Webinar Slides

Prompt Library for Educators

PD Workshop Planner Prompt

PD Idea Generator Prompt

Critical Thinking Prompt

How to Use AI for Lesson Planning

Student Guide for AI Use

Amanda on AI Detectors: Do They Really Work?

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.

Alex Muscat

Alex is the Head of Data and AI at Brisk Teaching. He made his start in EdTech on the teacher support team for the Summit Learning Program in 2016, serving 5,000+ teachers and 80,000+ students. Since then, he's maintained his passion for helping teachers learn how to effectively use technology in the classroom. A literature geek at heart, the advent of Large Language Models presented a perfect opportunity to combine his expertise in language, data, and education.

Amanda Bickerstaff: I'm Amanda, the CEO of and Co. Founder of AI for education really excited to have you here. This isn't a really cool topic, I think because I can help share the Pd strategies and best practices that I have seen in doing this work around the country, and not only the country, but we've now done this work in the Uk, the Australia, and in the Us. And so really excited to have you all here as always like, really want this to be something that we do together. I know I can't see your faces, but II know you're here, and I'll be watching the chat as much as I can on my own. But we always want you to be involved like II the thing that one of the things I love. The most about our webinars is that there's such a strong community that comes and talks, and the chat themselves become a resource. So please engage with each other. If you have a great PD. Strategy that you've used, please share it. If you have a question, there's gonna be someone in that audience gonna be able to help you out. So please do that. We're gonna do some prompting at the very end cause we just launched 2 new prompts around Pd. Itself. So we'll do some prompting together. But then, also, if you have a great resource, I know some people out there do please share it with each other. Right now. It's actually pretty funny. I'm at Microsoft in New York City. Kind of in a corner. at the end. Tech week Conference. This has been really interesting because it's been really focused on. What is happening not only in AI, but just a general education ecosystem. And it's been really interesting to see what's resonated with people. What kind of tools are being developed, but also been really surprised by how little we're talking about kind of AI literacy itself, and also about how are we preparing our teachers for this new technology? Because we know that school is changing around us. And so I'm really excited to have you all here.

Amanda Bickerstaff: as always like, really want this to be something that we do together. I know I can't see your faces, but II know you're here, and I'll be watching the chat as much as I can on my own. But we always want you to be involved like II the thing that one of the things I love. The most about our webinars is that there's such a strong community that comes and talks, and the chat themselves become a resource. So please engage with each other. If you have a great PD. Strategy that you've used, please share it. If you have a question, there's gonna be someone in that audience gonna be able to help you out. So please do that. We're gonna do some prompting at the very end cause we just launched 2 new prompts around Pd. Itself. So we'll do some prompting together. But then, also, if you have a great resource, I know some people out there do please share it with each other. Right now. It's actually pretty funny. I'm at Microsoft in New York City. Kind of in a corner. at the end. Tech week Conference. This has been really interesting because it's been really focused on. What is happening not only in AI, but just a general education ecosystem. And it's been really interesting to see what's resonated with people. What kind of tools are being developed, but also been really surprised by how little we're talking about kind of AI literacy itself, and also about how are we preparing our teachers for this new technology? Because we know that school is changing around us. And so I'm really excited to have you all here.

Amanda Bickerstaff: really excited to have you all here as always like, really want this to be something that we do together. I know I can't see your faces, but II know you're here, and I'll be watching the chat as much as I can on my own. But we always want you to be involved like II the thing that one of the things I love. The most about our webinars is that there's such a strong community that comes and talks, and the chat themselves become a resource. So please engage with each other. If you have a great PD. Strategy that you've used, please share it. If you have a question, there's gonna be someone in that audience gonna be able to help you out. So please do that. We're gonna do some prompting at the very end cause we just launched 2 new prompts around Pd. Itself. So we'll do some prompting together. But then, also, if you have a great resource, I know some people out there do please share it with each other. Right now. It's actually pretty funny. I'm at Microsoft in New York City. Kind of in a corner. at the end. Tech week Conference. This has been really interesting because it's been really focused on. What is happening not only in AI, but just a general education ecosystem. And it's been really interesting to see what's resonated with people. What kind of tools are being developed, but also been really surprised by how little we're talking about kind of AI literacy itself, and also about how are we preparing our teachers for this new technology? Because we know that school is changing around us. And so I'm really excited to have you all here.

Amanda Bickerstaff: I know, I say years, I mean decades, not like tens of year, like not ones of years. And so it's a really good place to ground this. And this is something that's been there before. It's just a major development that was a little bit faster or a lot faster than we than we anticipated.

Amanda Bickerstaff: And so the difference, though, is that these are tools that have very specific inputs and outputs. And so they're called classical or deterministic AI, and so means that I can predict what's gonna happen next in a set of roles based on training. And so you know, sometimes it does it really well, and sometimes it doesn't. We all have experienced a really bad chatbot before the last 9 months, where you just wanted a representative, and you kept saying representative and never got one like there. That's an example of a a very simple early stage. Non-generative AI tool. And so what I love about this is that you guys did a beautiful job. Thank you for playing along. But what you can do, though, is that you can see everyone you you have this understanding now that this isn't coming. This isn't magic. It's not coming from thin air, but what it is is it's building upon a body of work that's 70 years old.

Amanda Bickerstaff: And so the difference, though, is that these are tools that have very specific inputs and outputs. And so they're called classical or deterministic AI, and so means that I can predict what's gonna happen next in a set of roles based on training. And so you know, sometimes it does it really well, and sometimes it doesn't. We all have experienced a really bad chatbot before the last 9 months, where you just wanted a representative, and you kept saying representative and never got one like there. That's an example of a a very simple early stage. Non-generative AI tool. And so what I love about this is that you guys did a beautiful job. Thank you for playing along. But what you can do, though, is that you can see everyone you you have this understanding now that this isn't coming. This isn't magic. It's not coming from thin air, but what it is is it's building upon a body of work that's 70 years old.

Amanda Bickerstaff: So I don't know like what I would say if it was we were doing a Pd. Right right now. And if you play along with me and say, okay, so if that's an example of a I that's already been here, and you can't say ChatGPT. But what's another example of AI that you already have in your phone can. I'll give everybody just a minute to do that if you don't mind putting it into the chat. What is another example of AI, that Hardy is part of your life.

Amanda Bickerstaff: If everyone just submitted to to play along, you didn't know you'd have to play along. But you do so love it. Predictive text, absolutely. Theory. Spotify. Does a good job of being able to like your discover weekly. Got grammarly maps to plan a trip Google assistant. Even your fico square, Alexa, the algorithms and Netflix. Absolutely social media ads, ads, absolutely. If it remembers that you were looking for that one pair of pants that one time, and then, you see pants for the next. You know 3 years. You looked at one time. That's gonna be an example of AI and algorithm being able to follow you around and so these are all examples of AI that already exists in our life right?

Amanda Bickerstaff: And so the difference, though, is that these are tools that have very specific inputs and outputs. And so they're called classical or deterministic AI, and so means that I can predict what's gonna happen next in a set of roles based on training. And so you know, sometimes it does it really well, and sometimes it doesn't. We all have experienced a really bad chatbot before the last 9 months, where you just wanted a representative, and you kept saying representative and never got one like there. That's an example of a a very simple early stage. Non-generative AI tool. And so what I love about this is that you guys did a beautiful job. Thank you for playing along. But what you can do, though, is that you can see everyone you you have this understanding now that this isn't coming. This isn't magic. It's not coming from thin air, but what it is is it's building upon a body of work that's 70 years old.

Amanda Bickerstaff: And it's something that chatbots are not like, I said. Their first chatbots are like 50 years before us. Ibm. Watson is a generative AI tool that came out and beat jeopardy 5 years ago. But what's happened is we have we're actually gonna go here. What we've seen is this is on our website. And Dan, if you don't mind dropping in the chat the chat. But we have this really simple. Michael Kingston, who helps us build our curriculum, put this together as part of our student facing curriculum. But what we have here is a very simple, and it talked about only one who spend a little bit of time. But you can talk about how AI is the big field, right? Machine learning is a lot of what we just talked about which are going to be those experience, these algorithms that allow a computer to learn from experience or data without human programming meaning like you teach it something. And then it's able to recognize everyone's, you know, mostly PE faces. Unfortunately, bias occurs and it's less good at at minority faces. Unfortunately. But you have this opportunity of like training something, and then it can. It learns. And it's able to do with different inputs.

Amanda Bickerstaff: And so this is the idea that these, you know, they're they doesn't really work exactly like the brain. But it's using this idea of how brains are able to take many different centers and then generate new things. And so in this case we have generative AI and so generative AI, and we talk about Chatbot or midjourney or Gemini, that's just coming out is this ability to generate new text or audio images? The most common used case is going to be developing actual computer codes. So computer developing computer technology, Barry Metta, is like, that's where we have this really huge shift. And so this is an example of how you guys can bring that back to your school for institution and talk about what this looks like and really bring us down. So again, this is not magic. This is a computer program that is computing or predicting and not thinking. But it situates us in again. This is a big body of work, and even though it might seem very scary and new. And it is. It's I said. It happened really, really fast, and we were not prepared. I'm always like, think it's so funny when you see, like, you know, technologists, you know, talk about how you know it's extension event. And all these things. And one of the reasons why that reaction has happened is that you know, people thought that this was coming, but like 5, 1020 years in the future. And instead, when ChatGPT launched at the end of last year, and then more globally, beginning of this year, it happened so quickly, with so many more abilities than it than understood is that we all had to kind of speed up to understand it very, very quickly.

Amanda Bickerstaff: really want this to be something that we do together. I know I can't see your faces, but II know you're here, and I'll be watching the chat as much as I can on my own. But we always want you to be involved like II the thing that one of the things I love. The most about our webinars is that there's such a strong community that comes and talks, and the chat themselves become a resource. So please engage with each other. If you have a great PD. Strategy that you've used, please share it. If you have a question, there's gonna be someone in that audience gonna be able to help you out. So please do that. We're gonna do some prompting at the very end cause we just launched 2 new prompts around Pd. Itself. So we'll do some prompting together. But then, also, if you have a great resource, I know some people out there do please share it with each other. Right now. It's actually pretty funny. I'm at Microsoft in New York City. Kind of in a corner. at the end. Tech week Conference. This has been really interesting because it's been really focused on. What is happening not only in AI, but just a general education ecosystem. And it's been really interesting to see what's resonated with people. What kind of tools are being developed, but also been really surprised by how little we're talking about kind of AI literacy itself, and also about how are we preparing our teachers for this new technology? Because we know that school is changing around us. And so I'm really excited to have you all here.

Amanda Bickerstaff: about what this means in terms of like, what is AI? The next piece is, I'm just gonna go back. A slide is head on what are our concerns? So we do omits and fax. And so this is actually a set of slides, and if I'm in person I would have you stand up or sit down, because I mean that way. I want you to be interactive. If you're online, we would do thumbs up, thumbs down, or M and F in the chat. But here are some examples of like common missing facts, but also just to get deeply at what the concerns are. The first is around thinking like humans. And this is. This is something in which, you know, it's kind of it's deliberately, ambiguously worded because these are mimicking human, you know, like the human interaction, they're actually design, the ux and ui, that user experience is designed to feel like you're talking to like an intelligence. But in reality, what you have is that these are actually computing and predicting engines. And so while it mimics human experience, right? It isn't going to be human like. It isn't a human intelligence. It's not an intelligence. It's a predictive engine. And so what this does that kind of allows people to understand that like it isn't magic, and it's not Skynet. It's not Cal from space. Honestly, it's not the Terminator. But what it is is. It's a really powerful probability engine. That is meant and designed by the designers like Openai to feel like human interaction.

Amanda Bickerstaff: I know a lot of people in here already. So I know that you know that AI is actually very biased because it's that probability engine. That predictive engine is actually trained on the Internet in, you know. So ChatGPT 3.5 is trained on the Internet of text on the Internet, text and images. And what's happened is, I don't know about you, but I don't think of the Internet as a particularly unbiased place. So you know, it's something that is absolutely going to be something in which, we have these, like the responses that we get will have both explicit bias. So you can see those a lot, especially in image text image creators where you can put in you know, there's research shows. Tell me, show me a hundred judges what they would look like, and they come out where only 3% are women. But there are 34% of female judges in the Us.

Amanda Bickerstaff: I don't know if you've heard. But a lot of the rhetoric right now is AI is used for cheating. I don't know if you've heard that it's only been said, maybe one to a billion times in the last 9 months. But one of the things we also, we wanna acknowledge that actually, students and teachers potentially have can and will use these tools and ethically, and then sometimes they're using them in ways in which maybe actually, you know. maybe it isn't really cheating, because it's just busy work, or it's something that's become an outdated skill. But that there's this idea that these AI detectors work. And I'm telling you right now. when turn it in, says it's okay to have point 7% false positives. Which means that in 10,000 student responses you're gonna see 700 people be accused of AI written content. They say, that's okay. Not only is it okay, but you can still purchase it like you can buy it. knowing that those 700 are going to be students that probably are from non English language backgrounds, or really like purple prose, and try to be their best to try to be perfectionist, but that they say that's acceptable.

Amanda Bickerstaff: you may have heard this as well, especially if you've been an educator the last 20 years, or an Ed tech is that you know, technology will replace teachers. And so AI will place jobs like teachers. One, we we do quite a bit to get to that level of like existential questioning about what does this actually mean for my teaching? And you know, this one is really important to have a conversation, because there, there are jobs already that have been downgraded or degraded now in fields already based on the ability of AI. And so teaching, there are definitely some areas of teaching that can be routinized or brought out of the normal work day by an AI tool already. And so this is something in which we do have to have these conversations about what does teaching become. But right now there seems to be a very positive commitment to AI as a teacher, augmentation tool versus AI as a teacher, replacement tool. And so, while we know that schools will have to change.

it isn't a place yet that we are talking about replacement, especially if we really dig in and understand how we can change the way that school can be to make it better.

Amanda Bickerstaff: So this is a really good example of another way that you can kind of head on. This is a constructive way to talk about concerns instead of becoming this this opportunity, where everyone kind of

Amanda Bickerstaff: talks, you know, at each other, and you know not that this would ever happen, but, like I've been in the you know, a Pd. Where people get like cross talk, but this does it in a construction manner that allows us to really get deep and understand these sessions like what we're trying to to understand and talk about.

Amanda Bickerstaff: So I'm just going to move now

Amanda Bickerstaff: to a couple of things relatively quickly. And Nicholas asked about like the singularity as a myth or fact, at least singularities. This idea of, like the way that the you know

Amanda Bickerstaff: technology will go so fast that will outstrip our ability to understand it. And we're kind of at that point. I don't know, Nicholas, if I would really lead with that in A in a Pd. For teachers, especially not the first one, but definitely is for people that are thinking about that. They're usually a couple of people in the world that go that room that will go there for those people. I would definitely have a conversation about what that means, and because they're already kind of seeing around that corner of of what does this mean that we're going to be able to not only

Amanda Bickerstaff: have AI create new things for us, but also it's creating new technology. That's also feeding up everything.

Amanda Bickerstaff: behind about the idea of chatGPT being potentially harmful, it's not ChatGPT, but AI detectors. There's research that shows that it actually has false positives, more likely for students, or those that are non native English speakers. So if they're writing in English, they will be flagged as AI generated more commonly than their native English speaker peers. So that's what I mean, and I will put that in. I think then we'll put that in the email that comes after

Amanda Bickerstaff: just to kind of move a little bit is that you know we we use this a lot around like the idea of a brainstorming partner. The idea of 80 20. The idea that these are not thinking tools, that these are tools that can supercharge your productivity and planning, but they require the right approach. And this idea that they're you need to be conversational. How you can use this, but also that you're always checking for hallucinations and bias. And that hallucination is when that predictive engine, that probability engine says.

Amanda Bickerstaff: here's a citation or a URL that is completely made up because

Amanda Bickerstaff: that doesn't exist, because it doesn't even connect to the Internet. So this is a great way to kind of get into what it actually is good at right now, which is written communication, lesson planning, brainstorming, refining text translations. And even when I say formatting and structured data. I just mean, like, if you have, like an observation notes, if you're a leader, you can have it. Remove the the personally identifiable information. But you can use one of our prompts to use that

Amanda Bickerstaff: structure it and create a really beautiful formatted guide that helps you get that feedback to teachers more quickly. and so, and then on the other side, there are some best practices that show like this is idea of like what you always have to check for. And so you need to check for

Amanda Bickerstaff: the accuracy of what? What's happening like? Is it hallucinating? You need to look at? The starting point versus the final solution. If you're using a tool for like feedback or for lesson planning, or for even a Pd, what you wanna do is you wanna make sure that it's never a final draft. It's always gonna need to be reviewed. You always need to have you as the expert. You are the expert in this loop.

Amanda Bickerstaff: And so using that in terms of what the output is

Amanda Bickerstaff: the idea of knowing the limits of like these different tools. It doesn't do math. It can't do word count. It doesn't have citations. It doesn't. It actually isn't thinking it can't follow rules very well. But those are really important things to show. And and you know I can. We're gonna do a little bit of prompt engineering in a moment. But I'll show you an example of one that I use in schools. And then the final thing is protecting privacy. Please never like you really wanna make sure that your students, your teachers. Your leaders are not putting.

Amanda Bickerstaff: Personally, I didn't follow that information into these tools because you don't. They say like, if they're being trained on it, you don't want that to be there. But sometimes these tools don't work as expected, and so we don't want a situation like what happened with the Samsung engineer, where he used it to build a piece of code on intellectual property.

Amanda Bickerstaff: thinking it was safe, and then it got it was actually given to someone else. In another instance, like there are, these are new tools. And so it's really important to be as protective as possible.

Amanda Bickerstaff: So we have this prompt framework which will drop in the chat which we won't go into a lot. But we have these tools, because what I wanna do is actually go and look at 2 examples of like, we have some great Pd, we have some other pieces that will show. But I'm actually gonna stop sharing and then share one more time

Amanda Bickerstaff: share my screen.

Amanda Bickerstaff: So you may have seen a prompt library and it got an update in the face. Lift this weekend so we have now categories. We've added 12 new prompts. If Kelly's here. Thank you. Calli, who office put our a prompt library together, we have all kinds of different prompts. And so this is really great now, for if you're in a Pd. And you're talking to administrators or you're talking to, you're doing one with literacy training and then we have others that are going to be like I was gonna show you.

Amanda Bickerstaff: You all really quickly won't have time today. But play around like I actually used chat to ut to build that AI in your pocket. I was like, what are really good starting places for a Pd. And I use this idea of the idea generator

Amanda Bickerstaff: cut and pasted it in like, focused it on the kind of call to action, and then was able to build that out. Now I've used it all over, including with you. But we have these tools. But what I wanna do is like this. The prompt library is a great example of how you can get people really started and seeing the possibility and thank you to those that love our prompt library really appreciate it. We oh, wait, guys, you know, it's super cool. So the Jeff Rubenstein, who is

Amanda Bickerstaff: the head of education at Google Cloud on the main stage. Yesterday he actually shouted us out and said, AI for education, and their prompt library is a really good resource. So I bet with them today, and that was pretty cool to see. You know, someone that building these things thinks that we have some value here, and I know that we've seen that in the work that we do, it's a cool like, it's it's pretty awesome. And so yeah. But here we have these examples, and

Amanda Bickerstaff: you know, one of ones that I always start with is those that are really simple and starting places. So whether you wanna do a lesson hook you wanna do rubric, you wanna do quizzes like critical thinking. Let's you know what's today? Let's do critical thinking questions. I was really bad at questioning and so I have to be honest, like as an early teacher, I did not ask questions. Well, and so we're gonna let go and we use

Amanda Bickerstaff: chatGPT. And remember, we always create from real, from like, the the the non paid versions always wanna start a new context window. So it remembers me. And what I'm gonna do is I'm gonna take, if you if not seen it before. What we have is you've got this idea? That

Amanda Bickerstaff: you know it works with any chat. Bot. It also is like you. Can we have these like how to make it work for you. You actually always get to be in control. A lot of prompt and like prompt layers out there allow you to hit a button, and then it produces something, but you don't get to actually refine it. So we really like the fact that you can refine it, especially at the beginning stages of what you're doing. And it kinda keeps that control. But you're the expert. Remember, you are the expert in this loop.

Amanda Bickerstaff: So what I'm gonna do is I'm gonna call them over. And I'm gonna put this in. It says you're an expert educator, instructional designer. So I'm priming it with our like. Our prompt framework playlist of 10 open ended critical thinking questions from a tenth grade English class and the importance of using supporting documentation and argument of essays. So I'm gonna actually put this in. And remember, Gbt, 3.5 is our fastest model right now.

Amanda Bickerstaff: what it's gonna do is it's going to have this idea of here are top 10 open, ended, critical thinking questions. And so why do you think it's important? How can you distinguish? And you can see, like these are, consider a topic where there might be conflicting evidence. How would you handle this situation when you're crafting your argumentative essay? Oh, man, how great! That's a really good question. So I wanna do for that one. This is where it becomes that control and layer like, Oh, that's a good question. What you can do is now say for 7. Provide some responses that students

Amanda Bickerstaff: may have

Amanda Bickerstaff: from different perspectives. But I can do with some student exemplars, especially if I'm a new teacher. Okay? So you could like. So Sims may suggest that it's important to represent both sides.

Amanda Bickerstaff: The idea of choosing the strongest evidence. You know, like address and refute. So interesting, right? But like, this is idea of like, you can actually be prepared. But maybe if a kid or kids don't get there, you can kind of use good, you know, prompting to actually get them to think about these pieces. So I really really love this.

Amanda Bickerstaff: So I'm gonna do the I'm gonna do a similar one just with Claude to show you the like, how it works. It's, you know. I think it's actually it's not the best at some things, but it's starting to get better. I actually enjoy. If you watch our video earlier this week I enjoyed trash and trends as an opportunity to teach kids about fashion and recycling and so what we can do is, you see again. So these are, they are open, ended. And what I can do now is I can kind of use these 2 tools in concert.

Amanda Bickerstaff: and I can ask the same thing again where we have. I'm gonna pick a different number. So how can I?

Amanda Bickerstaff: Oh, how about 9? Nine's a good one? What? How would failing to document sources? That's a great one. Love it

Amanda Bickerstaff: and see like we have the

Amanda Bickerstaff: some ideas of soon soon responses and a little bit different, like again. It's gonna be like they're gonna have different strengths and weaknesses. But this is an example of like where we've got you know some good examples of how we can do this. So that's a good example of how it works like. And it's working. But I'm gonna show before, like, I'm gonna show where it fails. Because remember, we wanna show the positives, and we wanna show the limitations and capabilities. And so

Amanda Bickerstaff: I've done this live a couple of times and so remember, this is a predicting computing engine, not a thinking engine. And so what I'm gonna do is I'm gonna ask it, create a mnemonic.

Amanda Bickerstaff: You think I would be able to spell this. Now, wait, don't do it.

Amanda Bickerstaff: There we go on. Bloom's economy

Amanda Bickerstaff: and rate your output. And so what we wanted to do is have the first letter of every word batch moves, autonomy. And what we're gonna see is I'm asking it to rate its output as well. And so what it's done, is, it's

Amanda Bickerstaff: it's also just given me an mnemonic. Actually, it's just the framework. It's actually not a mnemonic at all. This is always really fun like this is definitely non mnemonic, like Roy G. Biv, but it's given itself a 7. Add 10 cause. I love. What I love about this is that not only is it wrong, but it's extremely confident. And it's wrong. Answer. So I'm gonna say, this isn't correct.

Amanda Bickerstaff: Because it's not a mnemonic

Amanda Bickerstaff: device. please, I always say, Live, please do again. I'm the one of the polite people and rate your answer.

Amanda Bickerstaff: And so now whew! Okay. So now we've got

Amanda Bickerstaff: raging unicorns always attack evil. So, Lisa, better. It's actually at least a bit more like this actually kind of works. But it's done, is it's giving itself a worse rating.

Amanda Bickerstaff: And so what you're gonna see is that this is an example where you can show those limitations. You know, it's Shadow. T does the worst job of identifying hallucinations. Where it says something that looks correct. But isn't the Endor does the rating, and if we do it one more time I put this in where we're gonna see is another answer, because the temperature of this. Ssl, I'm writing that temperature is very high. And so in this case, what you're gonna notice is it's gotten completely off track.

Amanda Bickerstaff: Do you see? There's no J, and Bloom's taxonomy and so and that's given itself a higher rating. And so this is going to be that example of like, how there are limitations again, is not a thinking tool.

Amanda Bickerstaff: So another. What I want to show example of a tool that actually does is trying to show hallucination. So I'm gonna ask for. Ask sorry it's been asked for provide 5 Urls for a lesson on argumentative essays.

Amanda Bickerstaff: So we know that both

Amanda Bickerstaff: chatGPT to current version and the free version and clawed or not connect the Internet. And so what's gonna happen is you're gonna notice is that you have right here

Amanda Bickerstaff: these these pieces. But immediately, if you see down here, it's a little bit small. It's given a

Amanda Bickerstaff: immediate warning that these links are most likely not true. And so it'll often happen is, even though it it's confident. And it's giving you 5 essays. What it's done is they know that it's gonna hallucinate when you ask that. So it's giving you an identification of where it is. So I always show this in a in a training. And so in this case, look, it's a it's an empty link, but the best part of it is, it looks like it's actually correct, because it looks completely right. But it doesn't work.

Amanda Bickerstaff: And so when we think about again, here are these capabilities. And here are these limitations like, what we wanna do now is like, go into you know what we're doing and having that be an opportunity to really, deeply understand what's possible.

Amanda Bickerstaff: And so Carl says about being chat about tendency, hallucinate random leaks, even even though it's connected. Internet. Google just released something that I really like. But so there is this idea that now you can, you hit a G in in barred, and it will at least highlight. What it can say is true, based on some resources. And so I think that this is gonna be something that's gonna be more

Amanda Bickerstaff: more and more possible. Thank you, Erin. Just like, let me like, see if I can share this, Erin, just put in ruthless elephants. 8. Can you guys see this

Amanda Bickerstaff: with us elephants, 8 juicy cares. This is a great example of how you can have fun with AI. Someone talked about being nerdy. I'm clearly very nerdy. But, like Erin, I love how you take it even further. And so this is a really good example of like how you can make it fun. But you are going. That kid is going to remember that kid is going to remember that that was wrong, or that teacher's gonna remember that that was wrong, and that I need to be careful and that I need to be a critical user of

Amanda Bickerstaff: this technology. So we're, gonna this is our, this is kind of our usual stopping place, and I just really appreciate those that are joined us already. I will stay on. We'll stay on. I do have to go to a dinner and a couple, and like. So Linda little bit earlier than usual. But I wanted to. Just always say, thank you.

Amanda Bickerstaff: we have something that I love for you all. If you're you know we appreciate your time and attention. We have a bit of you know, always connect with me and Linkedin but we also have, Dan's gonna put in the chat. We have a couple of questions that we'd love to ask, cause we're starting to think through. How can we help at scale? Unfortunately, I can't be in all your schools, even though I'd love to, and I'd love to meet you all in person. So we're thinking about like trying to scale this and like actually creating like a certified train, that training

Amanda Bickerstaff: or other opportunities. So if you have a little bit of time, if you are, be open to answering, that would be really great, and if you had a ha have to head off, please. You know, that's okay, too. We'll put this in the emails later. But I wanted to take this time like we have 5 good minutes that anybody have any questions that they like me to answer. If not, I just hope you have a really good night. Morning evening wherever you are. But if you have a couple of questions.

Amanda Bickerstaff: we can spend a little bit of time together. And so yeah. So

Amanda Bickerstaff: I don't know. Does anybody have any questions? We've got a couple in questions, but nothing we talked about. The inaccuracies. This will be recorded. Gemini is gonna be really interesting. So Gem and II just talked to the to Jeff from Google. And he said, it's much, much better. There's also Ed palm that's coming out this for educated education. As well. And then so Carrie has a question what a school is doing about terms

of you saying, students have to be of a certain age. Okay, man, that's a really important question.

Amanda Bickerstaff: 13. And up for chatGPT, and bard, and being with parental permission. Perplexity says the same thing. Claude is 18, and up. We we really say that at this stage, because the technology is extremely

Amanda Bickerstaff: early stages and can be unreliable and harmful to be very careful with letting students of younger ages, knowing that they're going to be using it in different places like Meta or Instagram or Tiktok. So what we suggest is we have this guide that we're building, that it's much more about together. So actually getting up in front of the classroom and doing this together, having kids prompt with you. We have a a great new, prompt about imaginary friends or creatures that students can get involved with, and I know that there are some people that are really starting to talk about AI with kids as young as grade one and 2. But there's a difference between building this kind of knowledge and understanding of like what it works, and then letting them loose into these knowledge

Amanda Bickerstaff: fit for purpose tools that are going to at times be an appropriate that we show hallucinate. So there's a difference between guided tools. And I think that we're gonna see more and more tools on market that are gonna have protections in place that make them safer. Like I said, Google, I just talked to them today. They are saying that they're creating a version that is safer for kids. So we'll see that happening more and more.

Amanda Bickerstaff: As so, Michael, we'll we'll definitely be able to share the slides. so to do train the trainer would be great. We'd love it. Yeah, so thank you for everyone with the nice feedback and so yeah, if you wanna, we are scaling up. I mean, we're early stages. But we're definitely gonna be looking for facilitators. But yeah, I just really thank you guys for joining. And thank you. This is our fifteenth one. I just think that it's been so lovely. And the fact that I get to know you guys is really cool, too. So I hope you have a wonderful evening morning. Wherever you are. It's so nice to have you be part of our journey. And

and yeah, I just really think that this is a great opportunity for us to work together, and good luck doing this in your own schools and contacts. But thanks, everybody have a really good wherever you are, and I hope to see you again.