How to Teach AI with Experiments & Ethics
Transcript edited for clarity and grammar.
00;00;00;03 – 00;00;03;16
John Davis
This is the 10 Minute Teacher podcast with your host, Vicki Davis.
Free Educator Wellness Toolkit from Show Sponsor, Advancement Courses
00;00;03;24 – 00;00;22;00
Vicki Davis
It’s back to school and educators. You can get your free educator wellness toolkit from today’s Sponsor Advancement Courses. Stay tuned at the end of the show to learn how to register for your free kit and how to receive 20% off your next Advancement Course.
Introducing Roberta Freitas
00;00;22;00 – 00;00;47;15
Vicki Davis
We’re talking with Roberta Freitas, EdTech coordinator at Ibeu, an English Language School in Rio de Janeiro, Brazil. And I was first introduced to Roberta as she presented at ISTE. I couldn’t attend ISTE this year in person, so I attended online; hers was an incredible presentation I attended virtually. Roberta, thanks for coming on the show.
00;00;47;24 – 00;00;52;04
Roberta Frietas
Thanks for the invitation. It’s a pleasure to be here, and I’m excited.
How Roberta Began Using AI in the Classroom
00;00;52;17 – 00;00;56;27
Vicki Davis
Roberta, let’s talk about how you are using AI in the classroom.
00;00;57;11 – 00;01;22;03
Roberta Frietas
Last year, I realized the importance of getting our students acquainted with AI and how much it is embedded into our daily lives. It’s closer than we even think. So I started thinking about how we could do it and bring this into the classroom.
00;01;22;09 – 00;01;53;25
Roberta Frietas
We are a language school. We teach English to these students. So I started to think about how we could bring those two things together, like English and AI, and we began developing activities for teachers.
So part of my job is helping teachers start using new technology and tools. So, I first began preparing activities for them because it’s easier if they have something ready. But, of course, they can adapt, and I want them to adapt to their students’ realities.
AI and Machine Learning Experiments
00;01;54;07 – 00;02;11;04
Roberta Frietas
But we got some activities ready for them where they experiment with machine learning or an AI experiment. They can look at algorithms, discuss them with their peers, understand how it works, and play a little bit.
00;02;11;04 – 00;02;20;17
Vicki Davis
With AI in your presentation, you talk about this AI experiment. Explain what an AI experiment is and give us an example or two.
Google AI Experiments
00;02;21;00 – 00;02;49;27
Roberta Frietas
So Google has this page full of AI experiments that you can try, and so you have like the teachable machine, which is an experiment of machine learning.
Experiment #1 – The Teachable Machine
So you can teach this machine whatever it is you want this machine to learn. So you input images, and then you train the machine to understand what the output should be. So I could train the machine to recognize if I’m wearing a mask or not.
00;02;50;05 – 00;03;10;01
Roberta Frietas
So I put lots of images of me wearing the mask, and I say “mask,” and the output could be a gif, a sound, or speech. And then I do the same without a mask. And you get lots of inputs, lots of images from different angles. And then the experiment is ready. You can even get a video of that.
00;03;10;01 – 00;03;34;28
Roberta Frietas
So students could be trained as the machine to do anything so they could use it to solve someone else’s problem. Create like dealing with empathy. Try to create something, a device that could make someone else’s life easier.
Experiment #2 – Shadow Art
Another experiment there is one which is called Shadow Art, and it deals with the signs of the Chinese zodiac.
00;03;35;19 – 00;03;51;19
Roberta Frietas
So there are lots of animals, and this is one we use a lot with kids because then they have to do the shadows with their hands. So they have to recreate the animal. So it’s a fun experiment to get the kids to start understanding AI.
How to Coach AI Experiments
00;03;51;26 – 00;04;09;20
Vicki Davis
So as your students, for example, you train it with the pictures as they’re doing these experiments. When you have an investigation, you’re trying something out, and then you’re making observations. How should that coaching process or that conversation work between teachers and students?
00;04;09;23 – 00;04;35;00
Roberta Frietas
So when I think of an activity, I like to think that first, they’re going to talk a little bit about AI or specifically machine learning or algorithm, whatever aspect of AI they’re going to deal with. Then they’re going to tinker, play with it, experiment a little bit to understand. Then they’re going to reflect on that experience.
00;04;35;03 – 00;04;58;02
Roberta Frietas
So how did it go? What went well? What was hard? What can machines do? What can they do very well? And then, they will create a prototype or think of improving. So there could be in this experiment or something like that. So I always tried this process for teachers.
00;04;58;04 – 00;05;00;16
Roberta Frietas
They go deeper into that.
AI is Only As Good as the Feedback It Gets
00;05;00;26 – 00;05;20;09
Vicki Davis
The challenge, and this is just an ethical issue that I talk to my students about with artificial intelligence, is that it’s only as good as the feedback it gets from humans. So many of the AI tools you probably know about and that I learned out about ISTE dealt give you a method for feedback. I’ll give you an example.
Experiment #3 – Verse By Verse
00;05;20;09 – 00;05;40;06
Vicki Davis
There’s this thing called Verse by Verse, which you probably know about where the famous poets of history help coach you in writing a poem.
So last week, when I learned about it, I wrote a poem that, as I was in this session listening, someone pointed out that there is not much diversity in these poets.
AI Should Have a Way to Report Flaws
00;05;40;19 – 00;05;58;09
Vicki Davis
While I looked on the page and there was no place to report that or say, You forgot a flaw in your AI because you’ve not programmed enough diversity in your poets that you’ve used. And there was no way to give feedback.
So what do we talk about? That angle of things?
AI and Flaws of Diversity and Bias
00;05;58;23 – 00;06;26;20
Roberta Frietas
Yeah. And this is one of the reasons why I think it’s so crucial for us to teach AI in the classroom. We’re talking about AI in the classroom because there is a considerable problem of diversity and bias with AI. So we need these students to understand the limits of their bias. And so whenever they see, they have to be critical of what they are dealing with, the idea of dealing with, because it is an issue.
00;06;26;20 – 00;06;57;29
Roberta Frietas
And you were right; I don’t see a place for feedback on any of the AI experiments or things like that. So we need to get students to think about that. And students who belong to different races or social classes need to understand that they have to be part of it because that’s it. Humans are the ones who are training, and nowadays, most of the humans doing it are probably white, rich people with more access to technology.
00;06;58;09 – 00;07;15;09
Roberta Frietas
And those are the first ones to get there to get these opportunities. And we need to have everybody all kinds of people, all kinds of classes and genders. We need to have everybody on there so that we have different kinds of points of view, right?
Other AI Experiments
00;07;15;09 – 00;07;29;01
Vicki Davis
Yes. And let’s talk about other AI apps you like to use in experiments. And let’s mention those, and I’ll try to include those in the show notes because it is a bit mind-blowing when you start playing with these tools.
Experiment #4 – Use Everyday Apps Like YouTube, Spotify, and Netflix
00;07;29;06 – 00;07;56;21
Roberta Frietas
I like to use daily apps like YouTube, for example. So understanding how the algorithm, YouTube, and other social networks, how does the algorithm work?
So, what kinds of recommendations are you getting? Suggestions from the app are you getting, and why do you get those? Students also have to understand that they can create their algorithm as they interact with different people and different accounts make comments they like.
00;07;56;29 – 00;08;21;18
Roberta Frietas
They are creating this algorithm. So they need to understand how it works because whatever shows up when they watch a video, there is a recommendation afterward that’s not like out of the blue. And that happens because of the things they are interacting with.
Also, on Spotify or Netflix, all kinds of recommendation recommendations on what you should watch next on Netflix.
00;08;21;18 – 00;08;48;27
Roberta Frietas
Why do you get that? Because of this thing, the similar order shows that you have watched.
Experiment #5: Voice Assistants Like Alexa, Siri, & Google Home
So I like to use these things that they these apps that they use daily and also voice assistants like Alexa, Siri, and Google Home to understand how they interact with that.
Languages and Voice Assistants
For us here in Brazil, voice assistants are interesting because they would only speak English when they first showed up.
00;08;48;27 – 00;09;18;16
Roberta Frietas
So there was no Portuguese on Alexa, for example, only English. And it’s quite interesting to see the evolution of the assistant because, like my husband, he is a learner of English.
Question Structure and Voice Assistants
He is not proficient, but he’s been improving and taking classes, and when he first talked to it, it was tough for Alexa to understand how he was speaking because he would have made a full question.
00;09;18;19 – 00;09;48;17
Roberta Frietas
Question formation for us in Portuguese is different than in English. Like the order. So when he was trying to ask questions, it was not perfect, and many times Alexa would not understand. But then, with the evolution of Alexa, she started understanding even when he did not make the question correctly. So it’s quite interesting to see this development. At first, your language would have to be the pattern language and then some changes.
00;09;48;17 – 00;09;51;07
Roberta Frietas
There are some machine learning, whether on or something like that.
The Need for “Undo” with Algorithms Programming
00;09;51;27 – 00;10;26;17
Vicki Davis
The other problem with algorithms is there doesn’t seem to be an undo. Still, if you accidentally watch a video or listen to music that you don’t want to hear more of, you are curious; it’s almost impossible to undo that impact. And then you start getting a whole slew of information.
“My Students Can’t Leave the War Zone”
I’ll never forget I talked to a teacher from Jordan who had many Syrian refugees coming into his classroom. At the time, this was just terrible; terrible beheadings were happening and were being posted to YouTube from Syria.
00;10;26;17 – 00;10;51;25
Vicki Davis
And he said the problem he had was that his kids were still in the war zone, that because of the algorithms of YouTube, his kids could not leave Syria, and the trauma that was happening daily. They could not separate the kids from the trauma because they logged into their Google accounts and couldn’t break the algorithms on YouTube from showing them what was happening in Syria.
Helping Kids (and Us) Break the Algorithms That Chain Us to the Problems of Our Past
00;10;51;25 – 00;11;10;12
Vicki Davis
And this conversation we had what’s been of a good five or six years ago, but it still just stuck with me of if you’re intervening and trying to help a child with an eating disorder, you need to intervene and help break the algorithm. But there’s no method to do that unless they start entirely over.
00;11;10;12 – 00;11;34;06
Roberta Frietas
There is no undo button, as you mention. You can start liking, commenting, and interacting with things that are more relatable to you. Something that you like and would like to be seen to try to teach the algorithm the things that make sense to you, but other things. There’s no one undo button. There’s nothing like, No, “I don’t want to see this anymore.”
00;11;34;06 – 00;11;41;14
Roberta Frietas
It will keep showing for some time, depending on the location. That’s a very foreign notion to you.
00;11;42;02 – 00;12;17;28
Vicki Davis
I highly recommend Roberta Session. Roberta Freitas If you just purchased the ISTE where you can attend and watch for a longer like, I can go back and look at her session. (This session is not available for post-viewing, unfortunately.)
It was excellent.
AI Experiments
This whole idea of AI experiments, I think, should become part of every computer application and every computer science course. So as we make theories, we have conversations, we test the AI, and perhaps we could even use a different account to test the AI that’s going on behind the scenes because we have to train all of this that surrounds us.
The Need for Feedback
00;12;17;28 – 00;12;42;10
Vicki Davis
And it seems we sort of release these algorithms into the wild, and nobody’s taming them and giving feedback. And there’s just a lot; it’s the wild west of AI, and a lot needs to change.
And Roberta, your kind of thinking is pushing us towards helping kids understand what’s going on. It’s artificial intelligence.
The Need for Human Morals and Ethics
00;12;42;21 – 00;12;51;06
Vicki Davis
It doesn’t have morals or ethics. Any of that has to come from us. And then we need to start having this conversation. Anything you want to add, Roberta?
We Don’t Want Students to Be Victims of AI
00;12;51;06 – 00;13;10;14
Roberta Frietas
I just want to add that we don’t want students to be victims of AI, So this is why it’s so important they can be protagonists in this and try to change this story, changing the algorithm and change AI and bring the good side of it, because there are lots of benefits in it for sure.
00;13;10;27 – 00;13;15;28
Vicki Davis
Thank you for coming on the show, Roberta.
00;13;15;28 – 00;14;11;07
Advancement Courses Wellness Toolkit
Go to coolcatteacher.com/wellness and download your free educator wellness toolkit. From today’s Sponsor Advancement Courses, you’ll get access to activities, strategies, and tools to improve the four domains of wellness.
Also, you can save 20% off your next course by using the code Cat20. Yes, that’s CAT20. Just for our listeners and readers of the Cool Cat Teacher blog. I love their courses because they are flexible, with up to six months to complete. As I did, you can receive graduate credit through CFP and regionally accredited university partners or continuing education units that meet your state requirement. With over 280 courses to choose from, Advancement Courses offers subjects in everything from educator wellness to social and emotional learning and more.
00;14;11;09 – 00;14;23;22
John Davis
You’ve been listening to the Ten Minute Teacher podcast. If you like this program, you can find more at coolcatteacher.com. Also, if you wish to see more content by Vicki Davis, you can find her on Facebook and Twitter under @coolcatteacher. Thank you for listening.