goal is to
160
00:04:15,455 --> 00:04:17,223
build a sustainable community
161
00:04:17,223 --> 00:04:19,559
of AI educators for the district.
162
00:04:19,559 --> 00:04:21,094
I'm thinking about some of the challenges
163
00:04:21,094 --> 00:04:22,095
that might come in here
164
00:04:22,095 --> 00:04:24,230
in the broader district especially,
165
00:04:24,230 --> 00:04:26,266
and thinking about getting the tech
166
00:04:26,266 --> 00:04:27,300
to the teachers
167
00:04:27,300 --> 00:04:28,101
maybe, maybe
168
00:04:28,101 --> 00:04:30,136
getting school boards on to the project.
169
00:04:30,136 --> 00:04:31,537
Are there limitations
170
00:04:31,537 --> 00:04:32,939
to the curriculum involved?
171
00:04:32,939 --> 00:04:34,674
Can you talk about some of the challenges
172
00:04:34,674 --> 00:04:36,175
in getting this implemented?
173
00:04:36,175 --> 00:04:39,178
So in terms of curriculum limitations,
174
00:04:39,178 --> 00:04:41,814
we worked very closely with teachers
175
00:04:41,814 --> 00:04:44,183
to identify the natural connection
176
00:04:44,183 --> 00:04:46,853
with their existing curriculum, like Dr.
177
00:04:46,853 --> 00:04:48,388
Barnes mentioned about CKLA
178
00:04:48,388 --> 00:04:49,656
which is a new literacy
179
00:04:49,656 --> 00:04:50,857
curriculum that they
180
00:04:50,857 --> 00:04:53,026
were adopting this last year.
181
00:04:53,026 --> 00:04:56,829
So that's not a really major limitation.
182
00:04:56,829 --> 00:04:58,331
But the biggest challenge
183
00:04:58,331 --> 00:04:59,098
that I saw
184
00:04:59,098 --> 00:05:00,900
is that a helping people
185
00:05:00,900 --> 00:05:01,267
see that
186
00:05:01,267 --> 00:05:03,936
AI belongs in elementary education,
187
00:05:03,936 --> 00:05:05,171
and it can fit naturally
188
00:05:05,171 --> 00:05:06,906
into existing classrooms.
189
00:05:06,906 --> 00:05:08,708
Teachers may and initially
190
00:05:08,708 --> 00:05:10,276
before we came in,
191
00:05:10,276 --> 00:05:11,377
they may worry that
192
00:05:11,377 --> 00:05:12,979
AI is too technical
193
00:05:12,979 --> 00:05:15,748
or too advanced for young kids.
194
00:05:15,748 --> 00:05:16,549
With many kids,
195
00:05:16,549 --> 00:05:17,717
they don't even have
196
00:05:17,717 --> 00:05:18,217
that many
197
00:05:18,217 --> 00:05:18,985
digital literacy
198
00:05:18,985 --> 00:05:20,553
skills and others
199
00:05:20,553 --> 00:05:21,554
might concern
200
00:05:21,554 --> 00:05:22,388
that teaching
201
00:05:22,388 --> 00:05:24,857
AI means adding another subject
202
00:05:24,857 --> 00:05:25,692
to a already
203
00:05:25,692 --> 00:05:27,026
very full curriculum,
204
00:05:27,026 --> 00:05:28,761
on top of the important
205
00:05:28,761 --> 00:05:30,630
skills like math and EOA.
206
00:05:30,630 --> 00:05:32,465
So our approach addresses
207
00:05:32,465 --> 00:05:34,200
those challenges or concerns
208
00:05:34,200 --> 00:05:36,035
by integrating AI into lessons
209
00:05:36,035 --> 00:05:37,303
teachers already teaching.
210
00:05:37,303 --> 00:05:40,273
And we are not asking to replace
211
00:05:40,273 --> 00:05:41,307
reading or mass.
212
00:05:41,307 --> 00:05:42,742
We reach those subjects
213
00:05:42,742 --> 00:05:44,410
with AI literacy skills.
214
00:05:44,410 --> 00:05:45,878
So like the example,
215
00:05:45,878 --> 00:05:47,980
these can learn about pattern recognition
216
00:05:47,980 --> 00:05:49,949
while they study birds
217
00:05:49,949 --> 00:05:51,150
doing a science activity.
218
00:05:51,150 --> 00:05:52,251
Because birds might have
219
00:05:52,251 --> 00:05:53,453
a certain features
220
00:05:53,453 --> 00:05:55,221
that can be patterns and doesn't
221
00:05:55,221 --> 00:05:56,823
connect to AI concepts.
222
00:05:56,823 --> 00:05:58,291
And another challenge
223
00:05:58,291 --> 00:05:59,158
I will say
224
00:05:59,158 --> 00:06:01,160
is a long term sustainability,
225
00:06:01,160 --> 00:06:01,794
which is something
226
00:06:01,794 --> 00:06:03,296
we are working on very hard.
227
00:06:03,296 --> 00:06:03,629
That's why
228
00:06:03,629 --> 00:06:04,731
we work very closely
229
00:06:04,731 --> 00:06:06,833
with district leaders and teachers
230
00:06:06,833 --> 00:06:08,334
through what we call
231
00:06:08,334 --> 00:06:09,836
research practice, partnership.
232
00:06:09,836 --> 00:06:11,938
And through this kind of partnership.
233
00:06:11,938 --> 00:06:13,473
We hope that AI
234
00:06:13,473 --> 00:06:14,574
continues long
235
00:06:14,574 --> 00:06:16,576
after the research project ends.
236
00:06:16,576 --> 00:06:17,510
I'd like to add
237
00:06:17,510 --> 00:06:18,411
that the district
238
00:06:18,411 --> 00:06:21,314
has been a real partner in this project.
239
00:06:21,314 --> 00:06:22,982
So not only have the teachers
240
00:06:22,982 --> 00:06:23,916
been real partners
241
00:06:23,916 --> 00:06:26,419
bringing new AI to their students
242
00:06:26,419 --> 00:06:27,653
much of the school year,
243
00:06:27,653 --> 00:06:29,789
but the district administration
244
00:06:29,789 --> 00:06:31,491
has been working with us.
245
00:06:31,491 --> 00:06:31,791
You know,
246
00:06:31,791 --> 00:06:32,191
we meet
247
00:06:32,191 --> 00:06:33,259
at least every month,
248
00:06:33,259 --> 00:06:34,560
and sometimes
249
00:06:34,560 --> 00:06:34,894
we meet
250
00:06:34,894 --> 00:06:36,162
with the superintendent,
251
00:06:36,162 --> 00:06:37,463
and sometimes we meet
252
00:06:37,463 --> 00:06:39,432
with the curriculum specialist.
253
00:06:39,432 --> 00:06:41,901
And the fact that we have these embedded
254
00:06:41,901 --> 00:06:43,035
teachers in the districts mean
255
00:06:43,035 --> 00:06:43,936
that we can regularly
256
00:06:43,936 --> 00:06:45,171
find out what's going on.
257
00:06:45,171 --> 00:06:45,972
And then
258
00:06:45,972 --> 00:06:47,573
we also get to hear, like,
259
00:06:47,573 --> 00:06:48,674
what are the important problems
260
00:06:48,674 --> 00:06:50,042
that the district is trying to solve.
261
00:06:50,042 --> 00:06:53,413
And this is a strong type of research
262
00:06:53,413 --> 00:06:54,714
that NSF is funding
263
00:06:54,714 --> 00:06:56,349
these research practice partnerships.
264
00:06:56,349 --> 00:06:58,618
And it makes me feel wonderful
265
00:06:58,618 --> 00:06:59,719
to work on this kind of thing,
266
00:06:59,719 --> 00:07:01,387
because I don't feel
267
00:07:01,387 --> 00:07:01,788
that I'm
268
00:07:01,788 --> 00:07:04,390
just bringing an extra thing to schools,
269
00:07:04,390 --> 00:07:06,192
but I feel that we are actually helping
270
00:07:06,192 --> 00:07:06,893
solve problems
271
00:07:06,893 --> 00:07:08,661
that the district wants to solve,
272
00:07:08,661 --> 00:07:09,662
and the initiatives
273
00:07:09,662 --> 00:07:12,665
that they are interested in doing.
274
00:07:12,832 --> 00:07:15,301
This partnership enables us to be able
275
00:07:15,301 --> 00:07:16,602
to listen, to be able
276
00:07:16,602 --> 00:07:19,639
to brainstorm solutions to problems
277
00:07:19,639 --> 00:07:21,641
in the school district or just,
278
00:07:21,641 --> 00:07:22,909
you know, desires.
279
00:07:22,909 --> 00:07:24,343
I'm not saying there are problems.
280
00:07:24,343 --> 00:07:26,145
I'm saying that, you know,
281
00:07:26,145 --> 00:07:28,214
whatever those priorities are,
282
00:07:28,214 --> 00:07:29,782
since we're there as partners,
283
00:07:29,782 --> 00:07:31,784
we actually get to talk together about
284
00:07:31,784 --> 00:07:33,753
what do we want to be excited about?
285
00:07:33,753 --> 00:07:34,854
What do we want to help other people
286
00:07:34,854 --> 00:07:35,521
be excited about?
287
00:07:35,521 --> 00:07:37,023
What do we want to help accomplish?
288
00:07:37,023 --> 00:07:38,758
And I think that it's key
289
00:07:38,758 --> 00:07:39,358
not only that
290
00:07:39,358 --> 00:07:40,359
the teachers are involved,
291
00:07:40,359 --> 00:07:41,260
but also that
292
00:07:41,260 --> 00:07:43,429
the administration is on board.
293
00:07:43,429 --> 00:07:43,663
You know,
294
00:07:43,663 --> 00:07:45,298
they're thinking about buying computers
295
00:07:45,298 --> 00:07:46,098
for every kid,
296
00:07:46,098 --> 00:07:47,500
and there's new initiatives
297
00:07:47,500 --> 00:07:49,435
across the state about,
298
00:07:49,435 --> 00:07:50,203
you know, determining,
299
00:07:50,203 --> 00:07:51,304
like how much screen
300
00:07:51,304 --> 00:07:52,505
time should kids have?
301
00:07:52,505 --> 00:07:53,940
So we had discussions
302
00:07:53,940 --> 00:07:55,041
in some of our meetings,
303
00:07:55,041 --> 00:07:56,375
like, what do researchers
304
00:07:56,375 --> 00:07:56,943
think about that?
305
00:07:56,943 --> 00:07:57,677
What does research
306
00:07:57,677 --> 00:07:59,612
say about how much screen time
307
00:07:59,612 --> 00:08:00,613
children should have?
308
00:08:00,613 --> 00:08:03,349
And we all had discussions and agreed
309
00:08:03,349 --> 00:08:05,184
that screens can be useful,
310
00:08:05,184 --> 00:08:06,586
especially when you get to do something
311
00:08:06,586 --> 00:08:09,222
like chatbots, Toy Talk that Dr.
312
00:08:09,222 --> 00:08:10,189
Tian and Dr.
313
00:08:10,189 --> 00:08:12,058
Jiang were bringing to the project.
314
00:08:12,058 --> 00:08:13,059
But you know,
315
00:08:13,059 --> 00:08:14,694
when they're not doing things like that,
316
00:08:14,694 --> 00:08:15,661
if they're just doing something
317
00:08:15,661 --> 00:08:16,596
that could have been done
318
00:08:16,596 --> 00:08:17,530
by reading a book
319
00:08:17,530 --> 00:08:19,265
or doing a worksheet,
320
00:08:19,265 --> 00:08:21,200
or doing something kinesthetic,
321
00:08:21,200 --> 00:08:22,502
where they're moving around the room
322
00:08:22,502 --> 00:08:23,803
and trying to make an analogy
323
00:08:23,803 --> 00:08:25,838
between how their physical body works
324
00:08:25,838 --> 00:08:27,306
and how some concept
325
00:08:27,306 --> 00:08:28,307
that they're learning in math
326
00:08:28,307 --> 00:08:31,477
or reading actually can be related to.
327
00:08:31,511 --> 00:08:32,078
There's different
328
00:08:32,078 --> 00:08:33,513
needs for different schools,
329
00:08:33,513 --> 00:08:34,947
different kids, different teachers,
330
00:08:34,947 --> 00:08:36,682
and being able to work together
331
00:08:36,682 --> 00:08:37,550
in this practice
332
00:08:37,550 --> 00:08:40,019
partnership means that we can address
333
00:08:40,019 --> 00:08:41,387
those as they come up.
334
00:08:41,387 --> 00:08:43,322
Doctor Tian, I had a question
335
00:08:43,322 --> 00:08:44,457
that kind of came out of here
336
00:08:44,457 --> 00:08:45,658
that you wrote a paper about
337
00:08:45,658 --> 00:08:47,393
and thinking about accommodating
338
00:08:47,393 --> 00:08:48,561
different learning styles.
339
00:08:48,561 --> 00:08:49,562
Can you talk a little bit
340
00:08:49,562 --> 00:08:50,830
about the approach
341
00:08:50,830 --> 00:08:53,199
to maybe somebody like myself?
342
00:08:53,199 --> 00:08:54,233
I've learned better
343
00:08:54,233 --> 00:08:55,902
with hearing in school
344
00:08:55,902 --> 00:08:58,905
and reading it as opposed to being shown.
345
00:08:58,938 --> 00:08:59,839
Can you talk a little bit
346
00:08:59,839 --> 00:09:01,073
about these kind of concepts
347
00:09:01,073 --> 00:09:02,341
and how you approach them?
348
00:09:02,341 --> 00:09:05,077
When we are working with the schools,
349
00:09:05,077 --> 00:09:07,313
one of the biggest challenges
350
00:09:07,313 --> 00:09:08,581
that we have seen
351
00:09:08,581 --> 00:09:10,182
is how do we design
352
00:09:10,182 --> 00:09:12,652
AI learning experiences
353
00:09:12,652 --> 00:09:15,221
when students come into the classroom
354
00:09:15,221 --> 00:09:16,923
with very different needs,
355
00:09:16,923 --> 00:09:17,323
you know,
356
00:09:17,323 --> 00:09:20,426
they have very different reading levels.
357
00:09:20,459 --> 00:09:22,628
They have different confidence
358
00:09:22,628 --> 00:09:26,232
level or trust level with AI technologies
359
00:09:26,232 --> 00:09:27,800
and different ways
360
00:09:27,800 --> 00:09:29,302
that they prefer to learn.
361
00:09:29,302 --> 00:09:32,538
So we are really thinking about,
362
00:09:32,572 --> 00:09:32,838
you know,
363
00:09:32,838 --> 00:09:34,240
the kinds of scaffolds
364
00:09:34,240 --> 00:09:35,341
that we can provide,
365
00:09:35,341 --> 00:09:36,943
both in the lessons
366
00:09:36,943 --> 00:09:40,346
and in the technology itself.
367
00:09:40,513 --> 00:09:42,615
On the instructional side,
368
00:09:42,615 --> 00:09:45,585
we try to provide multiple ways
369
00:09:45,585 --> 00:09:48,120
for students to engage with the content.
370
00:09:48,120 --> 00:09:49,855
So, for example,
371
00:09:49,855 --> 00:09:51,958
instead of only asking students
372
00:09:51,958 --> 00:09:54,694
to read the story or read the book,
373
00:09:54,694 --> 00:09:57,797
we gave the option for them to read
374
00:09:57,797 --> 00:09:59,832
or listen to the materials.
375
00:09:59,832 --> 00:10:02,401
And we also designed some
376
00:10:02,401 --> 00:10:04,370
unplugged activities
377
00:10:04,370 --> 00:10:05,237
where students
378
00:10:05,237 --> 00:10:06,772
learn about computing concepts
379
00:10:06,772 --> 00:10:08,174
without using computers
380
00:10:08,174 --> 00:10:09,709
or be on the screen.
381
00:10:09,709 --> 00:10:12,712
So one activity that we ask students
382
00:10:12,812 --> 00:10:15,715
when we are teaching this AI concept, one
383
00:10:15,715 --> 00:10:18,384
important concept is to understand
384
00:10:18,384 --> 00:10:19,719
effective prompts.
385
00:10:19,719 --> 00:10:22,021
When you're talking with AI chatbots,
386
00:10:22,021 --> 00:10:22,588
how do you
387
00:10:22,588 --> 00:10:24,657
construct a good prompt
388
00:10:24,657 --> 00:10:27,259
so that you can get your desired answer?
389
00:10:27,259 --> 00:10:29,562
So we had an activity
390
00:10:29,562 --> 00:10:32,732
that asks students to compare different
391
00:10:32,798 --> 00:10:33,866
AI prompts
392
00:10:33,866 --> 00:10:36,435
so they would move to different sides
393
00:10:36,435 --> 00:10:37,436
of the classroom
394
00:10:37,436 --> 00:10:39,038
to vote, which prompt
395
00:10:39,038 --> 00:10:40,773
they think is more effective.
396
00:10:40,773 --> 00:10:42,942
And then they will discuss why.
397
00:10:42,942 --> 00:10:45,611
So that's a very active way to
398
00:10:45,611 --> 00:10:48,514
engage them physically and mentally,
399
00:10:48,514 --> 00:10:49,148
to help them
400
00:10:49,148 --> 00:10:52,118
think critically about AI, prompt design.
401
00:10:52,118 --> 00:10:55,054
And also on the technology side,
402
00:10:55,054 --> 00:10:56,188
since I'm more of a
403
00:10:56,188 --> 00:10:59,191
technology designer developer,
404
00:10:59,191 --> 00:11:01,827
we developed a system called
405
00:11:01,827 --> 00:11:05,498
Toy Talk, which is a platform for kids
406
00:11:05,531 --> 00:11:09,301
to create their own AI powered chatbots
407
00:11:09,335 --> 00:11:11,070
that kind of like their
408
00:11:11,070 --> 00:11:13,139
AI version of their favorite toys.
409
00:11:13,139 --> 00:11:15,441
And so they got to define
410
00:11:15,441 --> 00:11:17,076
the job of their toy,
411
00:11:17,076 --> 00:11:18,778
the personality,
412
00:11:18,778 --> 00:11:20,513
the communication tone,
413
00:11:20,513 --> 00:11:22,782
and any rules or guidelines
414
00:11:22,782 --> 00:11:24,150
their toys should follow
415
00:11:24,150 --> 00:11:26,318
when they are talking to the users.
416
00:11:26,318 --> 00:11:30,322
So in this platform, we provide them
417
00:11:30,322 --> 00:11:33,859
with multiple ways to express themselves.
418
00:11:33,859 --> 00:11:36,395
So some students enjoy typing
419
00:11:36,395 --> 00:11:37,730
and you know
420
00:11:37,730 --> 00:11:40,032
they can just type the content
421
00:11:40,032 --> 00:11:42,201
when they are defining their toys,
422
00:11:42,201 --> 00:11:43,602
while others,
423
00:11:43,602 --> 00:11:46,305
especially those lower grade students,
424
00:11:46,305 --> 00:11:48,140
they will be much more comfortable
425
00:11:48,140 --> 00:11:49,175
with speaking.
426
00:11:49,175 --> 00:11:53,379
So we had those speech to text functions
427
00:11:53,512 --> 00:11:54,146
that they can
428
00:11:54,146 --> 00:11:55,281
simply press a button
429
00:11:55,281 --> 00:11:56,515
and say whatever
430
00:11:56,515 --> 00:11:58,350
they want their toy to talk about,
431
00:11:58,350 --> 00:12:00,019
and they can also
432
00:12:00,019 --> 00:12:02,955
upload the picture of their favorite toy,
433
00:12:02,955 --> 00:12:05,257
have the toy speak back to them,
434
00:12:05,257 --> 00:12:09,228
and define the voice and or communication
435
00:12:09,228 --> 00:12:10,763
style of their toy,
436
00:12:10,763 --> 00:12:12,531
which makes this
437
00:12:12,531 --> 00:12:13,699
experience feel
438
00:12:13,699 --> 00:12:16,302
much more personal and engaging.
439
00:12:16,302 --> 00:12:18,671
And so we build scaffolds
440
00:12:18,671 --> 00:12:21,907
to help students in critically about AI.
441
00:12:22,208 --> 00:12:23,943
You know, for example,
442
00:12:23,943 --> 00:12:27,747
we try to visualize the A's confidence
443
00:12:27,747 --> 00:12:30,316
level in each of the responses
444
00:12:30,316 --> 00:12:32,518
so that the students can see,
445
00:12:32,518 --> 00:12:32,852
oh,
446
00:12:32,852 --> 00:12:33,819
how confident
447
00:12:33,819 --> 00:12:36,589
the system thinks their response is.
448
00:12:36,589 --> 00:12:39,024
And we also included
449
00:12:39,024 --> 00:12:40,626
buttons for fact check
450
00:12:40,626 --> 00:12:41,560
so that
451
00:12:41,560 --> 00:12:43,763
if the students wants to compare this
452
00:12:43,763 --> 00:12:45,297
AI generated information
453
00:12:45,297 --> 00:12:47,299
with some external resources,
454
00:12:47,299 --> 00:12:48,667
they can do that.
455
00:12:48,667 --> 00:12:50,636
We also encourage students
456
00:12:50,636 --> 00:12:53,739
to compare the responses across different
457
00:12:53,773 --> 00:12:56,575
AI models, to see if the different
458
00:12:56,575 --> 00:12:58,310
AI models can produce
459
00:12:58,310 --> 00:13:00,379
same or different answer
460
00:13:00,379 --> 00:13:02,014
for the same questions.
461
00:13:02,014 --> 00:13:03,682
And so these features,
462
00:13:03,682 --> 00:13:04,917
these scaffolds
463
00:13:04,917 --> 00:13:06,652
become natural opportunities
464
00:13:06,652 --> 00:13:07,520
for students
465
00:13:07,520 --> 00:13:10,723
to teach them about how to evaluate
466
00:13:10,756 --> 00:13:12,491
AI instead of simply
467
00:13:12,491 --> 00:13:14,360
accepting all the answers.
468
00:13:14,360 --> 00:13:15,427
And of course, you know,
469
00:13:15,427 --> 00:13:17,630
designing these supports also comes
470
00:13:17,630 --> 00:13:19,064
with challenges.
471
00:13:19,064 --> 00:13:20,132
For example,
472
00:13:20,132 --> 00:13:23,402
one of our students said that the speech
473
00:13:23,402 --> 00:13:26,739
to text feature wasn't always accurate
474
00:13:26,839 --> 00:13:28,440
in capturing everything
475
00:13:28,440 --> 00:13:30,376
the students wanted to say.
476
00:13:30,376 --> 00:13:34,180
So we had to iterate on the interface to
477
00:13:34,213 --> 00:13:35,614
make it more reliable
478
00:13:35,614 --> 00:13:36,916
and more age appropriate
479
00:13:36,916 --> 00:13:38,984
for these students to use.
480
00:13:38,984 --> 00:13:40,986
And that's been an important lesson
481
00:13:40,986 --> 00:13:41,954
that we learned
482
00:13:41,954 --> 00:13:44,590
that effectively learning, AI learning
483
00:13:44,590 --> 00:13:47,893
design isn't just about adding features,
484
00:13:47,893 --> 00:13:49,195
it's about learning
485
00:13:49,195 --> 00:13:50,095
from students
486
00:13:50,095 --> 00:13:53,332
needs and be able to quickly adjust
487
00:13:53,465 --> 00:13:54,667
and modify
488
00:13:54,667 --> 00:13:56,869
the features to support these students
489
00:13:56,869 --> 00:13:58,737
with different learning needs.
490
00:13:58,737 --> 00:13:59,772
You spoke a little bit there
491
00:13:59,772 --> 00:14:01,841
about how their experience
492
00:14:01,841 --> 00:14:03,909
might react to the idea
493
00:14:03,909 --> 00:14:05,344
of AI getting something wrong,
494
00:14:05,344 --> 00:14:06,579
and I wanted to ask you
495
00:14:06,579 --> 00:14:08,147
about how you’re
496
00:14:08,147 --> 00:14:10,349
incorporating or responding
497
00:14:10,349 --> 00:14:11,917
to AI hallucinations.
498
00:14:11,917 --> 00:14:13,786
Like we hear a lot about these things,
499
00:14:13,786 --> 00:14:15,354
kind of just pulling information
500
00:14:15,354 --> 00:14:17,189
that maybe isn't accurate at all.
501
00:14:17,189 --> 00:14:20,025
How are you accommodating that?
502
00:14:20,025 --> 00:14:21,694
So we treat,
503
00:14:21,694 --> 00:14:24,196
you know, hallucinations in our system
504
00:14:24,196 --> 00:14:26,031
actually as a learning moment
505
00:14:26,031 --> 00:14:28,300
for them to be able to,
506
00:14:28,300 --> 00:14:30,135
you know, find out that
507
00:14:30,135 --> 00:14:32,171
AI actually make mistakes.
508
00:14:32,171 --> 00:14:33,272
And we treat that
509
00:14:33,272 --> 00:14:34,907
as a learning opportunity
510
00:14:34,907 --> 00:14:36,709
for them to realize,
511
00:14:36,709 --> 00:14:40,012
how would I detect hallucinations?
512
00:14:40,179 --> 00:14:41,547
What kind of strategies
513
00:14:41,547 --> 00:14:43,015
would I use to
514
00:14:43,015 --> 00:14:45,050
find out those hallucinations?
515
00:14:45,050 --> 00:14:47,786
And once I find out this hallucination,
516
00:14:47,786 --> 00:14:50,122
what should I do to mitigate that?
517
00:14:50,122 --> 00:14:51,957
So that's actually
518
00:14:51,957 --> 00:14:54,693
one of my favorite moments to observe.
519
00:14:54,693 --> 00:14:56,462
We see that students,
520
00:14:56,462 --> 00:14:56,795
you know,
521
00:14:56,795 --> 00:14:58,931
when they see the hallucination
522
00:14:58,931 --> 00:15:00,232
on the platform,
523
00:15:00,232 --> 00:15:03,535
they usually were surprised
524
00:15:03,535 --> 00:15:05,337
because they were expecting
525
00:15:05,337 --> 00:15:06,939
AI to know everything
526
00:15:06,939 --> 00:15:08,540
and be correct on everything.
527
00:15:08,540 --> 00:15:11,110
So when they see the AI, chatbots
528
00:15:11,110 --> 00:15:12,077
say wrong things
529
00:15:12,077 --> 00:15:14,079
like a wrong fact about the story
530
00:15:14,079 --> 00:15:17,816
that they are telling, they were confused
531
00:15:17,816 --> 00:15:20,819
because the response sounds so confident,
532
00:15:20,819 --> 00:15:23,155
but the fact was wrong.
533
00:15:23,155 --> 00:15:24,857
But very quickly
534
00:15:24,857 --> 00:15:26,091
their surprise
535
00:15:26,091 --> 00:15:28,961
actually turned to curiosity.
536
00:15:28,961 --> 00:15:30,596
So instead of asking oh,
537
00:15:30,596 --> 00:15:33,098
why is this AI so bad,
538
00:15:33,098 --> 00:15:35,467
they started thinking about
539
00:15:35,467 --> 00:15:37,303
why would the AI say that?
540
00:15:37,303 --> 00:15:40,439
And how can I make the AI better?
541
00:15:40,572 --> 00:15:42,708
And since they are building their own
542
00:15:42,708 --> 00:15:43,943
AI chatbots
543
00:15:43,943 --> 00:15:44,910
and so they're also
544
00:15:44,910 --> 00:15:46,812
experimenting AI outputs
545
00:15:46,812 --> 00:15:50,316
by refining their AI prompts, right?
546
00:15:50,349 --> 00:15:53,786
What we see often was that they would
547
00:15:53,819 --> 00:15:56,655
go back and rewrite their prompts.
548
00:15:56,655 --> 00:15:59,291
Some of them ask the same questions again
549
00:15:59,291 --> 00:16:00,492
to just to
550
00:16:00,492 --> 00:16:02,895
see if the AI would produce
551
00:16:02,895 --> 00:16:04,463
consistent answers,
552
00:16:04,463 --> 00:16:06,398
and some would add
553
00:16:06,398 --> 00:16:08,801
more specific instructions,
554
00:16:08,801 --> 00:16:11,704
and some actually went to upload
555
00:16:11,704 --> 00:16:13,572
better knowledge sources
556
00:16:13,572 --> 00:16:15,708
from those trustworthy websites.
557
00:16:15,708 --> 00:16:18,677
And they also compare their response
558
00:16:18,677 --> 00:16:21,680
from different models and say, oh,
559
00:16:21,780 --> 00:16:23,749
this model actually performs
560
00:16:23,749 --> 00:16:24,950
better than the other model.
561
00:16:24,950 --> 00:16:27,953
Maybe I would use this model more so
562
00:16:28,253 --> 00:16:31,490
we don't see these mistakes as failures.
563
00:16:31,590 --> 00:16:32,958
They start seeing them
564
00:16:32,958 --> 00:16:33,559
as something
565
00:16:33,559 --> 00:16:36,362
that they can investigate and improve.
566
00:16:36,362 --> 00:16:38,263
That's the kind of mindset
567
00:16:38,263 --> 00:16:39,031
that we are trying
568
00:16:39,031 --> 00:16:41,133
to teach the kids about.
569
00:16:41,133 --> 00:16:44,603
Not blindly trust the AI for everything
570
00:16:44,603 --> 00:16:48,240
or completely rejecting it, but knowing
571
00:16:48,774 --> 00:16:50,242
how to question it,
572
00:16:50,242 --> 00:16:51,744
how to investigate it,
573
00:16:51,744 --> 00:16:53,679
and how to improve it.
574
00:16:53,679 --> 00:16:57,583
So what's amazing was that these students
575
00:16:57,583 --> 00:16:59,318
actually started
576
00:16:59,318 --> 00:17:03,022
developing a habit of verifying AI’s
577
00:17:03,055 --> 00:17:04,790
answer whenever they see them,
578
00:17:04,790 --> 00:17:08,293
so they would actually go into
579
00:17:08,360 --> 00:17:09,395
their textbook
580
00:17:09,395 --> 00:17:10,295
or the books
581
00:17:10,295 --> 00:17:11,797
that they are talking about
582
00:17:11,797 --> 00:17:14,133
to cross-check those facts,
583
00:17:14,133 --> 00:17:15,834
and they will go on
584
00:17:15,834 --> 00:17:17,536
to those trustworthy websites
585
00:17:17,536 --> 00:17:19,271
to verify their answer.
586
00:17:19,271 --> 00:17:21,874
So when they are doing these practices,
587
00:17:21,874 --> 00:17:23,842
we see that there are also developing
588
00:17:23,842 --> 00:17:25,344
literacy skills.
589
00:17:25,344 --> 00:17:28,414
And what's fantastic about this
590
00:17:28,414 --> 00:17:30,015
is that they are actually
591
00:17:30,015 --> 00:17:31,717
intrinsically motivated
592
00:17:31,717 --> 00:17:33,719
to go read the book,
593
00:17:33,719 --> 00:17:34,820
not the teachers
594
00:17:34,820 --> 00:17:36,355
telling them read the book.
595
00:17:36,355 --> 00:17:39,258
They are motivated to go there
596
00:17:39,258 --> 00:17:40,659
and verify the answer.
597
00:17:40,659 --> 00:17:43,228
And that practice has improved
598
00:17:43,228 --> 00:17:44,897
significantly their reading
599
00:17:44,897 --> 00:17:46,065
and writing skills.
600
00:17:46,065 --> 00:17:46,865
I'd like to mention
601
00:17:46,865 --> 00:17:48,400
two things related to that,
602
00:17:48,400 --> 00:17:49,201
if that's all right.
603
00:17:49,201 --> 00:17:50,069
One is that I'm
604
00:17:50,069 --> 00:17:51,537
a mom of a third grader
605
00:17:51,537 --> 00:17:52,838
who just finished third grade,
606
00:17:52,838 --> 00:17:53,906
and is moving on to fourth.
607
00:17:53,906 --> 00:17:56,542
And, you know, with this, these ages,
608
00:17:56,542 --> 00:17:58,477
they start to have end of grade tests.
609
00:17:58,477 --> 00:18:01,080
And every parent is familiar with this,
610
00:18:01,080 --> 00:18:03,348
and the kids have to practice
611
00:18:03,348 --> 00:18:04,583
getting ready for those tests.
612
00:18:04,583 --> 00:18:05,984
And some of the things they have to do
613
00:18:05,984 --> 00:18:07,853
are the very things that Doctor Tian
614
00:18:07,853 --> 00:18:08,921
was telling you
615
00:18:08,921 --> 00:18:10,722
that the kids were inspired to do
616
00:18:10,722 --> 00:18:12,491
when they were making their own AIs.
617
00:18:12,491 --> 00:18:15,160
So the things the students need to do
618
00:18:15,160 --> 00:18:15,961
for the end of grade
619
00:18:15,961 --> 00:18:18,197
tests are tell what evidence
620
00:18:18,197 --> 00:18:19,731
in the text was,
621
00:18:19,731 --> 00:18:21,633
you know, supporting this idea.
622
00:18:21,633 --> 00:18:23,469
So the kids will be asked, like,
623
00:18:23,469 --> 00:18:23,669
you know,
624
00:18:23,669 --> 00:18:25,237
what is the main idea of this story?
625
00:18:25,237 --> 00:18:26,305
What is the main challenge
626
00:18:26,305 --> 00:18:27,873
for the main character?
627
00:18:27,873 --> 00:18:28,841
And then they have to
628
00:18:28,841 --> 00:18:29,908
actually give that evidence.
629
00:18:29,908 --> 00:18:32,144
So that extra motivation,
630
00:18:32,144 --> 00:18:32,778
when the children
631
00:18:32,778 --> 00:18:34,546
got to make their own chat bots
632
00:18:34,546 --> 00:18:36,515
and we inject hallucinations
633
00:18:36,515 --> 00:18:37,349
because models keep
634
00:18:37,349 --> 00:18:38,584
getting better and better
635
00:18:38,584 --> 00:18:40,152
so that students can learn
636
00:18:40,152 --> 00:18:41,587
that there can be hallucinations,
637
00:18:41,587 --> 00:18:43,555
we actually make sure that they happen
638
00:18:43,555 --> 00:18:44,623
so that they can have
639
00:18:44,623 --> 00:18:46,158
that experience of knowing
640
00:18:46,158 --> 00:18:47,359
that they need to check that
641
00:18:47,359 --> 00:18:48,760
there is such a thing as evidence,
642
00:18:48,760 --> 00:18:51,196
and you can go back and check that.
643
00:18:51,196 --> 00:18:54,099
And this is all really also very cool
644
00:18:54,099 --> 00:18:55,467
because we didn't
645
00:18:55,467 --> 00:18:56,702
originally plan
646
00:18:56,702 --> 00:18:58,637
to make chatbots with this district.
647
00:18:58,637 --> 00:19:00,172
This was research that Doctor
648
00:19:00,172 --> 00:19:02,508
Tian and I were already doing
649
00:19:02,508 --> 00:19:04,143
to make some chatbots.
650
00:19:04,143 --> 00:19:05,177
And then as
651
00:19:05,177 --> 00:19:06,178
we were working with the teachers,
652
00:19:06,178 --> 00:19:06,678
we said, hey,
653
00:19:06,678 --> 00:19:07,813
we'd like to show you this tool
654
00:19:07,813 --> 00:19:08,614
we're working on,
655
00:19:08,614 --> 00:19:09,915
and the teachers loved it.
656
00:19:09,915 --> 00:19:11,517
And then together
657
00:19:11,517 --> 00:19:13,652
in a research practice partnership,
658
00:19:13,652 --> 00:19:15,420
we worked together to design
659
00:19:15,420 --> 00:19:17,756
this toy talk experience for students,
660
00:19:17,756 --> 00:19:19,558
and the teachers helped implement
661
00:19:19,558 --> 00:19:20,425
it on the ground,
662
00:19:20,425 --> 00:19:22,261
like every day during a summer program.
663
00:19:22,261 --> 00:19:23,662
And this was a really inspiring
664
00:19:23,662 --> 00:19:24,596
part of this project.
665
00:19:24,596 --> 00:19:26,865
And I love how you're getting at critical
666
00:19:26,865 --> 00:19:28,066
thinking around
667
00:19:28,066 --> 00:19:29,468
computer information
668
00:19:29,468 --> 00:19:31,136
for kids at such an early age
669
00:19:31,136 --> 00:19:33,172
as they are going to go on to live
670
00:19:33,172 --> 00:19:35,474
even more online than we have.
671
00:19:35,474 --> 00:19:36,275
Just being able
672
00:19:36,275 --> 00:19:37,109
to think about what
673
00:19:37,109 --> 00:19:38,677
they're seeing or interacting with
674
00:19:38,677 --> 00:19:40,112
is really tremendous.
675
00:19:40,112 --> 00:19:42,381
So I want to pivot slightly here
676
00:19:42,381 --> 00:19:44,283
to kind of the NSF end of things.
677
00:19:44,283 --> 00:19:45,617
And you got an NSF
678
00:19:45,617 --> 00:19:47,252
grant to launch this project.
679
00:19:47,252 --> 00:19:48,120
And I understand
680
00:19:48,120 --> 00:19:50,355
you got to supplement to participate
681
00:19:50,355 --> 00:19:51,790
in the AI challenge portion.
682
00:19:51,790 --> 00:19:53,158
So can we talk a little bit
683
00:19:53,158 --> 00:19:55,794
about how that has impacted the project?
684
00:19:55,794 --> 00:19:56,361
Thank you.
685
00:19:56,361 --> 00:19:58,030
We were really excited to have a chance
686
00:19:58,030 --> 00:19:58,897
to work with the district
687
00:19:58,897 --> 00:19:59,665
on the presidential
688
00:19:59,665 --> 00:20:00,399
AI challenge,
689
00:20:00,399 --> 00:20:01,066
because we already
690
00:20:01,066 --> 00:20:02,501
had a group of six lead
691
00:20:02,501 --> 00:20:04,203
teachers who were working with us
692
00:20:04,203 --> 00:20:05,337
to learn about AI.
693
00:20:05,337 --> 00:20:06,638
So when we found out
694
00:20:06,638 --> 00:20:07,673
about the possibility
695
00:20:07,673 --> 00:20:09,741
to get a supplement for participating,
696
00:20:09,741 --> 00:20:11,076
we encouraged all the teachers
697
00:20:11,076 --> 00:20:11,610
to continue
698
00:20:11,610 --> 00:20:12,277
working with us
699
00:20:12,277 --> 00:20:13,478
to think about how they wanted
700
00:20:13,478 --> 00:20:15,280
to bring AI into their classroom.
701
00:20:15,280 --> 00:20:16,915
And then we scaffolded
702
00:20:16,915 --> 00:20:18,850
workshops where we help the teachers
703
00:20:18,850 --> 00:20:20,519
go through the presidential challenge
704
00:20:20,519 --> 00:20:21,520
materials, learn
705
00:20:21,520 --> 00:20:23,555
what it was about how to apply,
706
00:20:23,555 --> 00:20:25,224
and we supported that process
707
00:20:25,224 --> 00:20:27,492
for the teachers who wanted to do that.
708
00:20:27,492 --> 00:20:28,360
So we had
709
00:20:28,360 --> 00:20:28,760
sort of
710
00:20:28,760 --> 00:20:30,462
professional development workshops
711
00:20:30,462 --> 00:20:32,097
where we worked together with them.
712
00:20:32,097 --> 00:20:33,732
They would come up with ideas
713
00:20:33,732 --> 00:20:35,434
and tell them to each other
714
00:20:35,434 --> 00:20:37,536
and collaborate in our research team,
715
00:20:37,536 --> 00:20:39,137
you know, facilitated the teachers
716
00:20:39,137 --> 00:20:40,472
in the district to do that.
717
00:20:40,472 --> 00:20:42,574
And then as they were getting ready,
718
00:20:42,574 --> 00:20:43,775
we had practice sessions
719
00:20:43,775 --> 00:20:45,110
where the teachers
720
00:20:45,110 --> 00:20:47,112
would present their ideas to each other.
721
00:20:47,112 --> 00:20:48,680
We all gave each other feedback,
722
00:20:48,680 --> 00:20:50,949
and then they implemented their projects
723
00:20:50,949 --> 00:20:51,984
in their classrooms,
724
00:20:51,984 --> 00:20:53,819
and then brought those back to
725
00:20:53,819 --> 00:20:54,519
our teacher team
726
00:20:54,519 --> 00:20:55,721
and the research team
727
00:20:55,721 --> 00:20:57,356
to tell us about what happened.
728
00:20:57,356 --> 00:21:00,158
And then we all got to reflect on
729
00:21:00,158 --> 00:21:01,360
what the students were learning
730
00:21:01,360 --> 00:21:02,995
and thinking about
731
00:21:02,995 --> 00:21:04,396
what could be done next.
732
00:21:04,396 --> 00:21:05,664
And that was a
733
00:21:05,664 --> 00:21:07,065
really rewarding experience.
734
00:21:07,065 --> 00:21:09,034
And those extra supplements
735
00:21:09,034 --> 00:21:09,601
for stipends
736
00:21:09,601 --> 00:21:10,102
for teachers
737
00:21:10,102 --> 00:21:11,837
enabled us to help
738
00:21:11,837 --> 00:21:13,672
the teacher spend that extra time
739
00:21:13,672 --> 00:21:14,473
that they have to spend
740
00:21:14,473 --> 00:21:15,440
outside of the classroom
741
00:21:15,440 --> 00:21:18,277
to make a new activity plan,
742
00:21:18,277 --> 00:21:19,578
get the materials together
743
00:21:19,578 --> 00:21:20,879
and bring it to the classroom,
744
00:21:20,879 --> 00:21:22,147
and then collect all of that,
745
00:21:22,147 --> 00:21:22,381
you know,
746
00:21:22,381 --> 00:21:23,348
and create videos
747
00:21:23,348 --> 00:21:25,117
and content for that Presidential
748
00:21:25,117 --> 00:21:25,951
AI challenge.
749
00:21:25,951 --> 00:21:27,152
So, Professor Barnes,
750
00:21:27,152 --> 00:21:28,020
I want to ask you
751
00:21:28,020 --> 00:21:30,756
about the impact of NSF support
752
00:21:30,756 --> 00:21:31,390
over your career.
753
00:21:31,390 --> 00:21:32,491
You've got a number of grants
754
00:21:32,491 --> 00:21:32,991
over the years.
755
00:21:32,991 --> 00:21:34,293
Can you talk a little bit about what
756
00:21:34,293 --> 00:21:35,794
difference that has made for you?
757
00:21:35,794 --> 00:21:38,530
I'm extremely grateful and lucky
758
00:21:38,530 --> 00:21:40,532
to have gotten the amount of support
759
00:21:40,532 --> 00:21:41,700
over $26
760
00:21:41,700 --> 00:21:44,736
million over my career, and mostly
761
00:21:44,736 --> 00:21:45,771
it has all been
762
00:21:45,771 --> 00:21:48,440
because I have a strong belief
763
00:21:48,440 --> 00:21:50,309
that education is
764
00:21:50,309 --> 00:21:51,543
where we should be investing,
765
00:21:51,543 --> 00:21:53,345
but also in our research way,
766
00:21:53,345 --> 00:21:54,546
and also by building tools.
767
00:21:54,546 --> 00:21:55,981
So I'm a computer scientist
768
00:21:55,981 --> 00:21:57,883
and an educator in my heart,
769
00:21:57,883 --> 00:21:59,117
and so I'm always building
770
00:21:59,117 --> 00:22:00,319
educational technologies
771
00:22:00,319 --> 00:22:03,522
and AI to try to learn from data,
772
00:22:03,522 --> 00:22:05,590
to build systems that help students,
773
00:22:05,590 --> 00:22:07,225
but also to bring teachers
774
00:22:07,225 --> 00:22:09,394
into this sort of world of technology
775
00:22:09,394 --> 00:22:11,797
and how we can integrate computer science
776
00:22:11,797 --> 00:22:13,265
with education,
777
00:22:13,265 --> 00:22:15,467
but also to bring everyday people
778
00:22:15,467 --> 00:22:16,068
into learning
779
00:22:16,068 --> 00:22:17,336
about computer science and AI,
780
00:22:17,336 --> 00:22:18,704
because increasingly
781
00:22:18,704 --> 00:22:19,237
our lives
782
00:22:19,237 --> 00:22:20,572
have become more and more digital,
783
00:22:20,572 --> 00:22:21,540
and so everyone
784
00:22:21,540 --> 00:22:22,941
actually needs to have access.
785
00:22:22,941 --> 00:22:24,176
And that's
786
00:22:24,176 --> 00:22:24,743
really where
787
00:22:24,743 --> 00:22:26,211
the National Science Foundation
788
00:22:26,211 --> 00:22:30,549
has given me a platform to involve people
789
00:22:30,549 --> 00:22:33,552
from of all ages, from little kids
790
00:22:33,552 --> 00:22:34,419
up through,
791
00:22:34,419 --> 00:22:34,720
you know,
792
00:22:34,720 --> 00:22:37,723
K-12 college students, graduate students
793
00:22:37,723 --> 00:22:39,591
and faculty like myself
794
00:22:39,591 --> 00:22:41,026
and also my collaborators
795
00:22:41,026 --> 00:22:42,361
who are here and others.
796
00:22:42,361 --> 00:22:43,995
So I think that, you know,
797
00:22:43,995 --> 00:22:45,997
there's kind of a unified vision
798
00:22:45,997 --> 00:22:47,532
that many of us who apply
799
00:22:47,532 --> 00:22:48,400
for our National Science
800
00:22:48,400 --> 00:22:49,468
Foundation funding have
801
00:22:49,468 --> 00:22:50,068
is that
802
00:22:50,068 --> 00:22:51,403
we want to leverage
803
00:22:51,403 --> 00:22:54,506
not only the funds to not only advance
804
00:22:54,506 --> 00:22:56,375
research and technology,
805
00:22:56,375 --> 00:22:57,943
but also to bring
806
00:22:57,943 --> 00:22:59,711
that impact to other people.
807
00:22:59,711 --> 00:23:01,079
And I think you would see
808
00:23:01,079 --> 00:23:01,580
if you looked
809
00:23:01,580 --> 00:23:02,881
at my trajectory
810
00:23:02,881 --> 00:23:03,415
of projects
811
00:23:03,415 --> 00:23:04,015
over the years,
812
00:23:04,015 --> 00:23:04,616
that actually
813
00:23:04,616 --> 00:23:06,585
a large portion of the projects
814
00:23:06,585 --> 00:23:07,386
that I do
815
00:23:07,386 --> 00:23:08,553
have, this component
816
00:23:08,553 --> 00:23:10,622
of reaching out to teachers
817
00:23:10,622 --> 00:23:12,591
or students or the community
818
00:23:12,591 --> 00:23:14,426
and expanding access.
819
00:23:14,426 --> 00:23:16,061
And I'm super privileged
820
00:23:16,061 --> 00:23:17,295
to be able to do that.
821
00:23:17,295 --> 00:23:18,930
For my last question today,
822
00:23:18,930 --> 00:23:21,633
I want to kind of follow up with maybe
823
00:23:21,633 --> 00:23:22,834
a round robin question
824
00:23:22,834 --> 00:23:24,403
here for all of you.
825
00:23:24,403 --> 00:23:25,937
How did you feel
826
00:23:25,937 --> 00:23:28,173
or what was your reaction to the
827
00:23:28,173 --> 00:23:28,907
AI Challenge win
828
00:23:28,907 --> 00:23:30,308
coming out of this project?
829
00:23:30,308 --> 00:23:31,343
Very proud.
830
00:23:31,343 --> 00:23:35,280
And we see that what we have implemented
831
00:23:35,280 --> 00:23:36,214
in the district
832
00:23:36,214 --> 00:23:38,283
and our amazing teachers
833
00:23:38,283 --> 00:23:41,787
is when this Presidential Challenge were
834
00:23:42,120 --> 00:23:43,822
genuinely proud
835
00:23:43,822 --> 00:23:46,825
of what they did, what they accomplished,
836
00:23:46,858 --> 00:23:48,627
and the kind of impact
837
00:23:48,627 --> 00:23:50,462
that they are bringing
838
00:23:50,462 --> 00:23:51,463
to their students,
839
00:23:51,463 --> 00:23:52,431
to their families
840
00:23:52,431 --> 00:23:55,267
and to the district and to the nation.
841
00:23:55,267 --> 00:23:58,069
I think that it's very inspiring
842
00:23:58,069 --> 00:23:59,471
for everybody,
843
00:23:59,471 --> 00:24:00,739
for all the teachers
844
00:24:00,739 --> 00:24:03,742
to learn about their experiences,
845
00:24:03,742 --> 00:24:05,143
you know, implementing this
846
00:24:05,143 --> 00:24:06,945
AI lesson in our classrooms
847
00:24:06,945 --> 00:24:08,980
and other teachers,
848
00:24:08,980 --> 00:24:10,515
they can do the same thing
849
00:24:10,515 --> 00:24:12,851
and in their own classrooms.
850
00:24:12,851 --> 00:24:13,618
We are planning
851
00:24:13,618 --> 00:24:16,121
to share more about this experience,
852
00:24:16,121 --> 00:24:17,856
to empower
853
00:24:17,856 --> 00:24:20,025
other teachers to do the same thing
854
00:24:20,025 --> 00:24:21,193
in their own classroom.
855
00:24:21,193 --> 00:24:24,196
Yeah, and I echo that we are super proud
856
00:24:24,429 --> 00:24:27,132
about accomplishment of the teachers,
857
00:24:27,132 --> 00:24:27,899
not just Carrie,
858
00:24:27,899 --> 00:24:29,801
but also the other teacher teams
859
00:24:29,801 --> 00:24:31,136
who participated.
860
00:24:31,136 --> 00:24:32,337
It's a lot of effort
861
00:24:32,337 --> 00:24:33,071
to think about, to design
862
00:24:33,071 --> 00:24:34,039
the lesson,
863
00:24:34,039 --> 00:24:35,907
implement it, and also submit
864
00:24:35,907 --> 00:24:37,142
for the competition.
865
00:24:37,142 --> 00:24:38,810
So there are other teams
866
00:24:38,810 --> 00:24:40,812
also participating in this challenge
867
00:24:40,812 --> 00:24:43,582
is a big motivation for the district
868
00:24:43,582 --> 00:24:44,316
and the teachers
869
00:24:44,316 --> 00:24:45,984
that can be leaders in this space,
870
00:24:45,984 --> 00:24:47,052
and I think that we
871
00:24:47,052 --> 00:24:48,587
should give more space for teachers
872
00:24:48,587 --> 00:24:50,055
to demonstrate their own expertise
873
00:24:50,055 --> 00:24:50,856
in this area.
874
00:24:50,856 --> 00:24:51,756
And the teachers
875
00:24:51,756 --> 00:24:52,858
impact the teachers
876
00:24:52,858 --> 00:24:54,459
so they can see the impact
877
00:24:54,459 --> 00:24:55,994
on the district and the nation.
878
00:24:55,994 --> 00:24:57,229
So very, very proud.
879
00:24:57,229 --> 00:24:58,797
I just want to add on that
880
00:24:58,797 --> 00:25:02,133
I was actually in Montgomery in the PD
881
00:25:02,234 --> 00:25:03,668
in one of their summer
882
00:25:03,668 --> 00:25:06,037
teacher conference yesterday,
883
00:25:06,037 --> 00:25:09,074
and we talk about the
884
00:25:09,140 --> 00:25:10,242
things that we learned
885
00:25:10,242 --> 00:25:12,377
from the AI design challenge.
886
00:25:12,377 --> 00:25:15,480
And all the teachers in that same room
887
00:25:15,480 --> 00:25:16,414
said that
888
00:25:16,414 --> 00:25:19,217
I want to participate for the next year.
889
00:25:19,217 --> 00:25:20,719
So that's the power
890
00:25:20,719 --> 00:25:22,020
that this Presidential
891
00:25:22,020 --> 00:25:22,988
AI Challenge
892
00:25:22,988 --> 00:25:24,756
that brings so much impact
893
00:25:24,756 --> 00:25:26,258
to other teachers.
894
00:25:26,258 --> 00:25:27,826
Just to contrast two teachers
895
00:25:27,826 --> 00:25:29,461
we had in the presidential challenge.
896
00:25:29,461 --> 00:25:29,995
So Carrie
897
00:25:29,995 --> 00:25:31,396
Robledo was at the Friday
898
00:25:31,396 --> 00:25:33,431
Institute for a decade
899
00:25:33,431 --> 00:25:35,634
at least, and was working with
900
00:25:35,634 --> 00:25:37,035
actually teachers across the state,
901
00:25:37,035 --> 00:25:38,570
like helping them do digital learning.
902
00:25:38,570 --> 00:25:39,871
And she was the winner
903
00:25:39,871 --> 00:25:41,473
of the Presidential AI Challenge
904
00:25:41,473 --> 00:25:43,508
and did her second grade activity
905
00:25:43,508 --> 00:25:44,709
about using teachable
906
00:25:44,709 --> 00:25:46,111
machines with insects.
907
00:25:46,111 --> 00:25:47,112
And her students
908
00:25:47,112 --> 00:25:48,380
like understanding that
909
00:25:48,380 --> 00:25:49,681
even though you can train an AI
910
00:25:49,681 --> 00:25:51,116
to recognize different kind of insects,
911
00:25:51,116 --> 00:25:52,017
it can make a mistake.
912
00:25:52,017 --> 00:25:53,251
And then they started to learn
913
00:25:53,251 --> 00:25:55,253
that the data they put in is actually
914
00:25:55,253 --> 00:25:56,855
what the AI is learning from.
915
00:25:56,855 --> 00:25:58,323
But we had another teacher
916
00:25:58,323 --> 00:25:59,791
who this year was her first year
917
00:25:59,791 --> 00:26:01,660
doing anything like this.
918
00:26:01,660 --> 00:26:02,594
She was amazing.
919
00:26:02,594 --> 00:26:04,095
She was a first grade teacher,
920
00:26:04,095 --> 00:26:05,864
and she went above
921
00:26:05,864 --> 00:26:07,299
and beyond and was preparing
922
00:26:07,299 --> 00:26:08,133
for her Presidential
923
00:26:08,133 --> 00:26:09,935
AI Challenge, you know, ahead of time.
924
00:26:09,935 --> 00:26:11,236
And she was amazing.
925
00:26:11,236 --> 00:26:13,438
And, you know, collecting her
926
00:26:13,438 --> 00:26:14,272
students artwork
927
00:26:14,272 --> 00:26:16,074
about their imaginings, about AI
928
00:26:16,074 --> 00:26:17,309
and what they learned.
929
00:26:17,309 --> 00:26:18,977
And so we're seeing growth
930
00:26:18,977 --> 00:26:21,012
not just from teachers who are already
931
00:26:21,012 --> 00:26:23,315
well prepared to do this kind of work,
932
00:26:23,315 --> 00:26:24,716
but teachers who are just coming in
933
00:26:24,716 --> 00:26:25,717
and getting very excited
934
00:26:25,717 --> 00:26:27,519
because this is a new age
935
00:26:27,519 --> 00:26:29,154
and being able to get the chance
936
00:26:29,154 --> 00:26:31,056
to be involved with our team,
937
00:26:31,056 --> 00:26:32,057
but with other teachers
938
00:26:32,057 --> 00:26:33,224
who are excited to learn
939
00:26:33,224 --> 00:26:34,626
and then being able
940
00:26:34,626 --> 00:26:36,161
to share their progress on
941
00:26:36,161 --> 00:26:37,062
a national stage
942
00:26:37,062 --> 00:26:38,630
has made a big difference.
943
00:26:38,630 --> 00:26:40,332
Special thanks to Tiffany Barnes,
944
00:26:40,332 --> 00:26:42,334
Shiyan Jiang and Xiaoyi Tian.
945
00:26:42,334 --> 00:26:43,635
For the Discovery Files, I'm
946
00:26:43,635 --> 00:26:44,502
Nate Pottker.
947
00:26:44,502 --> 00:26:45,337
Watch video versions
948
00:26:45,337 --> 00:26:47,572
of these conversations on our @NSFscience
949
00:26:47,572 --> 00:26:48,440
YouTube channel.
950
00:26:48,440 --> 00:26:49,107
Please subscribe
951
00:26:49,107 --> 00:26:50,375
wherever you get podcasts
952
00:26:50,375 --> 00:26:50,942
and if you like
953
00:26:50,942 --> 00:26:52,477
our program, share with a friend
954
00:26:52,477 --> 00:26:54,179
and consider leaving a review.
955
00:26:56,214 --> 00:26:57,148
Discover how the U.S.
956
00:26:57,148 --> 00:26:58,316
National Science Foundation
957
00:26:58,316 --> 00:27:00,986
is advancing research at NSF.gov.
