< Return to Video

AI 101 for Teachers: Transforming Learning with AI

  • 0:00 - 0:02
    [Music]
  • 0:02 - 0:05
    foreign
  • 0:07 - 0:10
    hi welcome to session three of the AI
  • 0:10 - 0:12
    101 for teachers professional learning
  • 0:12 - 0:13
    series
  • 0:13 - 0:15
    in this session we are traveling to the
  • 0:15 - 0:17
    Wharton School at the University of
  • 0:17 - 0:20
    Pennsylvania to chat with Dr Ethan
  • 0:20 - 0:22
    Mullick a professor who teaches
  • 0:22 - 0:24
    Innovation and Entrepreneurship and Dr
  • 0:24 - 0:27
    lilac Mullick who works on interactive
  • 0:27 - 0:29
    pedagogy and AI research
  • 0:29 - 0:32
    they will help us explore how AI can be
  • 0:32 - 0:35
    combined with pedagogy to enhance
  • 0:35 - 0:36
    student learning
  • 0:36 - 0:39
    let's go meet the Molex
  • 0:44 - 0:46
    hi I'm Ethan Malek a professor at
  • 0:46 - 0:48
    Wharton who has been working on how we
  • 0:48 - 0:50
    democratize access to education through
  • 0:50 - 0:53
    tools like games and interactive tools
  • 0:53 - 0:54
    and AI
  • 0:54 - 0:57
    Nami and I'm director of pedagogy of
  • 0:57 - 0:58
    Wharton interactive and I've been
  • 0:58 - 1:00
    working at the intersection of AI and
  • 1:00 - 1:02
    education helping to democratize
  • 1:02 - 1:05
    education for everyone through effective
  • 1:05 - 1:08
    pedagogically sound use of AI and we
  • 1:08 - 1:10
    have been working together on the future
  • 1:10 - 1:12
    of education for a while thinking about
  • 1:12 - 1:14
    how to make education more interactive
  • 1:14 - 1:16
    to work at scale and with the Advent of
  • 1:16 - 1:19
    generative AI we've had a powerful new
  • 1:19 - 1:21
    tool that can really help in the
  • 1:21 - 1:22
    classroom but also carries some risks
  • 1:22 - 1:24
    today we'd like to talk a little bit
  • 1:24 - 1:27
    about the classroom use of AI upside to
  • 1:27 - 1:28
    the downside to give you some examples
  • 1:28 - 1:30
    to work with the first we like to start
  • 1:30 - 1:32
    with our three guiding principles the
  • 1:32 - 1:35
    first is that AI is undetectable there
  • 1:35 - 1:37
    are AI tools detection tools but they
  • 1:37 - 1:40
    are not effective the second principle
  • 1:40 - 1:43
    is that AI is ubiquitous it's everywhere
  • 1:43 - 1:46
    169 countries have access to Bing chat
  • 1:46 - 1:48
    and you and your students have access to
  • 1:48 - 1:51
    the most powerful AI available
  • 1:51 - 1:53
    the third principle is that AI is
  • 1:53 - 1:56
    transformative it will transform how we
  • 1:56 - 1:58
    live how we work and how we teach and
  • 1:58 - 2:00
    learn
  • 2:01 - 2:03
    foreign
  • 2:03 - 2:05
    [Music]
  • 2:05 - 2:10
    not going away but this is probably the
  • 2:10 - 2:11
    worst AI you're ever going to use so if
  • 2:11 - 2:13
    this feels disruptive now I kind of have
  • 2:13 - 2:15
    bad news which is that there isn't a
  • 2:15 - 2:17
    reason to suspect that AI development
  • 2:17 - 2:18
    will not continue and I think people
  • 2:18 - 2:20
    worry a lot about like the far future or
  • 2:20 - 2:22
    who knows how far it is where AI is
  • 2:22 - 2:23
    smarter than humans but even over the
  • 2:23 - 2:26
    next couple years even of the fact of
  • 2:26 - 2:28
    this Academic Year I would expect AI to
  • 2:28 - 2:30
    continue to improve five times ten times
  • 2:30 - 2:33
    we have no idea but if you're not
  • 2:33 - 2:34
    already thinking about these systems
  • 2:34 - 2:35
    what they mean for Education what they
  • 2:35 - 2:36
    mean for you what they mean for your
  • 2:36 - 2:38
    students careers I think we have to
  • 2:38 - 2:40
    think about it because these systems are
  • 2:40 - 2:41
    not going to disappear let me make the
  • 2:41 - 2:44
    very pragmatic case for why you might
  • 2:44 - 2:46
    want to do this for the first part of
  • 2:46 - 2:47
    the pragmatic case is your students will
  • 2:47 - 2:49
    be using this anyway so you have to come
  • 2:49 - 2:51
    to speed I don't think everybody wants
  • 2:51 - 2:52
    to be dragged along this technology no
  • 2:52 - 2:54
    one asked for education to be massively
  • 2:54 - 2:56
    disrupted but it is and unfortunately
  • 2:56 - 2:57
    we've got to figure out a way to get
  • 2:57 - 2:58
    around that all your homework
  • 2:58 - 3:00
    assignments can be done by AI now so you
  • 3:00 - 3:01
    have to think about that and then I
  • 3:01 - 3:03
    think the second thing is the pragmatic
  • 3:03 - 3:06
    argument about how AI can make your life
  • 3:06 - 3:07
    easier as a teacher if you put the hours
  • 3:07 - 3:09
    down you get them back later and if
  • 3:09 - 3:10
    you've worked on a number of prompts to
  • 3:10 - 3:12
    help make lives easier for teachers
  • 3:12 - 3:14
    yeah so one thing you can do prompts
  • 3:14 - 3:16
    like give me a lesson hook prompts like
  • 3:16 - 3:18
    create a lesson plan
  • 3:18 - 3:21
    um or Create a quiz for me so starting
  • 3:21 - 3:23
    to work with your material and the
  • 3:23 - 3:25
    different models can get you give you a
  • 3:25 - 3:27
    really good sense of how the AI Works
  • 3:27 - 3:29
    what is good at what it what it's not
  • 3:29 - 3:32
    good at and so and save you time in the
  • 3:32 - 3:34
    end right so that's where I would be my
  • 3:34 - 3:36
    push to teachers is a you have to and B
  • 3:36 - 3:38
    you're going to want to
  • 3:38 - 3:39
    I think the other important thing is
  • 3:39 - 3:42
    just to try it they're very simple to
  • 3:42 - 3:44
    use they're very intuitive because
  • 3:44 - 3:46
    they're conversational uh you can
  • 3:46 - 3:49
    continue a conversation it feels fairly
  • 3:49 - 3:51
    natural and I think the the key really
  • 3:51 - 3:54
    is experimentation see how it works with
  • 3:54 - 3:56
    you see how it works within your context
  • 3:56 - 3:59
    within your topic that you teach our
  • 3:59 - 4:01
    rule of thumb is you need about 10 hours
  • 4:01 - 4:03
    playing with AI to get what it's good
  • 4:03 - 4:05
    and what its limitations are so I would
  • 4:05 - 4:07
    actually start by suggesting that this
  • 4:07 - 4:08
    that the teacher throw their own
  • 4:08 - 4:10
    assignments into the AI and see what
  • 4:10 - 4:11
    kind of results they get back I would
  • 4:11 - 4:14
    think about asking them uh to ask their
  • 4:14 - 4:17
    students to create an assignment using
  • 4:17 - 4:18
    Ai and then critique that assignment
  • 4:18 - 4:21
    potentially even in class to see for the
  • 4:21 - 4:22
    students can get a sense of what the
  • 4:22 - 4:25
    gaps and abilities of AI are I have a
  • 4:25 - 4:26
    little bit of Freedom as an instructor
  • 4:26 - 4:29
    because I'm teaching college and MBA
  • 4:29 - 4:30
    students entrepreneurship so I have a
  • 4:30 - 4:32
    lot of I have points I want them to make
  • 4:32 - 4:33
    but they also are building things and
  • 4:33 - 4:35
    doing things and ai's absolutely
  • 4:35 - 4:38
    transformed how that works so
  • 4:38 - 4:40
    um my assignments now literally call for
  • 4:40 - 4:41
    students to do at least one impossible
  • 4:41 - 4:43
    thing in class if you can't code you
  • 4:43 - 4:45
    have to write working programs if you
  • 4:45 - 4:47
    have never if you can't do design work
  • 4:47 - 4:48
    you have to create a full graphic design
  • 4:48 - 4:50
    working prototypes that's literally that
  • 4:50 - 4:52
    part of the class so where it used to be
  • 4:52 - 4:53
    right up write a little bit of an essay
  • 4:53 - 4:55
    do a prototype on paper now you have to
  • 4:55 - 4:57
    create a full working product every
  • 4:57 - 4:59
    assignment that they turn in has to be
  • 4:59 - 5:00
    critiqued by at least five famous
  • 5:00 - 5:02
    entrepreneurs through history and they
  • 5:02 - 5:03
    use AI to invoke those there's a
  • 5:03 - 5:05
    pedagogical reason too which is that
  • 5:05 - 5:07
    entrepreneurs tend to be overconfident
  • 5:07 - 5:08
    so you want feedback from different
  • 5:08 - 5:11
    sources so to me it is let me teach 10
  • 5:11 - 5:13
    times more than I did I used to teach
  • 5:13 - 5:15
    Advanced intermediate entrepreneurship
  • 5:15 - 5:17
    course I can now in the intermediate or
  • 5:17 - 5:19
    basic course get all the way past the
  • 5:19 - 5:21
    advanced material and further so I think
  • 5:21 - 5:22
    we're going to see that shake out more
  • 5:22 - 5:23
    in the future but some of this is about
  • 5:23 - 5:25
    powering a past what we could do before
  • 5:25 - 5:27
    and I think that's exciting as well
  • 5:27 - 5:30
    apart from student tutors as AI
  • 5:30 - 5:32
    assignments teachers can certainly use
  • 5:32 - 5:35
    AI coaches and AI assistants to help
  • 5:35 - 5:37
    students prepare for discussions help
  • 5:37 - 5:40
    students outline help students do
  • 5:40 - 5:42
    research help students get feedback on
  • 5:42 - 5:44
    assignments and just help students
  • 5:44 - 5:46
    develop explanations I think there are
  • 5:46 - 5:48
    Myriad of approaches that are
  • 5:48 - 5:50
    pedagogically sound
  • 5:50 - 5:53
    that teachers can assign to students and
  • 5:53 - 5:56
    watch their work and ask for the back
  • 5:56 - 5:57
    and forth interaction to really see that
  • 5:57 - 5:59
    students are paying attention to and
  • 5:59 - 6:02
    focusing on the material
  • 6:03 - 6:04
    foreign
  • 6:04 - 6:07
    [Music]
  • 6:07 - 6:10
    teacher's perspective so because of the
  • 6:10 - 6:12
    ubiquity of AI you've got some choices
  • 6:12 - 6:15
    to make in terms of your AI policies in
  • 6:15 - 6:17
    your class so do you want to permit AI
  • 6:17 - 6:19
    do you want to forbid AI how are you
  • 6:19 - 6:20
    going to enforce these sets of things
  • 6:20 - 6:21
    we're going to assume that you want to
  • 6:21 - 6:24
    use AI to some extent and we'll dive
  • 6:24 - 6:26
    into a little bit of the details here so
  • 6:26 - 6:28
    as a instructor you should know a few
  • 6:28 - 6:30
    things one is there's obviously ongoing
  • 6:30 - 6:32
    ethical debates about Ai and those are
  • 6:32 - 6:33
    complicated debates there are debates
  • 6:33 - 6:35
    over whether or not the AI is trained on
  • 6:35 - 6:37
    the right kind of data about the biases
  • 6:37 - 6:41
    AI might have about the use of AI and
  • 6:41 - 6:43
    the outcomes for student learning and
  • 6:43 - 6:44
    it's worth acknowledging these sets of
  • 6:44 - 6:46
    things but this tool is out there and it
  • 6:46 - 6:48
    is worth thinking about how you want to
  • 6:48 - 6:50
    use it if you decide that that is okay
  • 6:50 - 6:51
    and how you want to communicate that
  • 6:51 - 6:54
    information beyond the initial ethical
  • 6:54 - 6:56
    concerns there's also concerns about how
  • 6:56 - 6:59
    AI actually works so the large language
  • 6:59 - 7:02
    models that power today's AI don't
  • 7:02 - 7:03
    actually have knowledge of the world
  • 7:03 - 7:05
    they're predicting the next word they're
  • 7:05 - 7:07
    predicting the right kinds of sentences
  • 7:07 - 7:09
    or information to give out and as a
  • 7:09 - 7:11
    result they make stuff up they would
  • 7:11 - 7:13
    call hallucinate so there are often
  • 7:13 - 7:14
    errors or mistakes now it's not always
  • 7:14 - 7:16
    clear those errors or mistakes are worse
  • 7:16 - 7:17
    than the errors and mistakes humans
  • 7:17 - 7:19
    would make but you need to be aware that
  • 7:19 - 7:20
    there's going to be those kind of errors
  • 7:20 - 7:22
    and mistakes and then finally you need
  • 7:22 - 7:25
    to think about as a instructor how
  • 7:25 - 7:27
    you're going to be using AI to Aid
  • 7:27 - 7:28
    learning which means being really clear
  • 7:28 - 7:30
    about what you want to accomplish with
  • 7:30 - 7:33
    an AI tool they can be boost for student
  • 7:33 - 7:35
    learning but ai's many possible uses in
  • 7:35 - 7:37
    classroom so do you want to use them to
  • 7:37 - 7:38
    have students generate ideas which I do
  • 7:38 - 7:40
    in my classes and we get better project
  • 7:40 - 7:42
    ideas as a result do you want them to
  • 7:42 - 7:45
    use them as tutors to explain Concepts
  • 7:45 - 7:46
    to them they don't understand do you
  • 7:46 - 7:48
    want these students to get feedback from
  • 7:48 - 7:51
    AI by asking it for questions about work
  • 7:51 - 7:53
    that they're doing do you want to be a
  • 7:53 - 7:54
    writing companion do you wanted to
  • 7:54 - 7:56
    explain why quiz answers might be right
  • 7:56 - 7:58
    or wrong and then once you've decided
  • 7:58 - 7:59
    what to do as an instructor you need to
  • 7:59 - 8:02
    decide what you're going to tell your
  • 8:02 - 8:04
    students
  • 8:06 - 8:08
    [Music]
  • 8:08 - 8:10
    AI detectors don't work they just don't
  • 8:10 - 8:12
    work you shouldn't use them and it's
  • 8:12 - 8:14
    worse than them not working because they
  • 8:14 - 8:16
    have a high false positive rate that
  • 8:16 - 8:18
    means they flag things as AI written
  • 8:18 - 8:20
    that aren't AI written and that
  • 8:20 - 8:21
    disproportionately falls on people whose
  • 8:21 - 8:24
    English is a second language this is
  • 8:24 - 8:25
    just not something that we can do and I
  • 8:25 - 8:27
    think trying to close the barn door here
  • 8:27 - 8:30
    after it's been opened and try and
  • 8:30 - 8:32
    detect AI is not the future for
  • 8:32 - 8:34
    responsibility in classrooms the the
  • 8:34 - 8:37
    other thing to note too is that students
  • 8:37 - 8:39
    were using shortcuts in the past it's
  • 8:39 - 8:41
    not that they weren't using Google it's
  • 8:41 - 8:42
    so that they weren't using you know
  • 8:42 - 8:44
    other students essays this was happening
  • 8:44 - 8:46
    in the past but this is a major
  • 8:46 - 8:48
    disruption and I think it does call for
  • 8:48 - 8:51
    a rethinking of how we do essays so
  • 8:51 - 8:53
    thinking a little bit more about the
  • 8:53 - 8:55
    learning goal for an essay or the
  • 8:55 - 8:58
    learning goal for any assignments
  • 9:00 - 9:02
    [Music]
  • 9:02 - 9:04
    one of the things that we're noticing as
  • 9:04 - 9:06
    we watch teachers do this is they all
  • 9:06 - 9:09
    feel an obligation to talk about Ai and
  • 9:09 - 9:11
    dive deep into the ethical implications
  • 9:11 - 9:13
    of AI and so on I think that's important
  • 9:13 - 9:15
    but I don't think it needs to be the
  • 9:15 - 9:17
    theme of every class I don't think every
  • 9:17 - 9:19
    class needs to be a discussion about AI
  • 9:19 - 9:21
    just like every class that uses
  • 9:21 - 9:22
    computers doesn't need to be a
  • 9:22 - 9:23
    discussion about computers I think it's
  • 9:23 - 9:25
    important to have that conversation and
  • 9:25 - 9:27
    right now we're all just reacting so
  • 9:27 - 9:28
    it's not clear who's supposed to have
  • 9:28 - 9:30
    that so I totally get teachers wanting
  • 9:30 - 9:32
    to have ai discussions but it's even
  • 9:32 - 9:35
    harder to come up to speed not just on
  • 9:35 - 9:37
    the use of AI but how it works it's you
  • 9:37 - 9:39
    know standards it's ethical implications
  • 9:39 - 9:42
    so I think teachers should feel a little
  • 9:42 - 9:44
    bit of um okayness with experimenting
  • 9:44 - 9:46
    with AI without having to make it the
  • 9:46 - 9:49
    subject of class
  • 9:49 - 9:51
    foreign
  • 9:51 - 9:53
    [Music]
  • 9:53 - 9:56
    first is as Ethan mentioned that AI can
  • 9:56 - 9:58
    fabricate that means that any output
  • 9:58 - 10:00
    that the AI gives a student may be made
  • 10:00 - 10:03
    up it may be mistaken it may be very
  • 10:03 - 10:05
    subtly mistaken and so students should
  • 10:05 - 10:08
    be responsible for their own work they
  • 10:08 - 10:10
    should at the very least check sources
  • 10:10 - 10:12
    check any number check any facts that
  • 10:12 - 10:14
    the AI gives them and check them with
  • 10:14 - 10:17
    credible sources the second principle is
  • 10:17 - 10:19
    that the AI is not a person it's easy to
  • 10:19 - 10:22
    imbue the AI with a personality or to
  • 10:22 - 10:24
    feel like you're talking to a person but
  • 10:24 - 10:25
    it's not a person and it doesn't know
  • 10:25 - 10:28
    you the third principle is really to
  • 10:28 - 10:31
    give it a lot of context the AI doesn't
  • 10:31 - 10:32
    know you it doesn't know your context or
  • 10:32 - 10:35
    your experience or your expertise the
  • 10:35 - 10:37
    more context you give it the more useful
  • 10:37 - 10:39
    it'll be for you and the fourth
  • 10:39 - 10:41
    principle is that you're in charge not
  • 10:41 - 10:43
    only should you evaluate and interrogate
  • 10:43 - 10:46
    its output but if it's leading you in a
  • 10:46 - 10:48
    conversation that is no longer useful to
  • 10:48 - 10:50
    to you or if it's stuck in a loop or if
  • 10:50 - 10:52
    you'd like to change the direction of
  • 10:52 - 10:54
    the conversation you should ask feel
  • 10:54 - 10:57
    free to take charge
  • 10:57 - 10:58
    thank you
  • 10:58 - 11:02
    [Music]
  • 11:02 - 11:06
    so when we talk about Ai and these
  • 11:06 - 11:08
    generative AI Solutions we tend to talk
  • 11:08 - 11:10
    about large language models and there's
  • 11:10 - 11:13
    actually only a few large-scale general
  • 11:13 - 11:15
    purpose large language models there is
  • 11:15 - 11:17
    the models created by openai which are
  • 11:17 - 11:21
    GPT 3.5 or gbt4 gbd 3.5 is the free
  • 11:21 - 11:24
    version that you get through uh through
  • 11:24 - 11:27
    chat CBT and gbt4 is either through the
  • 11:27 - 11:29
    page chat gbt or through Microsoft Bing
  • 11:29 - 11:32
    in creative mode and when we talk about
  • 11:32 - 11:34
    specialized apps almost all of them are
  • 11:34 - 11:36
    using one of these models and providing
  • 11:36 - 11:38
    prompts and other information on top of
  • 11:38 - 11:41
    it I generally think instructors should
  • 11:41 - 11:43
    get familiar with the models themselves
  • 11:43 - 11:45
    because those are the models that are
  • 11:45 - 11:47
    actually producing the answers and you
  • 11:47 - 11:49
    can manipulate them directly that way
  • 11:49 - 11:50
    and learn how they work so if you're
  • 11:50 - 11:51
    trying to buy an off-the-shelf solution
  • 11:51 - 11:53
    they're almost only using one of these
  • 11:53 - 11:55
    existing models and then providing some
  • 11:55 - 11:57
    sort of wrapper or other information on
  • 11:57 - 11:59
    top of it and it's often cheaper and
  • 11:59 - 12:00
    more effective and gives you more
  • 12:00 - 12:02
    control to use the foundation models
  • 12:02 - 12:04
    yourself but that's a choice you get to
  • 12:04 - 12:07
    make so when develop helping the prompt
  • 12:07 - 12:10
    we really and for all of our prompts we
  • 12:10 - 12:12
    really look at the science of learning
  • 12:12 - 12:14
    and try to combine that with the power
  • 12:14 - 12:17
    of the AI so for instance a good tutor
  • 12:17 - 12:20
    pushes you for information uh it doesn't
  • 12:20 - 12:22
    just hand it to you a good tutor finds
  • 12:22 - 12:24
    out what you know and builds on that
  • 12:24 - 12:26
    prior knowledge
  • 12:26 - 12:29
    a good tutor will also find out a little
  • 12:29 - 12:31
    bit about you a good tutor also knows
  • 12:31 - 12:34
    that you need lots and varied kinds of
  • 12:34 - 12:37
    examples and analogies and a good tutor
  • 12:37 - 12:40
    knows that the way that you show
  • 12:40 - 12:42
    evidence of Mastery is by being able to
  • 12:42 - 12:44
    explain something in your own words to
  • 12:44 - 12:46
    someone else and give an example of it
  • 12:46 - 12:49
    which is exactly these are exactly the
  • 12:49 - 12:51
    steps and the kinds of questions that we
  • 12:51 - 12:53
    use in the tutor prompt but you'll
  • 12:53 - 12:55
    notice when you look at our prompts that
  • 12:55 - 12:57
    they do things like provide context to
  • 12:57 - 12:59
    the AI as Leo has discussed already the
  • 12:59 - 13:01
    idea that it it asks you who you are and
  • 13:01 - 13:03
    we tell the AI who it is it's an
  • 13:03 - 13:05
    instructor with this kind of setting
  • 13:05 - 13:07
    you'll notice that it also tells it
  • 13:07 - 13:09
    exactly the scientific framework to use
  • 13:09 - 13:12
    this idea of context matters we provide
  • 13:12 - 13:14
    controls we ask it to go step by step
  • 13:14 - 13:16
    through sets of questions to ask
  • 13:16 - 13:18
    sometimes not in these prompts we
  • 13:18 - 13:19
    provide examples of good output and then
  • 13:19 - 13:22
    we test it a lot you can't do prompting
  • 13:22 - 13:24
    without testing and that's one of the
  • 13:24 - 13:25
    great things about testing your
  • 13:25 - 13:28
    expertise it's cheap to do and so you
  • 13:28 - 13:29
    get to experiment a lot and that makes
  • 13:29 - 13:32
    for good prompts and and we should also
  • 13:32 - 13:34
    say we tested not just on one model but
  • 13:34 - 13:36
    on several models so for instance these
  • 13:36 - 13:38
    two prompts we just worked with chat
  • 13:38 - 13:42
    gpt4 they also work with Bing Bing will
  • 13:42 - 13:44
    react a little bit differently and it
  • 13:44 - 13:46
    will um because it's connected to the
  • 13:46 - 13:48
    internet it will also look up citations
  • 13:48 - 13:49
    sometimes they're the right citations
  • 13:49 - 13:51
    sometimes they're not but that is
  • 13:51 - 13:54
    available it may or may not work with
  • 13:54 - 13:55
    some of the other models so you really
  • 13:55 - 13:57
    have to test it I think as an instructor
  • 13:57 - 13:59
    before you give it to your students in
  • 13:59 - 14:01
    the context of the topic that you're
  • 14:01 - 14:03
    teaching to see how it works so this is
  • 14:03 - 14:05
    all very theoretical but I think
  • 14:05 - 14:07
    important so let's let's get practical
  • 14:07 - 14:08
    let's talk about some examples of what
  • 14:08 - 14:11
    AI can do and again this isn't a
  • 14:11 - 14:13
    monolithical thing a AI has many
  • 14:13 - 14:14
    possible uses as we said it was
  • 14:14 - 14:16
    transformative earlier so we're going to
  • 14:16 - 14:17
    show you a couple of prompts that we've
  • 14:17 - 14:19
    created and those will be available to
  • 14:19 - 14:22
    you as well to work with and these are
  • 14:22 - 14:25
    just examples of the ways AI classroom
  • 14:25 - 14:27
    use can work so the first one we want to
  • 14:27 - 14:29
    show you is a prompt that I believe you
  • 14:29 - 14:32
    created that talks about uh feedback
  • 14:32 - 14:34
    that gives proper feedback and
  • 14:34 - 14:36
    one of the really interesting things
  • 14:36 - 14:38
    about the AI side of things is a more
  • 14:38 - 14:39
    sophisticated prompt that takes into
  • 14:39 - 14:40
    account some of the principles we were
  • 14:40 - 14:41
    talking about earlier will result in
  • 14:41 - 14:43
    better outcomes so students will often
  • 14:43 - 14:45
    ask for writing advice from an AI even
  • 14:45 - 14:46
    if you tell them not to do it but
  • 14:46 - 14:47
    they're going to ask for it in a way
  • 14:47 - 14:49
    that's fairly unsophisticated and is
  • 14:49 - 14:50
    going to give them fairly generic
  • 14:50 - 14:52
    sounding work and possibly more mistakes
  • 14:52 - 14:54
    if you give a more elaborate prompt you
  • 14:54 - 14:55
    can get more elaborate answers so in
  • 14:55 - 14:57
    this case could you explain what this
  • 14:57 - 14:59
    prompt does the feedback prompt yes so
  • 14:59 - 15:03
    we combine the principles of good
  • 15:03 - 15:05
    feedback which is feedback that takes
  • 15:05 - 15:07
    into account your prior knowledge or
  • 15:07 - 15:09
    what you already know from the student
  • 15:09 - 15:11
    perspective takes into account who you
  • 15:11 - 15:13
    are your learning level what grade
  • 15:13 - 15:15
    you're in uh whether you're in college
  • 15:15 - 15:17
    or you're a professional and it also
  • 15:17 - 15:20
    takes into account uh the idea that you
  • 15:20 - 15:22
    want to respond to this feedback so it
  • 15:22 - 15:24
    is going to be actionable it's going to
  • 15:24 - 15:26
    be balanced it's going to tell you
  • 15:26 - 15:28
    what's wrong and what you can improve on
  • 15:28 - 15:30
    and what you're doing well and it's
  • 15:30 - 15:32
    going to keep working with you but like
  • 15:32 - 15:34
    any good tutor or coach it won't
  • 15:34 - 15:36
    actually give you the answer it'll push
  • 15:36 - 15:38
    you in that direction ask you to explain
  • 15:38 - 15:41
    ask you to construct your own knowledge
  • 15:41 - 15:43
    and so you can see the prompt hopefully
  • 15:43 - 15:46
    on the screen here uh and uh as a place
  • 15:46 - 15:48
    to work from you don't need to take this
  • 15:48 - 15:50
    as an absolute answer this is something
  • 15:50 - 15:52
    you can play with but let's see it in
  • 15:52 - 15:54
    action so let's get started using this
  • 15:54 - 15:56
    prompt it says that it's a teaching
  • 15:56 - 15:57
    assistant because that's the
  • 15:57 - 15:58
    instructions we gave it and it asks us
  • 15:58 - 16:01
    for our grade level and subject we're
  • 16:01 - 16:03
    studying what should we say so I think
  • 16:03 - 16:05
    we're studying Macbeth and we're in 12th
  • 16:05 - 16:09
    grade in 12th grade okay great
  • 16:11 - 16:14
    okay and so we've told the AI as
  • 16:14 - 16:17
    information it's feeding it into the
  • 16:17 - 16:19
    logic that it's using here and it's
  • 16:19 - 16:21
    asking us about a specific assignment
  • 16:21 - 16:22
    um and it's asking if we have a rubric
  • 16:22 - 16:24
    or other information to work with or
  • 16:24 - 16:26
    what we're hoping to achieve with uh
  • 16:26 - 16:27
    with as much of Rage as possible I don't
  • 16:27 - 16:30
    have a huge amount here so I'll say I
  • 16:30 - 16:35
    have to write an analysis of Macbeth
  • 16:35 - 16:39
    it is graded
  • 16:39 - 16:41
    based on
  • 16:41 - 16:46
    writing style and depth of content
  • 16:46 - 16:48
    and you'll see what it's going is it's
  • 16:48 - 16:50
    asking us questions soliciting
  • 16:50 - 16:51
    information from us which makes it a
  • 16:51 - 16:52
    kind of a good prompt that you might
  • 16:52 - 16:55
    hand a student better than one that is
  • 16:55 - 16:56
    just they're just developing themselves
  • 16:56 - 16:58
    and it's asking about Specific
  • 16:58 - 16:59
    Instructions and as gets to share the
  • 16:59 - 17:02
    assignment uh here is what I have
  • 17:02 - 17:04
    written so far
  • 17:05 - 17:08
    and I have asked the AI to generate a
  • 17:08 - 17:11
    Macbeth essay so here we go I'm just
  • 17:11 - 17:13
    pasting that in
  • 17:13 - 17:15
    and we'll see what it says here and
  • 17:15 - 17:17
    you'll notice it's it's it's working on
  • 17:17 - 17:18
    the information it's saying it's taking
  • 17:18 - 17:20
    time to carefully read through it that's
  • 17:20 - 17:21
    a bit of an illusion it's obviously not
  • 17:21 - 17:23
    taking any extra time but uh it's
  • 17:23 - 17:25
    responding in this method and you'll see
  • 17:25 - 17:26
    it's giving a set of strengths and
  • 17:26 - 17:27
    weaknesses what's great about again
  • 17:27 - 17:30
    using a tutor that you've built or a
  • 17:30 - 17:31
    mentor that you built is that it can
  • 17:31 - 17:32
    give you the kind of feedback that's
  • 17:32 - 17:34
    educationally valuable that ties into
  • 17:34 - 17:36
    pedagogy rather than just students
  • 17:36 - 17:38
    asking make this essay better a nice
  • 17:38 - 17:42
    example of AI working in your favor uh
  • 17:42 - 17:43
    as an educator and not necessarily
  • 17:43 - 17:45
    working against you and undermining the
  • 17:45 - 17:46
    points you're making
  • 17:46 - 17:48
    so you'll notice at the End by the way
  • 17:48 - 17:51
    it gives a question that you for the
  • 17:51 - 17:54
    student's answer so how do you plan to
  • 17:54 - 17:56
    revise your analysis give me a plan on
  • 17:56 - 17:57
    specific changes you're going to make
  • 17:57 - 17:58
    again the kind of thing we would do as
  • 17:58 - 17:59
    an instructor in our classroom
  • 17:59 - 18:02
    soliciting changes or differences so I
  • 18:02 - 18:06
    think you can start to see why a uh tool
  • 18:06 - 18:09
    like this can be really useful when
  • 18:09 - 18:11
    properly applied now let's also talk
  • 18:11 - 18:13
    about one other potential use for ai ai
  • 18:13 - 18:14
    as a tutor what are some of the
  • 18:14 - 18:16
    advantages of disadvantage of that
  • 18:16 - 18:17
    approach so an advantage of this
  • 18:17 - 18:19
    approach is that you're getting students
  • 18:19 - 18:21
    to actually pay attention to the
  • 18:21 - 18:23
    material you're getting them to read
  • 18:23 - 18:26
    over the rubric to read over the purpose
  • 18:26 - 18:27
    of the essay and the audience and to
  • 18:27 - 18:30
    really think through it a disadvantage
  • 18:30 - 18:34
    is that you you certainly can ask the AI
  • 18:34 - 18:37
    to do it for you but if you work with it
  • 18:37 - 18:38
    and if you're given guidelines to work
  • 18:38 - 18:40
    with it it's one way to get feedback
  • 18:40 - 18:42
    that you would then have to evaluate
  • 18:42 - 18:44
    something else that a teacher could do
  • 18:44 - 18:47
    is to ask for the interaction and ask
  • 18:47 - 18:49
    for a reflection about the interactions
  • 18:49 - 18:51
    what about this feedback was good what
  • 18:51 - 18:53
    about this feedback was not as good and
  • 18:53 - 18:55
    again it's a higher order level thinking
  • 18:55 - 18:59
    about your essay and and your process
  • 18:59 - 19:01
    excellent and so I want to show you one
  • 19:01 - 19:03
    other example where the AI acts as a
  • 19:03 - 19:05
    direct instructor and we have a prompt
  • 19:05 - 19:07
    for that as well there are risks
  • 19:07 - 19:09
    associated with asking the AI to be a
  • 19:09 - 19:10
    direct instructor which is that
  • 19:10 - 19:12
    hallucination risk it doesn't know your
  • 19:12 - 19:15
    pedagogy your your your perspective but
  • 19:15 - 19:17
    I find in my classrooms that students
  • 19:17 - 19:20
    are increasingly using the AI as a
  • 19:20 - 19:22
    method of learning so they don't raise
  • 19:22 - 19:23
    their hands as much I'm going to ask
  • 19:23 - 19:24
    them why they're like well I'd rather
  • 19:24 - 19:26
    not show my evenings in class I could
  • 19:26 - 19:28
    ask the AI to explain like I'm five so
  • 19:28 - 19:30
    they're already engaging this Behavior
  • 19:30 - 19:32
    so something like a tutor both does a
  • 19:32 - 19:34
    useful thing of showing you what the
  • 19:34 - 19:36
    future of AI education might look like
  • 19:36 - 19:38
    uh like the way Khan Academy is building
  • 19:38 - 19:40
    AI interactive tutors to work and it
  • 19:40 - 19:42
    also might be a tool your students can
  • 19:42 - 19:44
    use to achieve more in class but you
  • 19:44 - 19:46
    should caveat that with the knowledge
  • 19:46 - 19:49
    that AI tutors are not 100 there yet but
  • 19:49 - 19:50
    let's let's use an example here so this
  • 19:50 - 19:52
    tutor is again trying to take the right
  • 19:52 - 19:53
    kind of format it says hello there I'm
  • 19:53 - 19:55
    your AI tutor and I'm excited to work
  • 19:55 - 19:57
    with you today what do we want to learn
  • 19:57 - 19:59
    about today
  • 19:59 - 20:01
    opportunity cost the concept from
  • 20:01 - 20:04
    economics let's see what happens
  • 20:04 - 20:06
    okay so we're telling the AI where the
  • 20:06 - 20:08
    opportunity cost it's saying it is a key
  • 20:08 - 20:10
    concept of Economics that's even throw a
  • 20:10 - 20:13
    little Emoji here which is cute uh can
  • 20:13 - 20:14
    you ask us about our learning level what
  • 20:14 - 20:17
    level are we at here 11th grade 11th
  • 20:17 - 20:18
    grade
  • 20:18 - 20:21
    now I wouldn't get too tied up on the
  • 20:21 - 20:23
    individual grade it's not amazing at
  • 20:23 - 20:24
    differentiating a 10th grader from 11th
  • 20:24 - 20:26
    grader but this is part of the context
  • 20:26 - 20:27
    in which it's working and so it's not
  • 20:27 - 20:29
    pulling from some sort of universal
  • 20:29 - 20:32
    standards here and it says what do we
  • 20:32 - 20:34
    know about opportunity cost
  • 20:34 - 20:37
    well we know that it has to do with
  • 20:37 - 20:40
    choices has to do with alternative
  • 20:40 - 20:42
    choices
  • 20:42 - 20:44
    that is it and of course the advance one
  • 20:44 - 20:46
    of the advantages of AI is this kind of
  • 20:46 - 20:48
    freeform texted interaction is the real
  • 20:48 - 20:50
    power of education and it's something
  • 20:50 - 20:52
    the AI can fake reasonably well again
  • 20:52 - 20:54
    not as well as a real human instructor
  • 20:54 - 20:57
    yet and you'll notice it's giving us
  • 20:57 - 20:59
    examples and explaining things in
  • 20:59 - 21:02
    different ways which is a powerful thing
  • 21:02 - 21:03
    that AI can do it's very good at
  • 21:03 - 21:05
    breaking things down in different ways
  • 21:05 - 21:06
    but you'll notice that it's now starting
  • 21:06 - 21:09
    to ask questions it's asking us to make
  • 21:09 - 21:10
    choices so one of the things we know
  • 21:10 - 21:12
    from the research on tutoring is you
  • 21:12 - 21:14
    can't just declaim things to people the
  • 21:14 - 21:15
    advantage of tutoring comes from
  • 21:15 - 21:17
    soliciting information making
  • 21:17 - 21:18
    connections and you can see the AI
  • 21:18 - 21:20
    starting to do this and asking us for
  • 21:20 - 21:22
    Connections in our own life the other
  • 21:22 - 21:24
    thing to mention too about the Tudor
  • 21:24 - 21:26
    prompt is that it is not assuming that
  • 21:26 - 21:28
    the student can judge their own learning
  • 21:28 - 21:31
    uh very often you'll see in a tutor
  • 21:31 - 21:33
    prompt that is very simple like explain
  • 21:33 - 21:35
    to me like I'm 10 It'll ask you if you
  • 21:35 - 21:38
    understand something instead here it's
  • 21:38 - 21:39
    not asking you to make a judgment about
  • 21:39 - 21:41
    your own learning which we know is
  • 21:41 - 21:43
    inherently flawed instead it's
  • 21:43 - 21:45
    soliciting as Ethan said soliciting
  • 21:45 - 21:47
    information from you to find out what
  • 21:47 - 21:49
    you know and to help you build on your
  • 21:49 - 21:51
    knowledge and these sorts of subtle
  • 21:51 - 21:54
    differences are what separates
  • 21:54 - 21:56
    using AI in sort of an expert way in a
  • 21:56 - 21:58
    classroom where we know what we want to
  • 21:58 - 22:01
    have happen from just the naive use
  • 22:01 - 22:03
    people are doing
  • 22:05 - 22:06
    [Music]
  • 22:06 - 22:08
    I I think that there is an advantage to
  • 22:08 - 22:11
    taking charge of your students AI usage
  • 22:11 - 22:12
    because they're going to be using it
  • 22:12 - 22:13
    anyway and thinking about directing it
  • 22:13 - 22:15
    either by giving them prompts having
  • 22:15 - 22:16
    these discussions and it's a really
  • 22:16 - 22:18
    powerful tool that in the future will
  • 22:18 - 22:21
    greatly boost classroom learning and is
  • 22:21 - 22:23
    not a replacement or threat for students
  • 22:23 - 22:25
    for teachers it is something that we can
  • 22:25 - 22:27
    use to improve the outputs of our work
  • 22:27 - 22:29
    improve student learning make our lives
  • 22:29 - 22:31
    easier while making students lives
  • 22:31 - 22:33
    better and I think that that's a very
  • 22:33 - 22:35
    powerful view of the future and I hope
  • 22:35 - 22:36
    that you at least Embrace that
  • 22:36 - 22:38
    experiment with AI before deciding
  • 22:38 - 22:40
    whether you want to use it or ban it in
  • 22:40 - 22:41
    your classrooms
  • 22:41 - 22:44
    wow there really are so many ways to
  • 22:44 - 22:46
    enhance student learning using AI
  • 22:46 - 22:49
    with AI technology advancing rapidly
  • 22:49 - 22:50
    there will be more and more tools
  • 22:50 - 22:54
    available as with any new tool Educators
  • 22:54 - 22:56
    have a responsibility to ensure they are
  • 22:56 - 22:58
    using age-appropriate tools
  • 22:58 - 23:01
    protecting student privacy and creating
  • 23:01 - 23:03
    spaces for students to critically
  • 23:03 - 23:05
    evaluate the potential of pitfalls of
  • 23:05 - 23:08
    the technology they are using join us in
  • 23:08 - 23:10
    session four ensuring a responsible
  • 23:10 - 23:13
    approach to AI as we explore these
  • 23:13 - 23:15
    topics thanks for joining us see you
  • 23:15 - 23:18
    again in session four visit the AI 101
  • 23:18 - 23:22
    for teachers website at co.org AI 101 to
  • 23:22 - 23:25
    sign up for Early Access and to explore
  • 23:25 - 23:28
    additional resources from code.org ETS
  • 23:28 - 23:31
    IST and Khan Academy thanks for joining
  • 23:31 - 23:33
    us
Title:
AI 101 for Teachers: Transforming Learning with AI
Description:

more » « less
Video Language:
English
Team:
Code.org
Project:
AI 101 for teachers
Duration:
23:40

English (auto-generated) subtitles

Revisions

  • Revision 1 ASR: YouTube automatic subtitles
    Amara Bot