Latest / Azadeh and Guy on AI / Same Work, Different Feedback: Melissa Warr on Hidden Bias in AI

Transcript

Same Work, Different Feedback: Melissa Warr on Hidden Bias in AI · Azadeh and Guy on AI

• 0:01
  SPEAKER_01: Hi and welcome to Azadeh and Guy on AI.
• 0:05
  SPEAKER_01: And today we have with us Dr.
• 0:09
  SPEAKER_01: Melissa Ward and she is an Assistant Professor of Educational Design
• 0:14
  SPEAKER_01: and Learning Technologies at New Mexico State,
• 0:18
  SPEAKER_01: where apparently they're getting the day off for at least K-12 schools are getting
• 0:23
  SPEAKER_01: the day off for voting.
• 0:25
  SPEAKER_01: And so this is really exciting.
• 0:27
  SPEAKER_01: We didn't get the day off.
• 0:28
  SPEAKER_01: I had to wake up really early in the morning and run and vote quickly.
• 0:33
  SPEAKER_01: And it's raining here.
• 0:35
  SPEAKER_01: I'm pretty sure it's not raining in New Mexico right now.
• 0:37
  SPEAKER_02: It's very sunny.
• 0:39
  SPEAKER_02: I didn't get the day off, but the children did, which...
• 0:43
  SPEAKER_01: Which makes you kind of have the day off and on at the same time.
• 0:47
  SPEAKER_02: Yes, yes.
• 0:48
  SPEAKER_01: As parenting tends to do.
• 0:50
  SPEAKER_01: And her research explores the intersections of AI, creativity, design,
• 0:56
  SPEAKER_01: and critical pedagogy with a particular focus on equity in education.
• 1:01
  SPEAKER_01: And this is one of the major reasons we wanted to have you on.
• 1:04
  SPEAKER_01: We've talked a lot about other aspects.
• 1:07
  SPEAKER_01: And we always run around the ethics.
• 1:10
  SPEAKER_01: And this is like Straight at the ethics and the way you explore it.
• 1:16
  SPEAKER_01: And you also examine how AI technologies can reinforce existing biases
• 1:21
  SPEAKER_01: and inequities, as well as how they can be used critically and creatively
• 1:26
  SPEAKER_01: to support more inclusive learning environments.
• 1:29
  SPEAKER_01: I got it all out.
• 1:30
  Nice.
• 1:32
  SPEAKER_01: So we're really excited to have you with us today.
• 1:37
  SPEAKER_01: Anything we didn't say that you're dying that we will say?
• 1:41
  SPEAKER_02: I don't think so. You covered a lot there.
• 1:43
  SPEAKER_01: Okay. Okay.
• 1:44
  SPEAKER_01: We try, but, you know, there's always something. All right.
• 1:49
  SPEAKER_01: Azadeh, do you want to start?
• 1:50
  SPEAKER_00: Yeah, sure.
• 1:51
  SPEAKER_00: Welcome again, Dr. War.
• 1:53
  SPEAKER_00: So I'm going to start with the first question.
• 1:56
  SPEAKER_00: In your research, you've demonstrated
• 1:58
  SPEAKER_00: that AI can evaluate identical student work differently depending
• 2:03
  SPEAKER_00: on contextual information about the student.
• 2:06
  SPEAKER_00: What should educators understand about the risk of using AI for assessment and feedback?
• 2:12
  SPEAKER_02: Yeah, that's a great question.
• 2:14
  SPEAKER_02: And one I think about a lot too, because these models are constantly changing.
• 2:20
  SPEAKER_02: And so the interesting thing is, you know, a slight change to what I'm seeing these days.
• 2:25
  SPEAKER_02: Right now, we're writing up a study where we compare the models from 2024,
• 2:31
  SPEAKER_02: then do the same thing in 2026.
• 2:34
  SPEAKER_02: And we're actually seeing the scoring isn't as biased.
• 2:39
  SPEAKER_02: What is biased is the textual responses.
• 2:42
  SPEAKER_02: So we look at measures like how if it gives feedback to students,
• 2:46
  SPEAKER_02: what's the reading level of that feedback?
• 2:48
  SPEAKER_02: So is it riding on like a 10th grade reading level or like a 7th grade reading level?
• 2:53
  SPEAKER_02: And if we say, for example, student likes rap music, it makes it much easier to read.
• 2:59
  SPEAKER_02: Humans adjust for the audience that they perceive.
• 3:03
  SPEAKER_02: We also see we use some metrics that show that it talks almost down more.
• 3:09
  SPEAKER_02: If, for example, we use rap, we did rap and classical for this one,
• 3:12
  SPEAKER_02: just saying the student likes rap music, the student likes classical music,
• 3:16
  SPEAKER_02: give the same passages.
• 3:18
  SPEAKER_02: And so it almost talks down to them more.
• 3:21
  SPEAKER_02: And so while scores seem to be evening out, which is a good thing,
• 3:26
  SPEAKER_02: The implications of the language it's using with students is really concerning
• 3:31
  SPEAKER_02: because that goes beyond, it's harder to see without different metrics to see it.
• 3:36
  SPEAKER_02: That's part of it.
• 3:38
  SPEAKER_02: And it reinforces kind of this, what they call the hidden curriculum of schooling,
• 3:42
  SPEAKER_02: where some students are given more direct instruction, they're seen as less capable,
• 3:49
  SPEAKER_02: and so they have to be told what to do.
• 3:52
  SPEAKER_02: And we've seen that, too, with how they support creativity.
• 3:55
  SPEAKER_02: They're going to give more direct recommendations to some things, some students,
• 3:59
  SPEAKER_02: and more open-ended to others.
• 4:01
  SPEAKER_02: And we know creativity, we want open-ended.
• 4:04
  SPEAKER_02: So I think going back to your question for teachers,
• 4:08
  SPEAKER_02: I think it's really important to think carefully about why you're doing it
• 4:14
  SPEAKER_02: and what it's doing.
• 4:17
  SPEAKER_02: I'm sympathetic with teachers. I tried being a teacher.
• 4:19
  SPEAKER_02: It was too stressful in K-12.
• 4:23
  SPEAKER_02: I do not want to tell teachers not to do things with AI because I did not survive.
• 4:30
  SPEAKER_02: But think carefully about what it is you're portraying and giving
• 4:35
  SPEAKER_02: when you're using this AI feedback and how students are perceiving it.
• 4:40
  SPEAKER_02: One of the biggest things with AI is it sounds confident
• 4:43
  SPEAKER_02: and it sounds like it knows what it's talking about.
• 4:46
  SPEAKER_02: It's good at getting those signals, right?
• 4:48
  SPEAKER_00: Yes.
• 4:49
  SPEAKER_02: Yeah.
• 4:49
  SPEAKER_02: And so a student can just automatically take it as this truth and this reflection
• 4:55
  SPEAKER_02: on them versus if they're thinking about it critically and they're thinking, OK,
• 5:00
  SPEAKER_02: you know, we're being transparent.
• 5:01
  SPEAKER_02: My teacher said that this was AI feedback.
• 5:04
  SPEAKER_02: And I need to try to evaluate it.
• 5:06
  SPEAKER_02: And if it's critical or if it's appropriate or what does this say,
• 5:09
  SPEAKER_02: that's very different than a teacher doing it without telling students,
• 5:13
  SPEAKER_02: which has a lot of implications too.
• 5:18
  SPEAKER_02: Yeah. So I hope that answers your question.
• 5:20
  SPEAKER_01: Yeah.
• 5:20
  SPEAKER_01: And I love the fact that you talk about disclosure, which is something we explored.
• 5:26
  SPEAKER_01: And one of the things that I'm really feeling strongly
• 5:30
  SPEAKER_01: about is Just like you're asking your students for disclosure.
• 5:34
  SPEAKER_01: You got to disclose if you're using AI.
• 5:36
  SPEAKER_01: And if you're not, first of all, they often sense that there's an AI.
• 5:41
  SPEAKER_01: And that's something we're hearing from students very often.
• 5:44
  SPEAKER_01: And I'm not even sure the teacher is always using AI.
• 5:48
  SPEAKER_01: But I think that just like I'm telling my students, I'm saying to them,
• 5:52
  SPEAKER_01: you got to have an AI disclosure.
• 5:54
  SPEAKER_01: Even if it says, I did not use AI, be explicit.
• 5:58
  SPEAKER_01: Because if you're not explicit, now we're all suspicious, right?
• 6:03
  SPEAKER_02: I was quite surprised when my 12-year-old said, yeah, my teachers just use AI for everything.
• 6:08
  SPEAKER_02: I had never thought that she would notice that.
• 6:12
  SPEAKER_01: Yes. And again, I'm not even 100% sure they are using AI.
• 6:17
  SPEAKER_01: But it is in the air.
• 6:19
  SPEAKER_01: It's in the water.
• 6:20
  SPEAKER_01: You cannot ignore the fact that we need to talk about it and say, here's how I'm using AI.
• 6:26
  SPEAKER_01: Here's how I'm not I'm not using AI because the assumption now,
• 6:30
  SPEAKER_01: almost the built-in assumption, I think,
• 6:32
  SPEAKER_01: if something feels even a little bit artificial, it's AI.
• 6:36
  SPEAKER_01: I've been doing education for a long time as the old person in the room, right?
• 6:41
  SPEAKER_01: And I remember I used to have templates for my responses to students because you see
• 6:47
  SPEAKER_01: the same error again and again and again.
• 6:49
  SPEAKER_01: I'm going to make this comment 12 times.
• 6:51
  SPEAKER_01: You better believe that I have it somewhere saved and I'm copying and pasting.
• 6:58
  SPEAKER_01: And putting it in and adding to details.
• 7:00
  SPEAKER_01: And that's part of my feedback.
• 7:02
  SPEAKER_01: So it would feel AI-ish, I guess, at least to me.
• 7:07
  SPEAKER_01: But yeah, so I think that the disclosure is really important,
• 7:11
  SPEAKER_01: but really that nuance of reading level and how direct it is and all
• 7:17
  SPEAKER_01: of that is really, really interesting as a way to think about this is not passive
• 7:23
  SPEAKER_01: in some ways, right?
• 7:24
  SPEAKER_01: It's doing some work.
• 7:27
  SPEAKER_01: Just to shift a little bit, because I know you worked on TPAC,
• 7:30
  SPEAKER_01: and TPAC is one
• 7:32
  SPEAKER_01: of these very formative frameworks that we've had
• 7:38
  SPEAKER_01: in educational technology for the longest time.
• 7:41
  SPEAKER_01: And I think it's the most cited, one of the most cited frameworks,
• 7:46
  SPEAKER_01: if not the most cited frameworks in edtech.
• 7:48
  SPEAKER_01: So it's really an important one.
• 7:52
  SPEAKER_01: And what do you think AI changes
• 7:57
  SPEAKER_01: In TPACK or the way we think about TPACK.
• 8:00
  SPEAKER_02: Yeah. And that's, again, something I think a lot about.
• 8:05
  SPEAKER_02: Dr. Punya Mishra was my doctoral advisor.
• 8:07
  SPEAKER_02: And so we talk and we were actually just chatting today on Slack
• 8:11
  SPEAKER_02: about an article we did a few years ago and what we've learned
• 8:14
  SPEAKER_02: since then as we're hopefully writing a new version of it pretty soon
• 8:20
  SPEAKER_02: or a revised response to us from three years ago, what we've learned.
• 8:27
  SPEAKER_02: I think, you know, a lot of early on we were thinking,
• 8:30
  SPEAKER_02: is this different than other technologies?
• 8:32
  SPEAKER_02: Because really, TPACK is kind of technology agnostic.
• 8:37
  SPEAKER_02: It's not supposed to matter what the technology is, if it's a whiteboard, if it's,
• 8:41
  SPEAKER_02: you know, it's pencil and paper, whatever this tool is that we're using,
• 8:46
  SPEAKER_02: it's supposed to apply to everything.
• 8:48
  SPEAKER_02: And so the question was, is this needed to, is it really going to change TPACK?
• 8:53
  SPEAKER_02: Or is it just applying another technology to it?
• 8:56
  SPEAKER_02: And there's some qualities, I think,
• 8:58
  SPEAKER_02: of generative AI specifically that are important to think about.
• 9:02
  SPEAKER_02: And the one we're landing on most lately is the social aspect,
• 9:06
  SPEAKER_02: that it kind of forms this other social being.
• 9:10
  SPEAKER_02: And so it brings in more of an importance of context and kind of the social emotional piece.
• 9:19
  SPEAKER_02: It really just if we have contextual knowledge in TPACK,
• 9:23
  SPEAKER_02: but it's just almost just needs to be strengthened a lot
• 9:28
  SPEAKER_02: because if teachers are using this tool, for example, even the disclosure,
• 9:31
  SPEAKER_02: like if I'm going to use AI feedback, I need to know enough.
• 9:34
  SPEAKER_02: How are my students going to interpret this?
• 9:36
  SPEAKER_02: And this goes more to a contextual knowledge.
• 9:40
  SPEAKER_02: Which has different definitions, but how are they going to interpret it?
• 9:44
  SPEAKER_02: And if I put an AI statement,
• 9:46
  SPEAKER_02: that's going to help them think differently about the feedback they get.
• 9:50
  SPEAKER_02: First of all, they'll just know that I'm being honest, but also they'll know, OK,
• 9:54
  SPEAKER_02: I got to think about whether I agree with this.
• 9:56
  SPEAKER_02: This isn't like the gospel coming at me and helping them understand that.
• 10:02
  SPEAKER_02: And the other thing is just like with these chatbots and stuff
• 10:05
  SPEAKER_02: that students can form relationships with and have had a lot of ethical problems,
• 10:10
  SPEAKER_02: personal, emotional problems with the need to think carefully about that in context.
• 10:16
  SPEAKER_02: So we can focus just on education, but the world around education changes.
• 10:21
  SPEAKER_02: And so what does that mean for what we're teaching?
• 10:25
  SPEAKER_02: In some ways, AI can embed TPACK itself in a way.
• 10:29
  SPEAKER_02: And it can make really interesting things.
• 10:30
  SPEAKER_02: Some of my students right now are creating different tools for teaching different content area.
• 10:36
  SPEAKER_02: And the vibe coding and being able to do that really lets you explore, like,
• 10:41
  SPEAKER_02: what are the affordances?
• 10:43
  SPEAKER_02: What do I put in here to support the content?
• 10:46
  SPEAKER_02: Really interesting things.
• 10:48
  SPEAKER_02: But that context piece, I think, is the big part to really think about now.
• 10:54
  SPEAKER_01: And I think we're seeing some of this in the way school districts, for example,
• 10:58
  SPEAKER_01: are thinking about AI, and that is they're thinking there's a difference
• 11:02
  SPEAKER_01: between giving AI to administrators and teachers and giving AI to students.
• 11:08
  SPEAKER_01: And that may be answering some of those.
• 11:11
  SPEAKER_01: And hovering over this is a whole host of questions,
• 11:14
  SPEAKER_01: but one of them is the question of, do I want to give them these tools,
• 11:19
  SPEAKER_01: access to these tools that they may be using in ways I cannot anticipate?
• 11:23
  SPEAKER_01: That will have detrimental effects.
• 11:27
  SPEAKER_01: And I don't want to be the place where that happens.
• 11:29
  SPEAKER_01: I mean, schools can't stop that from happening.
• 11:32
  SPEAKER_01: And it's just like social media from that perspective.
• 11:35
  SPEAKER_01: Kids are walking with it in their pockets and they're interacting with this all the time.
• 11:40
  SPEAKER_02: You know, with social media,
• 11:42
  SPEAKER_02: there was so much research around how do we use social media and learning?
• 11:46
  SPEAKER_02: And then the big impacts of social media come from outside of education.
• 11:51
  SPEAKER_02: They come from like the fake news and all these elements that we were so focused on.
• 11:57
  SPEAKER_02: How can we use this to help students share their knowledge, which isn't a bad thing.
• 12:00
  SPEAKER_02: It's fine.
• 12:01
  SPEAKER_02: But we're missing the bigger picture of how things change around teachers
• 12:07
  SPEAKER_02: and it's impacting them in major ways that are outside of their pedagogy.
• 12:12
  SPEAKER_01: Yeah, absolutely.
• 12:16
  SPEAKER_00: Another question that comes up a lot when we are talking about AI is about biases.
• 12:21
  SPEAKER_00: So I'm curious to know what hands-on activities can teacher educators use
• 12:27
  SPEAKER_00: to help teachers and students recognize hidden biases in AI?
• 12:34
  SPEAKER_02: Yeah, thanks for asking that.
• 12:37
  SPEAKER_02: We're working a lot on that with the teachers we work with.
• 12:41
  SPEAKER_02: In fact, we have equityandai.net, and we have a lot of activities on that.
• 12:46
  SPEAKER_02: They're in different stages of development.
• 12:48
  SPEAKER_02: Some are better than others.
• 12:49
  SPEAKER_02: But we're also developing some tools.
• 12:51
  SPEAKER_02: They're password protected right now.
• 12:54
  SPEAKER_02: But we're going to develop them more, I think, and make them more public.
• 12:57
  SPEAKER_02: But I think the big thing, and teachers have helped us with this,
• 13:03
  SPEAKER_02: is images are huge.
• 13:05
  SPEAKER_02: Students can see images and understand a lot.
• 13:10
  SPEAKER_02: And it's important whenever you do this. Well, let me let me back up.
• 13:13
  SPEAKER_02: So, for example, we have a teacher who is actually a special education teacher
• 13:19
  SPEAKER_02: and in third to fifth grade.
• 13:21
  SPEAKER_02: And they were working on like a standard.
• 13:23
  SPEAKER_02: And I think it's always important to embed it in the content. Right.
• 13:26
  SPEAKER_02: So they're they're working on setting and how to setting impact the story.
• 13:33
  SPEAKER_02: And that that's a specific standard she was working on.
• 13:36
  SPEAKER_02: And so they start with the three little pigs, and they say to AI,
• 13:39
  SPEAKER_02: to just ChatGPT image generator, create an image of the three little pigs in India.
• 13:46
  SPEAKER_02: And this class, they're teachers from India, and they have kind of like a sister class in India.
• 13:50
  SPEAKER_02: So they have a feeling of what the culture is like that a little bit.
• 13:53
  SPEAKER_02: And it produces these pictures that have these huts and things.
• 13:57
  SPEAKER_02: And they ask their teacher, wait, did you live in a hut in India?
• 14:01
  SPEAKER_02: Like, what is this?
• 14:03
  SPEAKER_02: And and then they ask for one in and we're right at the border with Mexico and in Mexico.
• 14:09
  SPEAKER_02: And they produce a lot of pictures.
• 14:12
  SPEAKER_02: And that's important because they need to know that this is about patterns.
• 14:15
  SPEAKER_02: It's not about a one-off thing. It's about patterns.
• 14:17
  SPEAKER_02: And so the tools we have online let you produce, you know,
• 14:20
  SPEAKER_02: five pictures of the same thing so you can look at the pattern across compared
• 14:24
  SPEAKER_02: to the pattern of the other thing.
• 14:26
  SPEAKER_02: But, you know, the students are very perceptive.
• 14:28
  SPEAKER_02: They say, look, AI doesn't treat every country the same.
• 14:32
  SPEAKER_02: It's strong on stereotypes.
• 14:33
  SPEAKER_02: And it's just why is it doing that?
• 14:34
  SPEAKER_02: What does that mean about society?
• 14:37
  SPEAKER_02: What does it mean about AI?
• 14:39
  SPEAKER_02: She also had them, some of their students created like a classroom and noticed
• 14:43
  SPEAKER_02: that there weren't kids in wheelchairs, right?
• 14:45
  SPEAKER_02: And maybe that's something, you know, they can fix with coding, but still like, why is that?
• 14:50
  SPEAKER_02: Why is the default not to have that?
• 14:52
  SPEAKER_02: And what does that mean?
• 14:53
  SPEAKER_02: And it leads to a much broader conversation about society.
• 14:57
  SPEAKER_02: And it's about AI, but it's bigger than that too.
• 15:02
  SPEAKER_02: One other example I want to give is a veterinarian class, a veterinarian science,
• 15:07
  SPEAKER_02: and They asked, the teacher asked the students to create a treatment plan with AI
• 15:13
  SPEAKER_02: for a cat with pneumonia.
• 15:17
  SPEAKER_02: And sometimes they said it was for a city vet and sometimes it was for like
• 15:20
  SPEAKER_02: a country, like they didn't use country vet, but that's how I imagine it,
• 15:24
  SPEAKER_02: like a low resource vet.
• 15:26
  SPEAKER_02: And they noticed that when they said the city vet, it usually said to give the cat oxygen.
• 15:32
  SPEAKER_02: And when they said that it was for like a lower income clinic, it didn't suggest that.
• 15:38
  SPEAKER_02: And that was eye opening about AI and bias.
• 15:41
  SPEAKER_02: It also embeds in the content saying, wait, what should you do?
• 15:45
  SPEAKER_02: Should we give it oxygen?
• 15:46
  SPEAKER_02: What if they can't pay for it?
• 15:48
  SPEAKER_02: Do we still offer it?
• 15:49
  SPEAKER_02: And so it opened a very complex discussion that is about AI,
• 15:54
  SPEAKER_02: but it's also about content and equity and all those elements.
• 15:58
  SPEAKER_02: So just producing stuff.
• 16:00
  SPEAKER_02: In a content area that allows them to see the differences and looking for patterns
• 16:05
  SPEAKER_02: or doing a lot of repetitions of something and finding the pattern in it,
• 16:08
  SPEAKER_02: I think is what really we've seen be successful with students.
• 16:13
  SPEAKER_01: All right.
• 16:14
  SPEAKER_01: And do you know when you're going to have some of those websites open so we can all
• 16:19
  SPEAKER_02: use them?
• 16:20
  SPEAKER_01: Just a general projection, not a date.
• 16:23
  SPEAKER_02: Right now we have equity in AI.net and it has a lot of learning.
• 16:28
  SPEAKER_02: So on it, there's critical AI explorers.
• 16:30
  SPEAKER_02: And if people are interested in using them, I can set up a password for you to use for a while.
• 16:34
  SPEAKER_02: The problem we have is that you can use it just to access ChatGPT all you want.
• 16:39
  SPEAKER_02: And so we need to work on putting some safeguards there.
• 16:42
  SPEAKER_02: And so right now we're giving classrooms individual passwords for set periods of time.
• 16:47
  SPEAKER_02: We do have some funding coming through that we're hoping to just,
• 16:50
  SPEAKER_02: we'll have to watch the usage,
• 16:52
  SPEAKER_02: but Just totally open it up and see what happens and maybe put some limits
• 16:57
  SPEAKER_02: on individuals or something.
• 16:58
  SPEAKER_02: But hopefully soon. We're hoping soon.
• 17:01
  SPEAKER_02: But people can reach out to me too if they would like access because we can do that too.
• 17:07
  SPEAKER_01: And that's a curious thing is we don't think about the usage
• 17:10
  SPEAKER_01: because if you're using FreeChat GPT,
• 17:12
  SPEAKER_01: it seems almost endless until you run out of it.
• 17:15
  SPEAKER_01: But you have to think about it in scaling up.
• 17:19
  SPEAKER_01: That becomes a problem really, really fast.
• 17:22
  SPEAKER_01: And so we are not really processing the costs from that perspective.
• 17:28
  SPEAKER_01: And the next question that kind of builds on that,
• 17:31
  SPEAKER_01: and that is how can teacher education programs move beyond just teaching
• 17:36
  SPEAKER_01: a little bit about AI or AI skills or AI literacy to really develop
• 17:41
  SPEAKER_01: that critical literacy that you talk about?
• 17:44
  SPEAKER_02: Yeah. I think a lot of it is the same things.
• 17:47
  SPEAKER_02: I mean, These activities that we have and the ones I'm describing
• 17:50
  SPEAKER_02: to you are teachers that were in our workshops and they designed these and we had,
• 17:55
  SPEAKER_02: you know, they're in service teachers,
• 17:56
  SPEAKER_02: but I've done the same thing with pre-service teachers said,
• 17:59
  SPEAKER_02: let's take a content area.
• 18:00
  SPEAKER_02: Let's take a standard.
• 18:02
  SPEAKER_02: What do you think you would see in this?
• 18:05
  SPEAKER_02: And really investigating and talking about it.
• 18:08
  SPEAKER_02: But really, it's the same thing.
• 18:09
  SPEAKER_02: It's designing those activities and then just an exploration.
• 18:13
  SPEAKER_02: So my students do like a critical AI exploration where they take their content area
• 18:18
  SPEAKER_02: and they think about how could bias show up in this and they explore it.
• 18:22
  SPEAKER_02: So recently I got one from a student where they said
• 18:25
  SPEAKER_02: that the students were Spanish speaking and it was it was creating like images
• 18:31
  SPEAKER_02: of students with different sounds, like making different sounds for reading. Right.
• 18:36
  SPEAKER_02: And the voice.
• 18:37
  SPEAKER_02: And if they had a secondary second language learner, it would do these little animations.
• 18:43
  SPEAKER_02: Versus more realistic looking people if they don't say that.
• 18:46
  SPEAKER_02: And that's a huge thing because you need the realistic looking people
• 18:49
  SPEAKER_02: to understand how the sound is made.
• 18:51
  SPEAKER_02: And so little things like that that they can explore,
• 18:55
  SPEAKER_02: it really provides an integrated view on their content in addition to the AI
• 19:01
  SPEAKER_02: and the critical perspectives.
• 19:05
  SPEAKER_01: All right. So it hides everywhere is what you're saying.
• 19:09
  SPEAKER_01: But what I love that you've said And that is that in some ways it's about AI,
• 19:16
  SPEAKER_01: but in other ways it's really exposing the hidden curriculum,
• 19:22
  SPEAKER_01: the way society is biased in one way or another.
• 19:25
  SPEAKER_01: And we've known that for a while and some people resist that and all of that.
• 19:28
  SPEAKER_01: So we're having this discussion.
• 19:29
  SPEAKER_01: But it's really hard to argue against that evidence that it's doing these things.
• 19:36
  SPEAKER_02: Right. And you have to think, why?
• 19:37
  SPEAKER_02: Where does this come from?
• 19:38
  SPEAKER_02: And it can open big conversations.
• 19:42
  SPEAKER_01: Yeah, absolutely.
• 19:44
  SPEAKER_00: So another question, which is again about AI biases,
• 19:48
  SPEAKER_00: how can we still use AI if it's biased in ways that are not transparent?
• 19:53
  SPEAKER_01: Because I'm going to still use AI.
• 19:56
  SPEAKER_02: Yeah, so am I.
• 19:58
  SPEAKER_02: That's a misunderstanding.
• 19:59
  SPEAKER_02: I guess sometimes with my research is like, so you're against it.
• 20:01
  SPEAKER_02: I'm like, no, it's not that simple.
• 20:05
  SPEAKER_02: Again, I'm not on the Illusion that as a person,
• 20:09
  SPEAKER_02: I can completely be critical and not be duped by it sometimes, right?
• 20:13
  SPEAKER_02: It's just human nature.
• 20:15
  SPEAKER_02: But really critically analyzing the best I can and thinking about it very carefully.
• 20:21
  SPEAKER_02: Part of it is in how you use it.
• 20:23
  SPEAKER_02: What uses are you choosing?
• 20:25
  SPEAKER_02: Part of it is that disclosure piece.
• 20:28
  SPEAKER_02: I want to make sure that I'm disclosing that I'm using it,
• 20:31
  SPEAKER_02: helping others see it not as the ultimate truth.
• 20:36
  SPEAKER_02: Another thing, and It's related, I guess, at least in my mind,
• 20:41
  SPEAKER_02: is I'm forcing myself these days to have time in the morning where I will not go
• 20:46
  SPEAKER_02: to AI because it helps me develop my own critical thinking.
• 20:51
  SPEAKER_02: And it's hard sometimes because it's like,
• 20:53
  SPEAKER_02: why don't I just go get a review of this article on AI really fast?
• 20:56
  SPEAKER_02: Because that'll be so much faster than reading this right now.
• 20:58
  SPEAKER_02: I'm like, no, I am not going to AI right now.
• 21:02
  SPEAKER_02: And it forces me to continually think and develop And it changes how I interact
• 21:07
  SPEAKER_02: when I do use AI because I'm grounding myself constantly in my own thinking.
• 21:15
  SPEAKER_02: And so I think teaching students to do that and just our own use,
• 21:20
  SPEAKER_02: really thinking about that is really important.
• 21:23
  SPEAKER_01: Yeah.
• 21:25
  SPEAKER_01: Kind of feeling similarly.
• 21:28
  SPEAKER_01: I use AI a lot.
• 21:30
  SPEAKER_01: It runs my life at this point, at least my professional life.
• 21:33
  SPEAKER_01: But at the same time, that time away,
• 21:38
  SPEAKER_01: today we sat together and we started writing and just composing sentences.
• 21:43
  SPEAKER_01: And it's a little bit harder, but it's also completely mine.
• 21:49
  SPEAKER_01: And it also makes me appreciate AI when it does solve some problems.
• 21:54
  SPEAKER_01: But at the same time,
• 21:56
  SPEAKER_01: it's definitely that time away helps frame things
• 22:01
  SPEAKER_01: and avoid the slop.
• 22:03
  SPEAKER_01: Because what I'm sensing is when I'm using AI in large quantities,
• 22:08
  SPEAKER_01: I have that sense of, I read some of it. It's good enough.
• 22:12
  SPEAKER_01: And it's like, no, it actually isn't.
• 22:14
  SPEAKER_01: It's mostly slop.
• 22:15
  SPEAKER_01: But it's easy to say, there's stuff there.
• 22:18
  SPEAKER_01: Here, you deal with it.
• 22:21
  SPEAKER_01: And so...
• 22:22
  SPEAKER_01: Helping our students develop.
• 22:24
  SPEAKER_01: And I'm thinking, I mean, you know, all of us grew up in the world before AI.
• 22:30
  SPEAKER_01: But our students, as they are emerging and climbing through, are going to be less and less so.
• 22:35
  SPEAKER_01: We've got the people with our students in engineering we're working
• 22:40
  SPEAKER_01: with and asking them questions, interviewing them.
• 22:42
  SPEAKER_01: And I mean, I keep going back to that moment where we ask them,
• 22:46
  SPEAKER_01: how did AI change the way you learn?
• 22:49
  SPEAKER_01: And I'm like, we've been using it since high school.
• 22:53
  SPEAKER_01: And we're like, so in a few years, they're not going to know a world before AI.
• 22:58
  SPEAKER_01: And that presents, they don't have that counter.
• 23:01
  SPEAKER_01: And we need to create those opportunities.
• 23:05
  SPEAKER_02: It reminds me of another thing I really try to do with anyone I'm working
• 23:08
  SPEAKER_02: with is we talk about one-to-one devices so much.
• 23:12
  SPEAKER_02: But one-to-three, have the students be talking to each other
• 23:16
  SPEAKER_02: while they're interacting with the AI, again,
• 23:19
  SPEAKER_00: helps them see the difference between a human being.
• 23:22
  SPEAKER_02: And it helps them see different perspectives on something instead of just
• 23:25
  SPEAKER_02: the one they're getting in front of them.
• 23:28
  SPEAKER_02: I think that's a big part of it, too.
• 23:29
  SPEAKER_02: It's just keeping those human connections as you're doing it and understanding the difference.
• 23:39
  SPEAKER_01: All right.
• 23:40
  SPEAKER_02: Yeah.
• 23:41
  SPEAKER_01: Thank you.
• 23:42
  SPEAKER_01: It's been lovely.
• 23:43
  SPEAKER_01: It was good.
• 23:45
  SPEAKER_01: I'm going to take my students to the website and try to see what we can do
• 23:51
  SPEAKER_01: because I think that it's an opportunity.
• 23:53
  SPEAKER_01: We did talk about the ethics of AI and we did talk about bias,
• 23:57
  SPEAKER_01: but talking about bias
• 23:59
  SPEAKER_01: and seeing bias are two very different things as we all know.
• 24:04
  SPEAKER_01: So I'm thankful that somebody worked on this and there's a way to get that Yeah.
• 24:11
  SPEAKER_02: Yeah.
• 24:12
  SPEAKER_02: And we'd love to hear, you know, if anyone tries it out, we'd love to hear.
• 24:15
  SPEAKER_02: We're going to put a survey on there soon.
• 24:18
  SPEAKER_02: Again, some of the activities are better than others because, you know,
• 24:22
  SPEAKER_02: some of them are AI created by teachers and we just put them up there to explore
• 24:25
  SPEAKER_02: and we're curious how they work.
• 24:27
  SPEAKER_02: And some of them are more carefully designed.
• 24:30
  SPEAKER_02: And so we're curious to see how the different versions and such work or don't work.
• 24:36
  SPEAKER_01: That's fantastic. So thank you.
• 24:37
  SPEAKER_00: Yeah. Thank you so much.
• 24:39
  SPEAKER_01: And we'll see you next time on Azadeh. And Guy.
• 24:42
  SPEAKER_01: On AI.
• 24:44
  SPEAKER_00: Thank you.