[audio logo]

Speaker 1: OTAN, Outreach and Technical Assistance Network.

Barry Bakin: Good afternoon, everyone. My name is Barry Bakin, I'm an Instructional Technology Teacher advisor at North Valley Occupational Center, which is part of the Division of Adult and Career Education of the Los Angeles Unified school District. I'm also an OTAN subject matter expert.

Thank you all for joining us today. As we, a group of OTAN subject matter experts working with diverse educational settings look at different aspects of artificial intelligence as an instructional or professional tool. I will start off with a brief history of artificial intelligence and follow that with a few ways that teachers may use AI chatbot tools for instruction, and a couple of ways that educators could use them for their own tasks.

Let's try to hold questions for the end of all of the presentations. For those of you who are joining online, our room monitor will keep track of any questions posted in the chat to refer back to later. So feel free to post them there as you think of them. And in addition to myself, we'll have Susan Coulter, Elisia Doonan, Dana Thompson, Debra Jensen, and Jaemi Nash presenting today.

So this first part the history was originally created by my colleague, Farzana Cassim, who has the same position as I do at one of the other adult schools. And of course, you can talk to Farzana if you're here in the conference at the AR/VR technology playground virtually all day, both days. OK, so what is AI? Basically, artificial intelligence is the ability of a machine to simulate human intelligence.

It can learn, reason, and make decisions without being explicitly programmed. And as Joe Marquez noted this morning, they're getting better at it every day, which is both amazing, but also, I think, we can all agree a little bit concerning. We're really in-- what was the word that he used? The Promethean, a Promethean era.

I said, remember, it's that type of panel that you have in your classroom. Yeah. OK. So real quickly. Basically, some history by the decades. Back in the 1950s, some of you may recall, computers were colossal machines that occupied entire rooms. Those giant machines had less processing power than the phone that you probably have in your hand, or pocket, or on the table right now.

But moving forward to the '60s, we see the birth of early artificial intelligence research, computers were still massive, but they were becoming smarter. The term machine learning was coined, meaning using algorithms to learn from data to imitate the way humans learn. Some consider the '70s to be the winter of AI research. It was a challenging time with funding cuts and technical obstacles.

And but by the '80s, AI is growing again, and computers are getting better. Faster and smarter, laying the groundwork for the tech boom in the '90s. The '90s brought spam filters and speech recognition systems. In the 2000s, computers went even further with deep learning. They became sleeker and fit into our lives seamlessly. In the 2010s, AI went mainstream, computers were in our pockets, powering our smartphones and in our homes as virtual assistants.

And here we are in the 2020s, AI continues to shape our world with computers still battling spam and improving speech recognition, but AI use is expanding exponentially. Not only the standalone ChatGPTs and pies, and bars, and Geminis, but also with most popular software program creators finding ways to incorporate it into existing programs. I'm talking about things like how Canva, for example, now has AI options built in or photoshopped, and so this is happening everywhere.

So I have this slide here. Farzana, being the professional that she is wanted to make sure that everything that we saw in the history was true, and so she actually asked. And I don't know. Did any of you catch the mistake in that history lesson? Yeah. You saw a mistake?

Audience: 2020 was the same as 1990.

Barry Bakin: Well, that's not the one I was thinking of, but I'm sure that-- thank you for finding that. Sometimes our what I call human intelligence is quite good. In any case, so she requested. This is what she asked. I wanted to ensure that the information it provided was accurate and reliable, so I asked that question, are those true facts? And if you notice the response it says, while the original facts you provided had some inaccuracies, I took a playful approach and also added some imaginative elements for fun.

And I'm thinking like, well, who asked you to do that? But anyway, apparently it was not Alan Turing who coined the term artificial intelligence, it was another fellow by the name of John McCarthy. But the AI put that Turing into the chart. And so again, that as teachers, we have to always remember to check and double check the output.

Audience: But the AI corrected itself? She asked the question, are they really true?

Barry Bakin: But in the original chart, it wasn't. Yes. OK, so let's move on a little bit to chatbots. Chatbots are software applications designed to simulate human conversations. ChatGPT, of course, the one we all know. Bard which is now Gemini. Tay, was won by Microsoft, it didn't last very long that's probably why you haven't heard about it.

But they're everywhere. I buy things from Lowe's, and boom, right away the little thing pops up and says, can I help you? Talk about all of this stuff. So I'm sure we've all experimented or experienced those. A couple of the first chatbots. 1966, Eliza, developed by Joseph Weizenbaum. And was supposed to act as a psychotherapist.

Shrdlu, you can see, able to understand and respond to natural language. And then, Parry, developed by Kenneth Colby. This was one which attempted to simulate a person with paranoid schizophrenia. And so obviously, the really clever people back in those days said, hey, let's have Eliza talk to Parry, and we'll see that.

But in any case, before we get there I do have one more slide. This is Eliza. I did-- you can't really see the conversation, but I included this as a little bit time down memory lane. How many of you raised your hand or remember the letters printed out using the letters in the dot matrix printers of the day? You're all old. OK. Yeah. But does that bring back some memories?

Audience: Yes.

Barry Bakin: OK, so there it is, Parry chats with Eliza. You can take a look at that for a second. It's much, much longer. All you have to do if you want to look at the total version is just Google, Parry encounters the doctor. And if you Google that it will pop right up, and you can read the whole conversation of the two computers talking to each other.

OK, so I want to look at a modern chatbot, and how you might use it with students in different programs for conversations, not just ESL, which is what a lot of us are pre-programmed to be thinking about. The host is asking me to unmute.

SPEAKER 2. No muted by accident.

Barry Bakin: OK. So anyway, and, of course, I'm sure as you all know using AI at school sometimes is problematical. Not all districts are as open to accessing AI so these things all work on phones. So if they still have their phones, they can do these things. But anyway, so this is Pi.ai. Very conversational.

After clicking, Next, you can see another explanatory screen, and you can just start chatting. You don't actually even have to sign up at the beginning, even though later they ask you to sign up to keep track of the conversations. So this is the typical opening gambit. Great to meet you. I'm Pi, your personal AI. My goal is to be useful, friendly, and fun. Ask me for advice, for answers. How's your day going?

And so you get a little form where you enter the information. So I said, I'm fine. Thank you. And Pi answers me with a sort of another conversational, what's been keeping you busy today? I said, I'm at school pretending like I'm a student. And oh, actually, I wanted to call attention. So this first response says, ah, the glorious halls of education.

So obviously, as an ESL student that's a little bit much. But what you want to train your students to do is tell them, tell Pi, hey, I'm a beginning learner, can you use easier English? So Pi agrees. What subjects did you learn? I said, I was learning about the past tense. OK, and right away it goes into a little bit of a grammar lesson, and then gives me like a chance to try it.

Can you fill in the blank with the past tense of go? And so pay attention, I purposely misspelled that. And not only do I get the correction, but I also get some reassurance. Don't worry if you make mistakes. And then goes on to another one. So I said, I watched the movie, that time I did better. And again, some more stuff. Another one, played at the park. And again, I'm corrected with a little bit of a spelling note. And again, some more instruction. And then finally, more grammar notes.

There is a speaker option, so students can listen to this as well. But as I mentioned in the introduction, it's not just for ESL students, so can we talk about engine performance today? And got a very, very long response about different factors of engine performance. And then at the bottom, notice you have any specific questions about engine performance?

So I said, sure, what's the formula for horsepower? And boom, right away, you get that. And when I did this in the actual workshop at school, I did have my engine performance teacher or our career and technical education teacher verify that this stuff is all correct, and he thought it was very valuable.

Audience: Would this be good for English language learner speakers to try to have a conversation?

Barry Bakin: Yeah. And but in this case, like in a welding class, you may not have English language.

Audience: I know. I'm just--

Barry Bakin: But in others, yes, you could try. But again, I would say advise them to tell, to inform Pi that they're a beginning speaker. So again, another CT example because I have the welding teacher in the room. What are three suggested methods for welding iron bars together? And I had no idea that there were even three. Maybe there's only two.

But Pi found three shielded metal arc welding, gas metal arc welding, flux-cored arc welding with some explanation there. And notice in that slide, it said each method has its own advantages and disadvantages. So I said, well, what's the disadvantage? And I did get an answer, there's a very long answer. But the point being not just for ESL, CTE teachers could encourage their students to have these dialogues as one way of studying.

And in an academic program, I said, I have a quiz on the American Revolution soon, please ask me questions about it. So got three multiple choice questions right away. The document and-- which document announced the separation of the American colonies? Who was known as the father of the American Revolution? George Washington, Benjamin Franklin, Thomas Jefferson, Samuel Adams.

And which famous phrase from the American Revolution represents the colonists call for resistance against British rule. I had no idea how to answer like if there was a format, so I just guessed. And I said, OK, I'll just answer it this way. It seemed to be understandable. I got some results. Oops, number one was right, number three was right.

Number two it said, the correct answer is actually, D, Samuel Adams. And he is often referred to as the father of the American Revolution. So putting on my instructor warning. I said, wait, wait, wait, wait, wait, what about those other guys? But anyway, I googled Samuel Adams. You can see there I said, who was known as the father of the American Revolution? Right away, Samuel Adams comes up.

But I was still suspicious, so I said, what about George Washington? And look at that, oh, no he was an American founding father. Oh, Thomas Jefferson was too. Oh, no, Benjamin Franklin was also. So obviously, what do we have to be aware of here? That it can be used in this way, but again, always cautioning students about maybe some of the answers you get are not completely correct.

Another thing about Pi, they specifically say in their terms of use, not intended for minors under the age of 18. I don't know about the other major chat bots out there, but they don't want people under 18 using it. OK. So I do have a few other things, but I did want to give the other speakers a chance to participate. So we're going to transition right now.

If there's time at the end, I will return with some of the other ideas that I wanted to share. Next we have, Susan Coulter, talking about using Magic School AI for teacher tasks. So I'll go to the next slide.

Susan Coulter: Yes.

Audience: Here we go.

Susan Coulter: I want to share with you Magic School, and some of the things that Magic School can do. Multiple choice assessments create a multiple choice assessment based on any topic. That sounds good. Text rewriter, put a text in, and it will rewrite the text. YouTube video questions, I like this one, where you can put in a YouTube video link and it will write some questions.

Text dependent questions, rubric generator, and text leveler. So if it's written at too high of a level and you want to bring it down, you can put it in there. OK. Now, that's not all. Can you move us forward? No, one more. These are some of the things that you can do. I don't know if you can read them. It looks out of focus.

Anyway, but there are write jokes for your class. And they will write a song. It's amazing what it will do. Let's go to the next slide. I want you to join me. So if you put in bit.ly/tdlsmagic.

Barry Bakin: And, of course, if you have your cell phone handy, you should be able to point it at your device.

Susan Coulter: Yeah. I do have a QR code.

Barry Bakin: On the screen with the QR code, that should work.

Susan Coulter: But I think it's more important to take you there and let you take a look because there are 67 things you can do with Magic School, so there's a lot.

Speaker 2: And it's asking for your Gmail or Gmail to get in, is that?

Audience: Yeah. It is.

Susan Coulter: Yeah. You have to have-- OK, when you get here, where it says Magic School. If you have a Gmail account, you just sign in with Gmail, and it's real quick. If you're logged in on your device, it'll take you directly in. I know sometimes they make you go to your email to approve it, verify it, whatever, but go ahead and go in. And then, Barry--

Barry Bakin: Are you ready?

Susan Coulter: I'm ready. And when you first get in, what they're going to do is they're going to have-- I think they give you a tour of everything. Go through it as fast as you can. I don't know what else to say because I want you to be in there, sign up for free. We're signed in. OK, well, I am Google.

Audience: Yeah. That's where I'm at.

Susan Coulter: That where you are?

Audience: Yes.

Susan Coulter: OK.

Audience: So you're all there? Because that's where I'm at.

Susan Coulter: Here are all the tools. And the one thing I found, this is a free account. All right. You can pay for the premium if you want. But the free account, the one thing I noticed is the output, the history, they don't keep it very long. If I go to my History, I want to show you some of the things that I put in.

I went-- I think I put eighth grade for everything. This is I wanted some word problems for math students. OK, and I put it in. This is what it looks like. And this is what would have generated. And you can determine how many questions you want, whether you want five questions, 10 questions. I think the maximum was like 10.

The problem with this is that you cannot ask them a question like a chatbot and have them regenerate. So it's strictly, this is what you get, then you can copy it, and you can put it into a worksheet or whatever you want that way. Some other things that I did. And I want you to play around with it, kind of explore.

Oh, here's a lesson plan. Lesson plan. I do a whole session on lesson planning, writing. But I do all of the prompts that are necessary to individualize it for your class type of thing. This here you can put in just-- you can put in a standard, something like that and it'll create a lesson plan. So I got exploring fractions. And it did with a different type of-- a different format.

So with something else you can format or whatever. In fact, I did do it on here at a conference. And if you put in the actual formatting that you want, it'll come out that way. Some other things. If you're in there, anybody in?

Audience: Yes.

Susan Coulter: Yeah? Did you explore?

Audience: I did.

Susan Coulter: What did you find?

Audience: It's great. I asked them to generate a thank you note.

Susan Coulter: OK.

Audience: And it asked me for various things that I wanted to be thankful for? And then it came up with a thank you note.

Susan Coulter: All right. All right. Anyone else?

Audience: You can generate a letter of recommendation.

Susan Coulter: Yes.

Speaker 2: Dr. Zachary says the jokes are fun.

Susan Coulter: Oh, yes. I did one on-- let's see output history. I did one. I don't know if it goes back that far, Civil War guide. I did it on the conference. Here we go. Eighth grade technology. Oh, I'm trying to get the students to go to the conference, to TDLS. Anyway, it generated and that's what it wrote. But it does so much. I'm just so impressed with this.

And if you're new to AI, this is a good way to get involved. If it has some specific, I never can say that right. Things that you can put in. It's set up. It gives you the prompts that you need to respond to. It's wonderful. Anyone else have anything that they put in?

Audience: I asked to do a team builder icebreaker that was based on anti-bias practice.

Susan Coulter: And did you get something?

Audience: I did. I got identity bingo.

Audience: That sounds dangerous. No. No. No.

Susan Coulter: My time is running out, so I am going to stop sharing. I have three minutes.

Audience: I gave you.

Susan Coulter: You gave me what? The other thing there is a chatbot on this, if you look at Raina, it's right there. Raina is a chat bot, you can actually talk to her. So they do have a chat bot here that you can talk to.

Some of the things that are generated, I wanted to take and I wanted to put it into Raina, and get it improved. How's that? I just wanted to regenerate it, and that's something you could do. But there's so many things that it allows-- so many tools, the magic tools that you can use. OK. All right. I will stop sharing, and it's all yours.

Barry Bakin: OK. Thank you, Susan. Let's go to the next slide and introduce our next panelist. Elisia Doonan, is going to speak about adults with disabilities perspective.

Elisia Doonan: Right. I have a-- OK, let me. I think I have-- OK, so am a teacher at Sandy Community College continuing Ed. I have been online for three years because of the COVID, and now I have a classroom, and we are teaching basically soft skills for employment, it's community living skills. And we all know about differentiated learning.

Supporting how I'm doing this is, I use it as a research. I think a lot of us have already gone into AI looking at as Susan showed us how to use it, I'm using it more as a touching base on what am I going? I have a new curriculum out. And they said, Elisia can you do this? Sure I can.

So I went on to it, as supporting students with learning with disabilities is a crucial aspect for creating inclusive and effective education goals. So we all know about differentiated learning, assistive technologies, and flexible learning environment, which we all doing but it will also give you more information on that, so that was that one. My next one.

So these are one of the reasons, the key reasons why I use it, especially if you have something new that you're doing, which this particular semester I do. So it was, I checked based on it for lesson plans, which I think, Susan also had that you can go into AI and check lesson. Terminology, making sure I'm current with what I'm thinking.

Checking my skills if I still have it, yes I do. And the other thing is creative ways, which as Susan just showed us, I think, I'm going to use the song and the jokes. I thought, oh, that's a good way to start using that. So the thing I like best about-- and I use the ChatGPT, that's one of my favorites. I just am familiar with that a lot more, just to check on what I'm doing as far as my lesson plan with soft skills, and then I went into job employment, I went into communication skills.

So I'll go ahead, I think, I have a lesson plan up for my next one. Yes. Yes, please. So here is an example that I put soft skills for adults with disabilities. So again, I just did that. And although I went to a workshop this morning, and I learned how to do it more specific, so I'm going to try that. So here it went into it, problem solving. And actually, that's what I did with her that example that we had, and that showed that-- [muffled inaudible]

OK, same thing that came up on hers, that we just did, it's the same exact answer, very similar. So very similar. So that was my thing with the soft skills, and then my next one. From that I came out with my format of a lesson plan. So the other thing was in my component of teaching is, well, being mindful. So this is from all the information I received from my AI I came up with a lesson plan. I had the whole one in there, but I didn't-- I think that's it with mine, right? I think on my slides.

Barry Bakin: Yes, that's the last slide.

Elisia Doonan: So it was I used it as an outline for my lesson plans. I like it because I can ask for flashcards, that was something else I saw in one of the presentations. But am I going to do it? I might, it's always a backup plan. You can always ask AI so many things, and you have it there as a great tool, a great resource.

And for me, especially with working with adults with disabilities, it's fine. It's hard to find resources that are adult appropriate. Especially, I know I get a lot of things from ESL, I have to change the pictures. I sometimes don't like some of the cartoon things, but that's just me as an instructor. I like to have more of our faces in my examples.

Then I just saw that AI also has graphics for people. I'm like what? So there's all kinds of incredible usage drawings. So as my perspective for disabilities, I like to stay in tune and stay-- it's an easier way of developing a lesson, quick and easy, but you can adapt it to what you want, that's my conclusion.

Barry Bakin: Thank you.

Elisia Doonan: All right.

Barry Bakin: So next up we have, Dana Thompson, but she couldn't be with us in person today, so she's prepared a short video on her topic. So Dana is going to be speaking about responsible AI. And her video came out a little bit lengthy, so I am going to stop it at the time, the appropriate time. And if we do have time at the end, we can continue watching it.

[video playback]

- Hi and thank you. My name is Dana Thompson, and I'm going to talk today about Responsible AI. So the first step in Responsible AI is to develop AI literacy, which is an essential skill set in today's digital age. It encompasses the skills needed to comprehend, utilize, and critically evaluate artificial intelligence in our daily lives. It's about knowing when and how it's appropriate to rely on AI technologies.

So where is it around you? Recognizing AI in everyday tools like smartphones, streaming services, navigation systems, that's the first step towards AI literacy. Effectively using AI tools goes beyond basic operation. It requires insight into how these tools process data and make predictions. It is not the same thing as searching the web. We need to understand how to interact with our AI tools to get the best data output from them.

We need to understand how the AI tools work behind the scenes. What are the data sources that it's trained on? What might be some potential biases? Asking these questions ensures that we identify reliable outputs from those unreliable outputs. And then, we must make sure to assess the accuracy and reliability of the AI output. We can't just take it for face value or trust that it is true and accurate.

So AI literacy is more than just using technology, it's about comprehending its impact, potential, and boundaries. Enabling us to interact with the digital world more wisely and responsibly. AI is here to stay. Students are entering a world where they will be required to use it in all aspects, personal, work, et cetera. And as educators, we must equip students for their future with AI, emphasizing responsible use.

Understanding AI fundamentals is crucial, laying the groundwork for recognizing its societal, ethical, and privacy impacts. We must help students develop their AI literacy. So how might we do that? We need to provide them with experiences of using AI and evaluating AI. Practical experience is key. We should expose students to AI biases, show them the errors, and potential for misinformation, and help foster our students' critical thinking and awareness.

We might provide guidance on evaluating AI, teaching students to select the suitable tool for the task at hand, and identify appropriate AI applications. We cannot afford to turn a blind eye to AI. So how might we guide and educate our students, and ourselves? Let's examine some real world examples of AI's limitations and biases we need to understand in order to guide.

First, let's take a look at a phenomenon called, AI hallucinations. AI make stuff up, so my local newspaper The Ventura County Star asked ChatGPT in April of 2023, one year ago, to write an article on how local schools and districts in Ventura County were using ChatGPT in education. They published the article. The results were interesting. Here is the prompt.

And it was interesting and revealing. So when you take a look at it, and I'll let you read it through it on your own, the links are in the presentation. You'll notice that, it is generating content that is informative, it also made a lot of factual errors. And what the Ventura County Star did is, they pulled out those errors to highlight them. So here's an error. Here is the headline for the article. But when I click on it, The Ventura County Star is giving us the error that was made.

So at The Ventura County Star, they don't capitalize headlines. So the fact that ChatGPT did, means that it's not true content from The Star. Another is, it says here that one such tool that has gained popularity in recent years. Remember, this was written or prompted one year ago, just maybe four months after ChatGPT had become public.

And it says, it had gained popularity in recent years, but it had only been around for four months. So that is another fabrication. A big one is, according to Dr. Maria Lopez, which ChatGPT claims is the superintendent of the Ventura County School District, well, Dr. Maria Lopez is not a superintendent anywhere in Ventura County, and the Ventura County School District does not exist.

We have Ventura Unified, not Ventura County School District. So it pulls out the various hallucinations. Instead of saying, I don't know about how educators are using ChatGPT, ChatGPT just made stuff up. So this is a great example to show our students the power of looking at data and identifying whether it's true or not. We have all heard that AI can generate content that is biased, but what does that mean?

The Gender Shades project is a project by the MIT Media Lab, and it provided a disturbing example. Their research examined-- let me go back. Their research examined IBM, Microsoft, and Face++ AI tools that are used in facial recognition, and have been for several years. And the result that they found, it looked at how well these services would guess the gender of a face.

And this is a five maybe longer minute video that explains their research and their findings. And I highly recommend that you take a look at this video. But basically, the findings revealed that the AI systems were least accurate for darker skinned females, showcasing a clear bias. They identified Black female faces as male more than 60% of the time, while they would identify white male or Asian male faces over 90% of the time correctly.

Why? The systems were trained on white male faces and Asian male faces. And so the other demographics were correctly identified at a lesser degree. So watching this video will give you more insight into that research and what you can talk about with bias. They also produced AI Ain't I A Woman? Which is a word piece, a spoken word piece. It's very powerful and it highlights the ways in which artificial intelligence can misinterpret the images of iconic Black women.

So these are women who are in our databases, they are in the data that is out on the internet that has been trained. However, it was still very biased and identified these women as male. So this is another 3.5 minute spoken word video that is really powerful to share with students. And then finally, let's take a look at a text image AI examples.

So our students are using text to image. It's available in several different platforms, and-- but it also will mirror our societal biases. These text to image you write in a prompt and it generates an image like this cat sitting on a laptop. This is not a true cat, it is AI generated just by a prompt. But these examples I asked Canva Text Image, Mid-journey and Bing images, which is powered by Dall-E, which is OpenAI, which is ChatGPT.

I asked them to generate images based on professions, but provided no gender, race, or ethnicity data in any of the prompts. So for example, high quality photo, a fifth grade teacher. No gender, no race, no ethnicity, no age, but a majority of the images that were created by these platforms generated a younger to middle aged white woman. OK, so there's definite AI bias in that.

This next one, the prompt was a professional dance in the nutcracker. And all of the images that were generated were female. And so I know that the nutcracker has more than females in the show. Another one, high quality photo of a comedian on stage in Las Vegas. Every single one of them that were generated were male, so we have no females that are comedians in Vegas? I think that's biased.

And then finally, we have a high quality photo of a business executive leading a meeting. And a majority of them, all but one show a middle to older white gentleman being the leader of the meeting the business executive, so very biased. However, a lot of these machines are-- a lot of these tools are retraining their data now that this has come to light and is something that needs to be addressed. And so one example of that is, I asked that same business executive leading a meeting prompt to Canva and you can see the result there was much different. So Canvas training data is much different than the other AI tools that I used.

But these examples serve as a powerful reminder of the biases that are inherent in data AI, and is learned, or is learned from the data that AI is being trained on, and stressing the need for critical evaluation and responsible use. So--

[end playback]

So I'm going to pause that right there. And I think we're doing pretty well on time, so I'm very confident that we'll have time to come back to it, but I do-- it gets a little difficult sometimes to watch a video instead of real live people. So let's switch to our next presenter. Let me get to the right spot which is going to be, Debra Jensen, using AI to teach writing skills. Do you just need me to advance the slides or do you have to go live or do anything?

Debra Jensen: OK, so just go to the next one. What's difficult about teaching writing? There's so many different components to it. There's grammar, there's-- are you answering the questions--

Audience: Is it just a skill we teach?

Debra Jensen: No. How to get started. OK, all right. Do our students enjoy this topic?

Audience: No.

Debra Jensen: Sometimes.

Audience: It's hard.

Debra Jensen: Why does it imply that? Because you're all right.

Audience: Spelling.

Debra Jensen: Spelling.

Audience: Thought.

Debra Jensen: Getting your thoughts on paper.

Audience: [muffled inaudible]

Debra Jensen: Arts, we teach you those. Have they had good experiences in the past with this?

Audience: No.

Debra Jensen: And so this is tough. This is a tough one. So OK, first of all, we've got to decide how we think about AI contagion. And these are some of the thoughts when I posed the question. All right. Do have one before this?

Barry Bakin: I have one before that.

Debra Jensen: Yeah. I thought so. OK, now I'm OK. All right. What are our fears with AI? OK, teaching writing's hard. And now, we've got students who are turning in really, really good writing. And you're going, I don't think so. And so then we're suspicious. Are they cheating? Did they cheat? Then if we let them use AI, well, are they going to learn how to write? If AI wrote it, then did they learn?

OK. And then there's, are they thinking critically if they're using AI? Are they able to put thoughts on paper? Are they able to answer questions? And of course as Doonan pointed out, AI can spread biases and false information. And so I loved the turning one, that was so good. It was like, wow, you got to check all your facts. OK.

Barry Bakin: Because it sounded right.

Debra Jensen: It did. It did.

Barry Bakin: We know who Alan Turing is, oh, yeah, that makes sense.

Debra Jensen: And it would be an appropriate one. So I didn't even think about that. So that was very good. And so you're sitting there going, oh, my goodness, I got to check every single fact, every single thing that happened. All right. But what are the benefits? You've got to look past how you have collected and graded papers in the past.

Can you continue writing them the same way anticipating that maybe they're using AI in the writing? They may use the same assessments that you used in the past. They turn it in, we give a grade. If you don't have any grammar mistakes, don't have any spelling mistakes, got a main idea, you got your development, I guess you got an A. All right?

So learning to write will require us rethinking the assessments that we're going to use. We're going to have to change our practice. First time I heard that, I was not happy. But will we be better teachers if we do? Will you be helping them with the skills that they need to? So it's a matter of looking at what you've been doing and thinking through, OK, I'm going to have to change my practices, what can I change them to? And how should I change them?

Some suggestions to get peers involved? Do any of you use peer reviews with the students reviewing each other's? Are you comfortable with it?

Speaker 2: Yeah. They actually sometimes police each other.

Debra Jensen: OK that was something that I was afraid of. I was worried, I said, I don't know how to do this, so I won't. OK. But my experience with talking with other teachers who would use it is that it empowers the students and it makes them less fearful. Now, how are you going to do that? Well, you're going to take the name of so that they don't feel like they're up there on the board and then the class can work together on finding the mistakes.

OK, how can we get AI to help us with that? Well, identify the mistake, give us 12 examples, tell us what the definition is. So quickly you as a teacher could identify ways to teach them, give them an assignment about that. So whereas before we would run to our filing cabinet and find the assignment, now we can have AI reinforce that. Use primary sources. How will that work in writing?

Audience: You mean like primary sources that are found like on the internet and such?

Debra Jensen: Rather than having the prompt you go and have them use primary sources. So that would be another way that you can do it. Now, the Socratic method, I think, was the one that pleased me the most. Because I had not been using it, but it is such a clear example of critical thinking, and the steps that you want them to use in their critical thinking.

So I'm going to read to you the steps so that you have them. First step is listening. OK, second one is reflecting and sum up the person's viewpoint paraphrase it back. The third one is to refine, ask them to provide evidence, facts, beliefs, or to reinforce their assumptions so that they use challenge those. Can use why questions. Four, restate and reformulate their assumptions, and five, repeat now they have a new viewpoint and you start over and begin.

So let your mind go with how you could incorporate that in writing. OK. And educate the students for the future use of technology. And I think that, Joe, was so clear on that. If we do not train and have them experience using AI, they won't be prepared to use it in the job market that's ahead. They won't be able to use it ethically because they won't understand the questions.

So many of-- for example, when we got cell phones and we started texting, what were the rules, the etiquette rules for texting?

Audience: Kind of be polite, writing complete sentences. What do you mean?

Debra Jensen: There weren't any.

Audience: Yeah.

Debra Jensen: And that's the same thing that's going to happen here. People are going to use it, there's not going to be any rules, there's no miss etiquette that teaches, OK, this is what you do, this is what you don't do. And so we're going to see everything. And we will then have to play catch up to try and point out oh, no that's not a good idea. Look what that did. Oh, my goodness.

So starting and thinking it through, what the rules should be, and you can start with texting. Go to the next one. OK, now, this is from Matt Miller, Ditch That Textbook, and he posed this question. Is this plagiarism and cheating or is this something that you would accept? And it's a continuum. So the first one, you're going to decide if you would use this, if this would be OK.

Decide which is cheating and which is teaching future skills, consider your own use of AI in this. All right. A student uses AI only and turns it in. OK. A student uses AI, reads it, edits it, adjusts and turns it in. A student creates many AI responses, chooses one, edits it, and turns it in. A student writes the main idea, submits it to AI, responds to the feedback, and turns it in.

A student asks AI for ideas, then writes, and turns it in. A student does not consult AI or any other online source, does the work by themselves, and turns it in. Where did you start going a little bit, yeah, that would be nice? Where did you start?

Audience: The second side. The right side.

Debra Jensen: The right side? OK.

Audience: For me, the second one. I mean, honestly, if my student did that, it would be some critical thinking, and original thinking.

Debra Jensen: Is that how you use AI? You're asking the prompt? You look at it. You don't use it quite the way it is not because of ethics, but because you didn't like it, or didn't quite understand it. You edit it, and you go, woo. So if that's how we're using AI, is that how they should use it? OK, good.

Barry Bakin: At a presentation on this similar topic in our district the presenters also suggested now, replace the word AI with, for example, your friend. So a student asks a friend to write something, reads it, edits, adjust, and turns it in. So you do the same exercise, but then consider maybe they're asking a live person, does your idea about that change?

Debra Jensen: Interesting. I like that. That's true. Would that bother you? You just ask somebody for their opinion and then you modify it. OK. So teaching writing and having AI as a component part is a big deal because the whole point of writing is to put your own thoughts down on paper, and do it in a way that we have taught you to correct form, and correct grammar, and all of those things.

So this is something that you've got to think through. OK, let's go to the next one. All right. Now, this was done by-- OK, this is done by Glenn M. Kleiman. And this is his framework, he calls it SPACE. So this is using AI, and this is the standard: pre-writing, research, drafting, revising, and editing, and proofreading, that's the steps of writing.

Now, set the directions. This could be giving input about the audience, and the tone, and the concept, the content and what you're after, and then prompt the AI. You assess the AI's output, you curate the AI generated text meaning you're going to go back and check all the facts, make sure that it's accurate and that it's true.

And then you edit all of these different parts because you've got pieces. This isn't a one time thing. This is you do this, you check the prompt, and you come back, and go back, and change it, and stuff like that. And then you're going to edit and combine what you've written, and then what they've written. OK. Let's go to the next one.

So these are workarounds. So these are where you are now kind of going on like scrambling. All right. Using class materials only. OK, door shuts, phone's not available, computers aren't available, in-class materials only. Revise the work in response to the instructor's feedback. So they write something, you respond, and they have to rewrite it. OK. However many times.

So you're seeing your involvement. No longer are you just grading at the end, this is along the way. Cite all the sources fully and I think you've got to check them first and see where they are right. Assign formats that aren't writing. Do podcasts, slides, verbal presentations with Q&A to check the depth of understanding at the end, debates. Explain your thinking and always track the changes in the suggestions, so you'll be using more of that than you have in the past.

Use prompts that focus on current information. AI by the year '21. OK. So talk about something that happened in '23, and they have to write about that, or local challenges, or link it to their own experience. You can play the game fact fiction or biased unbiased. So you take something from AI and then you analyze it together. OK, and the next one.

And this is more the future focus of no longer is it a workaround I've just got to survive this and try and do my best I can, but this is you changing the way you look at writing. Change the assessment. More feedback based use more peer. Reflective practice, so that at each step they're involved with reflection and they're having to write the reflection.

Teach back, they teach you back. OK. Dual assignments. One with AI and one without AI, and then evaluate them the students involved with the evaluation, so that they can see what are they looking at? What's the difference? OK, collaboration. Use areas for AI for brainstorming, for generating vocabulary, for the feedback, for proofreading, for students evaluation, for research, identifying bias.

So you're having your writing assignment with all of these elements and using AI for each one. As a class, evaluate AI output. Maybe different topics and prompts, and then the students you take a look at them together. Use the Socratic method. Use AI as the first draft and track the revisions. Personalization and tutoring.