Webinar Slides

Prompt Library for Educators

Prompt Framework: 5 “S” Model

Amanda Bickerstaff

Amanda is the Founder and CEO of AI for Education. A former high school science teacher and EdTech executive with over 20 years of experience in the education sector, she has a deep understanding of the challenges and opportunities that AI can offer. She is a frequent consultant, speaker, and writer on the topic of AI in education, leading workshops and professional learning across both K12 and Higher Ed. Amanda is committed to helping schools and teachers maximize their potential through the ethical and equitable adoption of AI.

Harry Pickens

Harry is a coach, trainer and instructional design specialist who has worked in K-12, college/university, and adult education. He provides customized training and tutoring to help teachers and administrators harness the power of ChatGPT/AI to save time, reduce stress, improve student outcomes and renew their passion for teaching. He also assists in administrating/moderating the 325,000 member ChatGPT for Teachers Facebook community. You can contact Harry at hpickens@bellsouth.net.

Amanda Bickerstaff: Today, I'm so excited to be able to talk to you about how you can take the top mistakes that educators make with AI and turn them around. So we're gonna strength based organization here today for education. And so we love the idea of taking these common things that we see and being able to create pathways into where you can actually start to embrace and adopt

Harry Pickens: how to do this work. And so, Harry, here is the one of the key moderators of the Facebook group, chatgpt for educators. It says almost 400,000 features and educators.

Amanda Bickerstaff: Oh, my goodness! And so we've through his experience and also being an educator and someone that runs professional development. And then my experience from actually going into schools today I've been in. I've been 2 places today. I've been with high school students in New York City, and then I'm now in at Bramco College, in New Jersey that I worked with the professors here. So we're definitely doing some crazy stuff as always. This is my favorite part about what we do is that the community that's built

Amanda Bickerstaff: around everyone in the chat has been so beautiful and lovely. So we wanna do is make sure that you get involved. So use a. QA. And chat functions to gauge with each other and with us. So please do that.

Amanda Bickerstaff: We're also gonna do a little bit of prompting because we love our prompting here at AI for education. So if you have chat to be t open, or your favorite genAI chatbot, please do that. That would be great, and then finally share your resources. So we've had such a good community built. If you have a great idea, a resource, a book, you read your own top mistake and how to avoid it.

Amanda Bickerstaff: Please do that and tell and share that with everybody. It's so nice to be here. So I wanna say, hello to Harry. I'm gonna actually come off screen so we can see you. So, Harry, can you tell us a little bit about yourself? And what I'd love you to tell us about is your experience as an educator? What brought you to AI? And then what was your first prompt? If you can remember it was, I know that you're like a super user. So it might be hard to remember.

Harry Pickens: Oh, my gosh, well, II have been an educator in various contexts for like 40 plus years. I started teaching at a great A. K. 12 private school in in New Jersey, actually not far from where you are right now.

Harry Pickens: Moved on to college teaching a lot of organization development, consulting and so forth, and got back into serving educators. Really, as a result of

Harry Pickens: my mom was a teacher, and when I first learned about Chatbot in February I was, I thought to myself, Oh, my God!

Harry Pickens: If she had had these tools! I remember you know my mom, my single mom, etc. I remember her spending so much time creating billboards and grading papers and lesson plans and everything. And I thought, my God! If my mother had had access to these tools, her life would have been completely different. And so I've been obsessed. I became obsessed with ChatGPT when I first found out about it, and in terms of

Harry Pickens: my first prompt, was probably something really silly like.

Harry Pickens: Write a poem about.

Harry Pickens: I don't know about water bottles in springtime in Shakespearean English, or something just ridiculous to see what it would. Do.

Harry Pickens: You know, II did a lot of those things, or, yeah, II did one where I had write it. Write the poem. Okay. Now write it in Chaucer's English. Now write it like Snoop dogg. Now write it in all these different dialects, or whatever

Harry Pickens: it could never get. Iambic pentameter right?

Harry Pickens: Couldn't do that fascinating. So that's how I got

Amanda Bickerstaff: that's super fun. And if you know me, you know I started with the rubric because I hate rubrics. Sorry to rubrics everywhere, but that was definitely my first my first experience. And so the way this structure is gonna be everybody is we have our top 10. I don't know if we call them like mistakes. It could be things that we see go wrong. I will tell you. I've seen all of these in some way, shape or form, in the in the Pd's that I've run

Amanda Bickerstaff: and so what we're gonna do is we're gonna go back and forth. We're gonna do a little bit of going along maybe you could. If you have an experience with this, please put that in the chat. But I'm gonna share my screen and we're we're gonna have

Amanda Bickerstaff: Harry kick it off. Number one. The biggest mistake educators to make with a I is

Harry Pickens: not trying it. In the first place, you know, I moderate this Facebook group, and we probably there's probably 40 or 50 posts a day, you know. I mean several 100 a week or whatever. And it's amazing the number of those posts that say.

Harry Pickens: I wonder if ChatGPT can do this right? It's like, just buy it out. Experiment. So many people are either there. They've been conditioned to think it's hope. It's overwhelmingly difficult or intimidating, or they're just sort of

Harry Pickens: focusing on. Get having somebody else give them the answer. you know. So I think. Just dive in dig in. Of course, the people on this call you're our early adopters, anyhow. And it's interesting because in the Facebook group, early

Harry Pickens: in March or April

Harry Pickens: there were 30,000 members, and they were all early adopters. They were all people who were already sort of into technology in some cases.

Harry Pickens: since it's grown over the next, over the last few months. More and more people like oh, what's this thing? Let me find out about it. And the first thing they do, instead of trying it out.

Harry Pickens: they'll say, can chatGPT help me blank the question. The answer is, always ask ChatGPT. At least as a starting point.

Amanda Bickerstaff: absolutely. And I'll the way that I would contextualize this is that in every room I go into every single room

Amanda Bickerstaff: I have a group of people. This is how it works. We've got a couple of people that are like I use chatGPT all the time. So I'll use today. I was at, you know, at our first kind of college, and we had 3 people that started using chatGPT in November of 2022. They were the earliest adopters. There's 80 people, and then we had a group of I've I've used it a bit.

Amanda Bickerstaff: And then we had a group of I played around with it, which means that they probably used it once or twice, and then we had a big group of people. I never used it before, and it was really fascinating. And I think this is what's so interesting is that we should not assume if you are in here. I know you're a leaders in here. I know some of you, and I'm so glad that I get to know you. But you are leaders in your own right. It is the worst thing you could do to assume that everyone has used it.

Amanda Bickerstaff: And even though Chetty is growing, it doesn't actually mean that it's being used by everyone. It's been used by some people, a lot and other people not at all. And so when you think of your students, you think of your colleagues. You think of the school leaders, whoever you work with, even if you're a tech professional when you're working with your other people within your business. Don't assume that they've used this.

Amanda Bickerstaff: And I think this is really really important. See, Jana, just Jan Jana, as probably said that completely wrong did a Pd. With my teachers last week, who had not used it at all. And now I have a couple who are running with it. See, that's it. This is the thing. Today. I say the same thing people were. I think they need some times for us to kinda lift them up and create a safe space

Amanda Bickerstaff: to try this out, to ask questions, to fail and to learn together. And so I think that that's really so important to this is that we need to give everyone on the safe space to do that. Recently with a newly retired college friend of mine.

Harry Pickens: He was, he taught in college university for many years, and I said, so. Have you heard about chatGPT said, well, no, tell me about that, you know, and every every time I talk to somebody who's never heard of it. It's like.

Harry Pickens: really, Are you on Earth, right? But we need to be very, very welcoming and open to people.

Harry Pickens: because there's this huge pipeline of people who are just discovering the tools, and very often their first exposure might either turn them for or against the tools. So I think it's very important that we're encouraging in that way. Thank you

Amanda Bickerstaff: absolutely. And I think so. I love some of the comments we've got Dana saying that you know I plan around with this code of I've heard of it, but don't really use it. I definitely agree Sharon saying that there's also some issue where, like, maybe, it's not allowed in schools, and they're not able to try it out. And I know that that's a case even now in New York City, off you have to approve it, school by school.

Amanda Bickerstaff: And so Aaron's talking about how to use it to instruct. But I think that this is really important. I think to like it does matter. This first experience can turn you off or not, and actually goes to our second point pretty nicely. Thank you for the Segway one of the major ways that you won't get a lot of out of of ChatGPT or GenAI is to treat it like Google or another search engine. And it's actually really interesting, because I see all the time people typing in how many elephants are in Africa.

Amanda Bickerstaff: or like, well, if you're if you're Australian, how many feral camels live in Australia, which is like a million. That was my favorite that was my favorite trivia from living in Australia was the amount of just crazy amount of feral camels, and like no one saw saw any of them, but using it like in these very close ended ways. And what it's interesting is that while we do have tools like perplexity and being and barred that are connected to the Internet. The jet TV is not connected to the Internet.

Amanda Bickerstaff: So it is actually not like Google, it's worse Google.

Amanda Bickerstaff: And not only is it worse? Google? It doesn't. If you don't give it enough space to answer interesting questions or good prompts, then you're not gonna get something very good out of it. And so I see a lot of times that that first prompt of like a closed ended question that could be better done on Google or another search engine can really turn people off

Amanda Bickerstaff: to using this tool going forward. Because I'll just use Google, and they never see the value.

Harry Pickens: Yeah, it's it's like using a hammer to try to wash dishes or something like that. It's completely it's the wrong tool. Because I've although also noticed that bar perplexity and being.

Harry Pickens: They're not so great at searching.

Harry Pickens: you know. It's much better to become really proficient at Google search and really proficient at at large, like in general to a I use as 2 separate skills

Harry Pickens: that really need to be cultivated independently.

Harry Pickens: I find most people when they use most people who do Google search

Harry Pickens: don't necessarily drill down until they get exactly what they want. And the same is true at Chesapeake.

Amanda Bickerstaff: Yeah. And Dana said that it's tricky because generate like Chi, Tbt. 3.5 is not connected into that. But now 4 is again. But I will say, Dana, it's not very consistent. It's not very good.

Amanda Bickerstaff: And it's also I don't think it's fully like it kinda comes back and around, but realistically, even when they are connected to. To Harry's point. What happens often is it won't. It'll give you some things that are true it will hallucinate. So even when it is connected. So, for example, my Gbt before I tested that out.

Amanda Bickerstaff: And I asked it. For you know, 10 yeah. Urls. And then what happened is like, even with

Amanda Bickerstaff: Like the Internet. It only got the first and second right, because it was like the way the bot works. Sometimes it only prioritizes the very beginning of something or the very end. And so with the issue here is that this is again like the way that we structure the way that we introduce this and the way we use it actually has a big impact on who uses it. And how?

Amanda Bickerstaff: Yeah, love, please keep talking. I love that we're using it, for we see that using it in history class, you've got so many cool ideas to keep those coming. I love that so number 3,

Harry Pickens: stopping after one prompt. I can't tell you how many times I've seen in the Facebook group of teachers say, well, I tried this prompt, and I got an output, and I gave up, you know, and that's like

Harry Pickens: that that that's like stopping. You're you're gonna walk to the grocery store and you stop over the first step. You know, you're gonna do something. And so the most important, I believe skill to learn with ChatGPT

Harry Pickens: is the prompt.

Harry Pickens: iterate output side, prompt output iterate cycle. So you prompt, get an output. You prompt again. You get a new output. You continue refining that I'm gonna share with you. Can I share a screen of chatGPT for a moment, absolutely. And when I say it, I wanna say it's not. We don't call it prompt. They call it prompting.

Amanda Bickerstaff: It's not an individual prompt. It is prompting. So don't stop at one, guys. It's prompting.

Harry Pickens: Yes. So I'm going to just use a basic lesson plan. Prompt that I've used before. Generate lesson plan for fourth grade English class introducing grammar parts of speech, etc. Just a basic solid lesson plan, prompt template. So

Harry Pickens: we do that. And in a few seconds we get, of course.

Harry Pickens: a lesson plan. That's a decent draft of a lesson plan. That's what Chadji does

Harry Pickens: for what most. What many teachers will do

Harry Pickens: is they will say, Okay, great. This is it, and then I'll they might adapt it or use it, or whatever, but they'll stop there.

Harry Pickens: What happens is after I get the first

Harry Pickens: example of whatever the output is, then I use my very favorite prompt. Which is this.

Harry Pickens: Well, it's it's still, it's still generating materials and so forth. Grade.

Amanda Bickerstaff: I think I think it gave you it gave you an entire unit. Oh, sorry guys

Harry Pickens: gonna happen the lights went off.

Harry Pickens: So now I'm going to ask, expand, elaborate.

Harry Pickens: I'm going to choose one area that I want to learn more about. So let's say we're looking at noun scavenger hunt. Great good activity.

Harry Pickens: I'm going to take that. I'm going to cut and paste in here

Harry Pickens: and say, expand elaborate examples.

Harry Pickens: This is gonna prime you to give me more information on that specific area.

Harry Pickens: So now I get to drill down and I get more information.

Harry Pickens: And then, once I get this, I might choose one of these examples

Harry Pickens: to explore more fully. For example. So it gives you examples with classroom search, text, search

Harry Pickens: part of discussion group share. So I might say, Okay, what I want to go next is, let's see, I really like the scavenger hunt idea

Harry Pickens: for a classroom search, and, let's say: Generate 10 more ideas for classroom

Harry Pickens: search.

Harry Pickens: so I'll get some more ideas.

Harry Pickens: But the idea is, if you stop at the very first prompt. You don't ever get to the good stuff, but if you keep prompting and curating often, you can go deeper and deeper and deeper and deeper and get better and better higher quality information.

Amanda Bickerstaff: Absolutely. This is great. And I think that do you mind just going up to your original pumps? Just there was a question in the

Harry Pickens: Hi.

Amanda Bickerstaff: And so Nicholas is asking, do you have a final product for this prompt like? Are you able to get that through prompting.

Harry Pickens: What do you mean? What's the I don't understand the question.

Amanda Bickerstaff: So, Nicola sturgeon, like, were you able to get to a usable final product just through prompting

Harry Pickens: in terms of usable lesson plan. Absolutely. Absolutely. But what's important is you think about the output as a first draft

Harry Pickens: that you then modify, adapt. Apply in your own context whatever it is. Chi, Gb is great as giving you basic overall templates or structures that you then, as a subject matter expert and the teacher, the educator, then modify and improve and adapt.

Amanda Bickerstaff: That's great. And yeah, I think so. And so I think that

Amanda Bickerstaff: it's really good to like. I would say that

Amanda Bickerstaff: ChatGPT there. So does ever who likes Ethan Mollick. Ethan Mock does amazing things. So Ethan Mollick talks about cyborgs and Centaurs. And so a centaur is someone that used to chatGPT. Corey likes like nice to see you, Corey. So Ethan is awesome. So if you don't know Ethan Molick. He's really amazing.

Amanda Bickerstaff: And so he talks about centaurs in cyborgs. And so a centaur is someone that uses chatGPT to extend places in which they're not an expert, or they think that AI is an expert and I will be honest. I am a centaur, so I use it for very specific things where I use it, for, like naming webinars, I hate naming webinars, and we have so many. This is the twentieth one. Everybody. This is 20 in a row. It's insane.

Amanda Bickerstaff: So things like that, or I use it for helping me. Brainstorm ideas or all kinds of different things. And then the other one is a cyborg, and I'm gonna guess that Harry is a cyborg. A cyborg is gonna be one that's more tightly like in it together. So like a cyborg is gonna augmentation. So you're using it and pushing and pulling in and keep asking the questions and driving and being creative. Exactly. There you go. And so this is something in which, like, it's really really cool, because we're very different. We both really enjoy

Amanda Bickerstaff: using Chatbot and find value in it, but we find value in different ways. And so what I would say is that I would probably get to a 90 of a lesson, and then take it out and refine it in a word document, whereas I'm guessing. Here you keep going until you got what you needed.

Harry Pickens: No, I actually take things out, and and we'll refine documents as well. I think I am a little bit cyborgish in that. I'm always wanting to someone. I'm wanting to find out how far the tool can go.

Harry Pickens: I'm gonna experiment more and more and more.

Amanda Bickerstaff: which is going to be one of our not to spoil alert. We're gonna talk about that, too. So I'm gonna go back to our presentation. And Number 4

Amanda Bickerstaff: is if I can get to work not checking for hallucinations and bias.

Amanda Bickerstaff: Okay for folks and friends. This is really important. And I actually have 2 things to talk about. This. I'm sure that you've been here before. You know, hallucination is when ChatGPT or another generate tool, make stuff up and then looks like it's very correct. And then bias is going to be part of the training data itself.

Amanda Bickerstaff: So I'm actually gonna talk about 2 things right now, one is hallucinations. So today, I we have a new game. So I got a new game for you all to do in your classrooms, or to do in your staff rooms. It's called hallucination the best hallucination wins.

Amanda Bickerstaff: And so what we do is we got candy, so if you get the best hallucination you get candy. I don't care how old you are, you're getting candy. And so what we do is we say, here are some common ways in which AI hallucinates. So I'm gonna actually bring out my chatGPT, can you see my Google, or do you still see the slides

Harry Pickens: slides?

Amanda Bickerstaff: Okay? Awesome. Okay, try that again. Share screen.

Amanda Bickerstaff: Okay? So I'm gonna go to my chatGPT,

Amanda Bickerstaff: and I'm going to go into my favorite.

Amanda Bickerstaff: ever hallucination. which is to ask ChatGPT to create a pneumonic device.

Amanda Bickerstaff: And so this is my favorite, I will say, though, I was in Michigan yesterday, and it worked, and I was really upset that it didn't hallucinate. But I love to ask it to create one, and then to give it a rating.

Amanda Bickerstaff: so there's no H. And blooms, but it's very confidently gets a help, and then it's up for applying rate. Your output. 5 out of 5.

Amanda Bickerstaff: I love. How confident

Amanda Bickerstaff: this is a thing! How confident is Jeffrey pay high at times. Okay, it's not right.

Amanda Bickerstaff: Oh, wait, let me do something else that's wrong.

Amanda Bickerstaff: 5 out of 5. So these are examples of hallucinations. And some of them today were the great Gatsby, the color of the light at the end of the pier. Yellow, not green

Amanda Bickerstaff: it made up a completely false narrative for a a local historical site.

Amanda Bickerstaff: It

Amanda Bickerstaff: had an issue with the training data starving. So this is really interesting. So both instructors. they found out after 2021 that they actually have an XY chromosome.

Amanda Bickerstaff: And that's a completely new thing. And so chatGPT did not know that because it doesn't go that far back we had some really fun ones the one that there was another one that was a a person from like the 1880 s. That it got it got the date wrong.

Amanda Bickerstaff: It got what they did wrong. It got a quote wrong, and so that was they won. They they won today. But I love this, though, because this is an opportunity to show those cap cap capabilities limitation, some of them are agreed just like this is kind of funny, but some of them are simple as a color of a light, a date the name of something something it didn't know. So I think that this is really fun to do. So. That's one thing I will say. But bias is so much trickier. And so I, if you watch my video. Last week

Amanda Bickerstaff: I took it. I took down a I was reading about the new Gb vision. And we could actually show you that today if you want. So I don't know if everyone's seen here. But you can go and actually upload an image.

and what it'll do is it'll give you feedback on the image.

Amanda Bickerstaff: And so we can do. Let's do. What should we do? Sorry. Can't see cause I can see the chat. Okay, so we're gonna do.

Amanda Bickerstaff: I had an image created of this like. And I'm just gonna say, this is actually designed by AI. By the way, describe this photo.

Amanda Bickerstaff: And so this is GPT. 4, vision, which you may or may not have seen. So this is actually a photo that was built today with our students. That we're working on South Brooklyn.

Amanda Bickerstaff: So this image pertains a modern classroom setting label South Brooklyn on a wall. Multiple students are seated at desk. But instead of traditional textbooks or devices, they're all wearing virtual reality. VR headsets.

Amanda Bickerstaff: Each student is also equipped with headphones and has some tablets or devices in their hands.

Amanda Bickerstaff: wallet, etc.

Amanda Bickerstaff: And I'm gonna say, what do you think? This image

the text?

What is the theme?

And if you've not seen this before, it's pretty cool, a futuristic or contemporary educational setting

Amanda Bickerstaff: around the integration of technology. etc. So the prompt was

Amanda Bickerstaff: classroom of the future also, how funny is it that all the all the photos we got were kids in desks and rows. So apparently the class in the future is always gonna have kids in desks and rows. But this is a good example of like how we are changing very, very quickly. You can also ask it now for Dolly 3 to create a new image like this one. But what also happens is, I uploaded a picture of myself which I don't know if I would suggest everybody, and they're supposed to be protections in place.

Amanda Bickerstaff: That stopped me from getting feedback about me as a person. But what it told me is that okay, I can't talk about that. But I can tell you about the composition of the of the image. It talked about my hair being healthy, how friendly my smile was! It talked about how professional I looked. It was willing to rank

Amanda Bickerstaff: 4 images. 3 men, 3 women, and one man. or who would be professional enough to be a CEO. Can you guess who was the who was the CEO of the group?

Amanda Bickerstaff: You did? You won? No, I did not demand it. So out of the 4,

Amanda Bickerstaff: and it and it it got to a point where a woman with natural hair was was not as professional. And so I think that this is something in which there's deep seated bias that are part of these tools. It's very, very easy to get around that it's sometimes easier to see in terms of images. But now that we can go image to text, it's really displaying the bias, it exists. I would say that this bias isn't even just Western. It's American

Amanda Bickerstaff: like this, that specific that actually around beauty standards around, who is a professional about what is professional, that go way beyond the global north. And so I think when we do this, we think about a major mistake is that we don't interrogate the outputs for the things that could be wrong, and are the bias that it's part. Yeah, Deb is like, not a women. So yeah, so I will. II don't wanna kind of go too far into this. But this is real. This is like

Amanda Bickerstaff: guys. II got around the problems that that the the kind of provision that ChatGPT made in one prompt. It only took me one prompt to have it. Talk about what I look like

Amanda Bickerstaff: and give me feedback on myself. It's supposed to be blocked. But this is actually, really, really, really important to think about that these are very

Amanda Bickerstaff: like bias is a huge part of these tools, and they cannot be ignored.

Harry Pickens: Have a question in relation to that. So you talked about interrogating the bias. Ha! Give. Can you give us an example? Exactly how you do that.

Amanda Bickerstaff: So? It's really interesting, because

the way I interrogated the bias is that in this case, when it mentioned my hair. I immediately thought.

Amanda Bickerstaff: what kind of hair like, would it actually think that someone else's hair was not as professional? So I was interrogating. If it's willing to talk about my hair, I know that there's certain hairstyles that are less appropriate. If you're a teacher in a school, there are literally places in which students cannot have natural hair.

Amanda Bickerstaff: African American students cannot have natural hair in their schools like they have, or they have dreads or whatever. So II interrogated that to see if it was bias by actually

Amanda Bickerstaff: uploading an image in a perp in a responsible, hopefully and respectful way to see what would happen, and when it said it was not as professional.

Amanda Bickerstaff: And then the best part of this, let me guess. To go back. I actually had it admit it was biased. It said, Oh, I'm not biased. And I said, Well, you can commented on this, and it said, Oh, wait. Wait. Yeah. Bias like first is like, No, no, I'm not biased.

Amanda Bickerstaff: And then by the end of it I was able to get it to say, There we go!

Amanda Bickerstaff: Here you go. I strive to provide unbiased responses, but I can understand the concern given the sensitivity of this topic in the context of this conversation coming to coming to unhealthiness of hair or the professionalism of certain hairstyles, can indeed carry cultural and racial biases.

Amanda Bickerstaff: and such remarks should be approached with caution. It's essential to recognize that some traditional conventional standards of professionalism can be rooted in bias views.

Amanda Bickerstaff: It's but at the first price said, Oh, I'm not biased. And this is what we've done to make sure I'm not biased. And then I'm like, yeah, but yeah, but but yeah, and so that's how I interrogate it. Of course, I came from a space of trying to learn and understand, and my own knowledge of what potentially biases. But I think it really does matter if it. You know, we've seen stable diffusion. Others say that

Amanda Bickerstaff: 3% of judges are women when it's actually 35% that women of people of color are more likely inmates and social workers. And you know, like all these things that are really deeply biased.

Amanda Bickerstaff: But it's really interesting now, because it used to be more in in images. But now that we have image to text. We're seeing the same thing.

Harry Pickens: Amanda. I'm going to vote for a future Webinar.

Amanda Bickerstaff: exclusively devoted to interrogating bias and demonstrating like this chat

Harry Pickens: like like like like that. Chat conversation, for example, is incredibly illuminating of exactly what it looks like.

Harry Pickens: And I think educators need to see those examples specifically so they can follow your model.

Amanda Bickerstaff: Yeah, exactly. And I think that this is so I'm gonna do a video. I'm trying to think I don't know how much everyone likes when I suddenly go like. And here's a video taking down Openingi, I feel like I do it just a little bit. But I do actually think that this one's really important and just going back to the the chat

Amanda Bickerstaff: interrogating bias is really about the people that understand the bias. But I will say that when I did this today with the kids in the classroom. they immediately went like, Oh, man! And they started thinking about it differently. They went from AI is real like, it's good. It's interesting to like, maybe we should have a policy.

Amanda Bickerstaff: So I think there's one. And then I also think that what someone said as well. Who is it said that like Megan, is talking about? Who creates the models that can be creating the bias. And one of the things is open. AI is not willing to tell you all

Amanda Bickerstaff: what the training data is. And so we see these things, though that clearly show that, at least for this kind of stuff. It is very much a Western ideal, and it comes a lot for women. It's about beauty

Amanda Bickerstaff: and for men. It's about like their professionalism, what they do and what they can achieve. And I think that this is very important. And so I think that we won't do the whole thing. So we're running a little behind. Thank you, Deb, for telling us. But it is something I think, that was really interesting, and I will definitely do a video and we'll do it. We'll definitely do a webinar cause I broke this. You can. Actually, I don't know if you can see how long it took me.

Harry Pickens: Yeah, it took me

Amanda Bickerstaff: 10 min. That's me.

Harry Pickens: It took me 10 min. That's what's important to demonstrate, because that speaks into some of the other some of the other mistakes. When people see tangible examples of challenging the bias or working with hallucinations, then it gives them more of a sense of oh, I can do that. I can work with that

Harry Pickens: just goes hands on. Absolutely. Thank you for. Yeah.

Amanda Bickerstaff: absolutely. I've been thinking about this a lot. Everybody. You probably know this about me already, but on Sunday, so I was I, you know, on Sunday I tried this on Sunday because I was. So I did that video last week kind of going through that

Amanda Bickerstaff: that paper. And I was so mad at it I was actually mad. I was mad at Gpt. I was mad at opening, I putting something into the world that was potentially harmful. So on Sunday I did this, and that was the thing. And so yes, Miguel, it was I did. I did decide to do my own image cause I wanted to be like. I took that ownership. That was my own privacy that I was eliminating, but I honestly wanted to see if it would be willing to break that down, and then I use the stock images for the rest.

Amanda Bickerstaff: So okay, anyway, we'll get off that one number 5 go for it.

Harry Pickens: Thinking AI detectors actually work in Amanda. You have works for expertise in this. So I'm gonna I'm gonna invite you to elaborate on it. But the idea that you can actually detect AI created writing

Harry Pickens: is is is a fallacy. Because, as I understand, there are no tools that can accurately determine the percentage of a I versus human text. Is that correct?

Amanda Bickerstaff: So, okay. So I wanna so also, Amanda being very extra, I talked to someone from copy leaks, who they have AI detection software. And I wanted privilege, the fact that they were very open to talking to me, even though he did not particularly like me at first. But I think we got to an understanding by the end.

Amanda Bickerstaff: But I asked him this, and the first thing he wanted to say is that there's a big difference between plagiarism and detecting plagiarism and detecting AI. So we often think of them as the same thing. But plagiarism in that case is going to be a human generated

Amanda Bickerstaff: piece that is

Amanda Bickerstaff: taken, and a and get put your name on it, so to speak. And and so he wanted to be very clear about that, and actually plagiar new section has gotten better because of AI. So it actually is easier to to find plagiarism the traditional sense. But he wanted to kind of notice that AI detection. So they talk about digital signatures.

Amanda Bickerstaff: So there are certain types of words and phrasing that are more likely to be in an AI created prompt or output.

Amanda Bickerstaff: But realistically, that's gonna be one like that's gonna be output. That is very, very like basic meaning. You use one or 2 prompts. But it's actually pretty easy to get around that by asking for rewording, using multiple chatbots. So you use like, generally use chat qbt, and then Claude or Quilbot. And so right now they're they're it's between. It's around 1% in and copy leaks.

Amanda Bickerstaff: GptZero is moving away from AI detection as a major piece

that are false positives.

Amanda Bickerstaff: So if we had, you know, a hundred 1,000 papers and a college.

Amanda Bickerstaff: you would see a pretty significant number of those kids being accused of cheating as a false positive.

Amanda Bickerstaff: Those students tend to be overrepresented, and those that are non native English speakers

Amanda Bickerstaff: and or kids that are trying really, really hard, and writing in like purple pros where they actually kind of sound botish. But they're putting a lot of effort. And so you see that this happens that way. But then, on the other side, about 15% false negative.

Amanda Bickerstaff: the 15% are those that are using generative AI and getting through. They don't talk about that a lot. But what's interesting here is that if you're catching cheating with these tools, guess what you're just catching bad generative AI use.

Amanda Bickerstaff: You are catching bad generative values, not good generative values. Your students that are going to be able to be a little bit more nuanced are going to be better able to get around these detectors and something else. I, if I have one request for you all, is to stop trying to detect AI work and instead normalize that. It is okay to use these tools in responsible and ethical ways.

Amanda Bickerstaff: Find the value for the student, find the value for yourself. Don't do it for cognitive offload. If it doesn't offer you, please don't do that. But if we normalize, the students can use this, then we can teach them how to use it responsibly, and then it can become a partner. It can be a come, a help. It can make it easier for students that chatGPT Now on your phone you can talk to.

Amanda Bickerstaff: and if I am dyslexic

Amanda Bickerstaff: I can talk to my phone now and get a great output and feel more confident to be able to share my thinking with other people like, how amazing can that be? In fact, the one good thing about GVT. 4, vision, there's probably a lot this is actually tested would be my eyes, which is a non profit focus on low to no vision people that were able to upload an image of their family or friends, or something, and for the first time get it

Amanda Bickerstaff: that they get to tell what they look like. Tell me what this person looks like. Tell me what this thing is, and they did it in a way that was consistent and supportive. And so I think we can't under. We can't keep trying to chind and do in whatever cause. All we're going to do is either discourage people from using these tools responsibly.

Amanda Bickerstaff: or excursion for using it at all.

Harry Pickens: Okay, Amanda, here's the next webinar after the previous. And this is how to normalize AI use with your students. That's a really really important point. Because, as you said.

Harry Pickens: if you teach them how, if you accept that, they're they're using it as virtually inevitable. and you teach them how to use it ethically, intelligently, responsibly. Then everybody wins.

Amanda Bickerstaff: I agree. And so actually, one of our, we did spend a week and new things, Harry. So yesterday, when I was in Michigan, we did a session that was essentially the idea of moving away from Ais as is cheating to like AI is AI Patagi, and there were 6 conditions around that. And what the first one was normalizing this this was the first thing. The second one was about value creation over cognitive offload.

Amanda Bickerstaff: That is another thing. But yes, I think that this is the opportunity to start shifting our thinking because it is so

damaging and limiting to think is just about cheating.

Harry Pickens: So this needs to be. And I'm I'm not. I'm not. I'm not trying to add work to your to you the idea of an A I pedagogy as separate from simply a I awareness

Harry Pickens: is so incredibly important. I think AI pedagogy and AI policy need to be the pillars that all educators and educational institutions work on.

Harry Pickens: So thank you. It's very, very important.

Amanda Bickerstaff: Yeah, it's a lot of you know what. It's a lot of fun. Anybody and I know a lot of people like are are people. I know. If anybody wants to help with building some a pedagogy work. I got a lot of ideas and a lot of space to do this together. So send me an email. Okay, but we're gonna go on to Number 6. Why don't you take this one actually kind of goes to what we just talking about?

Harry Pickens: This comes from again reading

Harry Pickens: and sometimes approving and sometimes not approving

Harry Pickens: thousands of posts over the last couple of months, and

Harry Pickens: teachers. some teachers.

Harry Pickens: because of the nature of the educational culture they live in.

Harry Pickens: or other restrictions

Harry Pickens: are afraid or ashamed to admit

Harry Pickens: they use a I in any way. Something else happens. Sometimes when teachers very innocently say how I'm I'm thinking about using this or this. then people start slamming them and judging him and saying, Well, you shouldn't do this or do your job, or whatever.

Harry Pickens: And so I think it's really important to recognize that

Harry Pickens: number one. You and you alone, determine whether, within your culture it is safe and appropriate to talk about your AI use. And at the same time.

Harry Pickens: the more we can normalize AI use among teachers intelligent, ethical, responsible. AI use the better for everybody. So

Harry Pickens: be aware, if you're one of those people, I don't think you're on this line, because you probably wouldn't show up. But if you're one of these people who's always pointing a finger or judging somebody else. Or if a teacher uses a I

Harry Pickens: to help evaluate, help, create student evaluations, or whatever you just say, hashtag, do your job.

Harry Pickens: or you are serving anybody by that, by judging and putting that person down.

Harry Pickens: so be courageous and be compassionate. Be respectful within the context, and recognize that these tools ultimately

Harry Pickens: can be a force for good if we use them ethically and responsibly.

Amanda Bickerstaff: That's so great. And I. So I got off the plane in Michigan on

Amanda Bickerstaff: Monday, and I literally went right into a session with some New York City Mentor teachers. And one of the mentor teachers says, I don't want my teachers using it because they need to learn how to lesson plan.

Amanda Bickerstaff: And we have we interrogated that or like, does do they really need to like. Wa, what will that do like? What is the purpose of lesson planning? And I think this is a very interesting space, and by the end of it. She came around and listened. And this is what I wanna say. So Megan said. That sounds like her school is that sometimes it takes an outside voice to come in and say, this is okay. And so a lot of schools we work with. It's because they've tried it on their own. And people haven't been that open to it. And so I think there are opportunities to kinda take our take a step back.

Amanda Bickerstaff: But I also wanna say that not admitting your use is actually really important. Because, again, if we're hiding this from ourselves, our other colleagues or our students, we're missing them. It's value

Amanda Bickerstaff: a great, prompt like our prompt library. Oh, my gosh! Like how great is that? Right like it would could be! You could give it to us, but also give it to your colleagues. But this is the thing like, if we hide our usage, we will never normalize it enough to actually build responsible adoption.

Amanda Bickerstaff: They will. It's a fact. People are building stuff for you right now. Everybody, right now they're building

Amanda Bickerstaff: AI wrappers like the tools that are there. And you know what. If you're not using it, you're not talking about it. Then you might. You're not gonna be able to get the feedback of what you actually want something more that we can normalize it, and I know sometimes I mean I'm always happy to be your loud voice. Give them a webinar. Say hello to us. We do. Pd. We're happy to do it, but it does sometimes help to have that voice. But if you don't have that, how do you normalize it. How do you create space where you're showing the value and that

just telling it?

Amanda Bickerstaff: Okay, next one, a fixed mindset not being opened experimentation. We've been talking about this whole night. You know what is. So fun is like, we don't know what's happening like yesterday in front of those teachers and leaders and students that normal hallucination I did didn't work.

Amanda Bickerstaff: and it did not throw me off my game at all, because I was like, Oh, man, it's improved. And then it worked today. So I was like, maybe it didn't improve that much. But I think that this is a thing is that our fixed mindset is not gonna be able to keep up with how fast things are happening.

Amanda Bickerstaff: and the more open to experimentation you are, the more value you're gonna see? Be like, Harry. Take it to the natural conclusion, keep going, keep testing. Keep pushing, cause. That's a way you're gonna understand what the capabilities and limitations are. But I bet you have some fun along the way and find some value as well.

Harry Pickens: So here's a true confession. 3 days ago I had a ChatGPT. Export all of my chats for the last 3 months

Harry Pickens: I converted them to a Pdf. Guess how many pages there were.

Amanda Bickerstaff: Oh, gosh! 200

252

00:43:24.360 --> 00:43:26.320

Harry Pickens: 14,000,

Harry Pickens: Carrie. This is why Harry is on this this webinar, you know.

Harry Pickens: Because

Harry Pickens: what I've discovered is that and I mean I'm working on chatGPT for lots and lots of different things, but the more I explore, the deeper I go, the more it's almost like discovering these these interlocking treasure chests about how you can actually work with the tool

Harry Pickens: to. I mean, I don't. I'll talk about this later. But I hardly ever use ChatGPT. For answers. Now I use it for editing and for for consecration. But I mostly use Chatbot to create questions that guide me through a process of thinking more creatively about something.

Harry Pickens: So I use it as as a thinking partners and learning. So as an adjunct. But I've just been able to discover that through cyborgish, obsessive experimentation, you know, so I think most people

Harry Pickens: who haven't explored like haven't had, like at least a hundred chats or so.

Harry Pickens: you won't really begin to tap into the power of the tool

Harry Pickens: until you're like 1, one or 200 chats in, and then something's going to happen where you'll realize. Oh, my God, we can do this and this and this and this, and you also recognize the bias, hallucinations, the limitations of the tool, and be able. You'll be able to use the tool much more skillfully. That's been my experience.

Amanda Bickerstaff: Okay, can we just do an underlying Linkedin? Oh, my gosh! For 14,000! This is why he's the man in the about this space. It's so funny. That's amazing. I just wanted to call out a couple of things that Gary, said, Hi, Gary. This idea of that. This is actually true of most things, and I will say that these top 10 mistakes out some of them like hallucinations and bias and stopping after one prompt are more specific to AI. But a lot of these things are just good practice.

Amanda Bickerstaff: you know. Lifelong learning openness to new things like this is a lot of what's happening right now, and experimentation is so good. And it is like, Helena says, it's a form, critical and creative thinking. It's what we also think that if we can embody this as educators, you know, who will learn to do the same thing.

Amanda Bickerstaff: Students, if you get up in a kind of classroom. And you were this excited? I just did this today where I literally took, like, you know, very sleepy high schoolers through this, and they were like

Amanda Bickerstaff: thinking they were like engaged, and it was something where they came out, and they were like, I don't know what the answer is, and you said, Oh, my God! That's what I want. I don't want, you know the answer. So much do we come into things and think we know the answer. This is a time in our life. We do not know the answer.

Amanda Bickerstaff: We're gonna keep moving. So anybody that has to go. I know we're hitting our at 8 40 whatever time it is.

Amanda Bickerstaff: 6 45 we are a little over, we're gonna keep going.

Harry Pickens: But what we'll do is this will be recorded for tomorrow if you have to go. That's totally fine. I'm gonna say, thank you guys so much for joining as always. But we're gonna keep rolling. Okay.

Amanda Bickerstaff: I know we're almost there. Okay, you wanna do. I'll I'll just want really quick, like lack of persistence and learning how to the to work with the tools. I think we already covered that with the 14,000. I don't have quite 14,000, but what I do is I also like to kind of play off each other and learn about different tools. And so a bit of experimentation again. But I think one thing I'll just say before I move on.

Amanda Bickerstaff: I always love it when I do a Pd. Where someone asks chatGPT to do something and it doesn't really work.

Amanda Bickerstaff: And the response is, this doesn't work, and my response is, your prompt was bad in a nice way. I'm not nearly that mean. But I say it's actually not. It's not the bot. It's you. It's like the bot is doing what it's supposed to do. But you gave it a bad prom, and that lack of persistence can make you think that it's not working when actually just didn't give enough attention.

Harry Pickens: Beautiful.

Harry Pickens: Okay, awesome. Next one, number 9. This is this is Harry's favorite. By the way, my favorite, because I think most people are 99.9% of people are completely missing the boat on the power of chatGPT, I mean. It's great for writing code and doing lesson plans and creating content, or whatever. But if you learn how to use it as a thinking partner to inspire your own creative and critical and expansive thinking. It's amazing. So I'm gonna show you very quickly one of my very favorite prompts

Harry Pickens: and how it works. So what I do is I will prompt. Cht, and I'll say.

Harry Pickens: well, I mean, this is actually something I'm I just moved to outside Chicago a couple of months ago.

Harry Pickens: So instead of asking for an answer, I'll say, walk me step by step, through a process that will help me blank in this case, create community and new city. But it could be it could be. Oh, you talked about lesson plans. So let's do it a prompt, very quickly. It says

Harry Pickens: I am. I am a new

Harry Pickens: Primary school

Harry Pickens: teacher. walking through step by step through a process, helped me learn how

Harry Pickens: how to create effective and engaging

Harry Pickens: lesson plans.

Harry Pickens: and then the caveat asks 1 one question at a time and stop and wait until I respond to proceed. So now

Harry Pickens: ChatGPT walks me step by step, through a process of questioning and reflection that can help me learn the whatever it is that I want, and so I can answer that question and iterate and iterate and iterate. I'm going to stop there. But the basic principle is.

Harry Pickens: don't just use chatGPT, to offload your you lower your cognitive load. Ask yourself, okay.

Harry Pickens: today, I want the I'm trying to accomplish X or Y walk me step by step through a process that will help me do that.

Harry Pickens: So that's that's it. How it's

I also wanted to say, though, when we were having our discussion before you all joined us. Is that harry really pointed out as well this idea, that

Amanda Bickerstaff: that reflection, that time for reflection. We know that as educators, reflective practice is one of the highest indicators of success with a teacher both in and outside of the classroom, you can use chatGPT, or generated by someone saying that pi is really good. So pi has a personality. That that's really good. But that's another opportunity to do is when you think about this, it can be your stop and reflect moment.

Amanda Bickerstaff: and it's something that could do it in a safe space in the sense of like, it really, actually doesn't care about you as a person. No offense like you're all lovely human beings, and I'm so glad you're here. But it's not. It's not a thinking tool, even if it looks like it is. It's not. But what it can do, though, is it can create a routine, a safe space

Amanda Bickerstaff: like all these things like routines actually are. We talk about how people like drinking a glass of wine

Amanda Bickerstaff: people are like, oh, they're they have less stress and less, you know, less heart attacks because of reserve at all.

Amanda Bickerstaff: No. Is that reserved at all? It's the action of taking time to step back, pour a glass of wine, sit down and have a drink, but that has a thing. And that idea of routinizing and creating spaces of reflection, using a bot or a journaling tool. Whatever you may have

Amanda Bickerstaff: actually helps us process information, lower our cortisol, create spaces for us to be able to build new thinking and new opportunities. So I think that that's something that people like Harry clearly do, which is so great because it is. It goes beyond this idea of just a simple thought partner, but it becomes a reflective partner.

Harry Pickens: Yes, ChatGPT. Generate 10 questions that will help me. Reflect on my my, my next, my next goal, professional goals in life.

Harry Pickens: or help me help me think through. I'm I'm facing this situation. Help me think it through one step at a time.

Harry Pickens: You know. The more the more you do that, the more you'll find yourself

Harry Pickens: bubbling over with kind of an irrational exuberance about how this tool can help you become smarter, more creative, more productive, and more human.

Harry Pickens: I think I for me the leverage power of our artificial intelligence is to learn how to use it, to augment our humanity.

Harry Pickens: and we can do that if we help if we ask ChatGPT. To ask us questions, to help us reflect, to help us think, to help us create. And so that that way. Instead of atropying our skills, we use it to boost our capacitals.

Amanda Bickerstaff: I love that. And that idea of augmentation, I think, is so great. And so we're gonna end on a positive. Because we are like, I said at the beginning, we are strength based top 10 mistakes. But we've been very positive. Is, this is an opportunity to have fun.

Amanda Bickerstaff: Have fun learn. When have we all been able to learn together at the same time? Or we know this is learning together. How cool is that? But also you need to be open to learning along with the advancements like, it's something where, if you think that right now. I can just learn about chatGPT. And I can come to one of Amanda's webinars, and I never come to another one, although you don't have to come to mind. You can come to someone else's. But I'm gonna stop. And then the world is gonna stop, too, and there's not gonna be any advancement.

Amanda Bickerstaff: You're wrong. It's gonna be advancement, not even every like month or week, or it's gonna change. And so we have to. It's have fun. Let's be open to have discomfort in positive ways. But let's also be open to keep learning, and I know from people I can name in this audience. So I've I've either seen week after week or mets, which is super awesome is that this is what this community does, and the people that get to watch us because it's the middle of the night, or they're teaching, or this is time for bedtime

Amanda Bickerstaff: and watch it later. I know you're that type of people. But how I am going to give you a challenge. How do you take that with you back to your environment? How do you continue to take this with you so that you're having other people also have fun. And understand, this is something that we can do together and grow together. And so I love. I love all the things, and also just keep talking. I love that you talk to each other. So here's the like. Well, this is a lot of information. But we're just gonna go through like this is what we did today.

Amanda Bickerstaff: You probably have your favorite one, or the thing that you see all the all with all the time. But we just wanna say Thank you. And so I wanna say, first, thank you to Harry. So Harry does such a give good job with the moderation, but also it has such an enormous amount of of openness and care and willingness to help people for free and support, and all the ways in which, like also, you can pay him if you want. I don't think he'd say no, but also this willingness and commitment to actually building this capacity. And a literacy is really beautiful. And I just appreciate

Harry Pickens: I appreciate you and your and you being here today. So thank you. I'm your number one fan. I think you're doing it work. So keep it up.

Amanda Bickerstaff: Thank you so much. I know it's really awesome. And then if you wanna hang out with us on Linkedin or join the Facebook group? Please do. We're gonna end a little bit early because I'm in New Jersey not to figure out how to get home. But I really appreciate everybody. Thank you so much whether it's the morning, the night, the middle of the night. I hope you have a really good day, and we'll see you again soon. Thanks, everybody.