Starting the School Year Strong with AI Adoption

Common Sense Media Report

Beyond the Tip of the Iceberg

FOSI Teen Study

CDT Research

5 Key Questions to Ask Teachers & Students About AI

SEE GenAI Literacy Framework

GenAI Misconceptions

Guidance on AI Detectors

Student Guide for AI Use

How to Use AI Responsibly EVERY Time Framework

Prompt Library for Educators

FREE GenAI Literacy 101 for Students Course

Parent Guide

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.

Dr. McGee is an educator, researcher, and advocate dedicated to creating or promoting access and opportunities for students navigating a future driven by artificial intelligence. She has served as an educator for nearly two decades. She taught middle school mathematics and was recognized as a district finalist for teacher of the year. Nneka has also served as District Gifted and Talented Coordinator, Director of GT and Advanced Academics, Executive Director for Learning and Innovation, and as a Chief Academic Officer.

Most recently, Dr. McGee served as project lead and principal co-author of “AI + Learning Differences: Designing for a Future with No Boundaries,” published by Stanford Accelerator for Learning. Her latest work, “AI for Young Learners: Balancing Tech Use with Authentic, Off Screen Experiences,” was published as a Jump Start Guide by ISTE in January 2026. She also served as an AI practitioner advisory board member for the Engage AI Institute and was selected as an EdSAFE AI Alliance Fellow in its augural cohort.

‍ 00:00
Amanda Bickerstaff
Hi everyone. Super happy to have you here. It's our first webinar in a little bit of time. We've actually taken a little hiatus because were in lots of schools and lots of districts and talking to lots of leaders and educators this summer. But I'm really excited to have our back to school webinar. This is kind of crazy. We started in August. I would say that August of 2023 was really the kicking off point. So this is the fourth school year. I can't believe to say that we've been doing this work and I think it's really interesting things have changed, but they also feel pretty similar in some ways. A lot of opportunity to dig into what we've seen work.

00:41
Amanda Bickerstaff
And I'm really lucky to be joined by Dr. Nneka McGee, who this is her first the coming out party that Nneka is going to be joining us full time at the end of the month to be our new director of partner strategy. I'm sure you've seen her great work in terms of releases from Stanford, including the last assessment paper that came out, I think just a couple weeks ago, Nneka. But we'll have an opportunity for you to introduce yourself in a moment. But I want to set the scene. We find it's just so important for us to have these opportunities to learn together. This is the opportunity for us to share what we see on the ground and some of the resources that we have been building.

01:21
Amanda Bickerstaff
But also it's an opportunity for you all to share with each other anything that you've seen work. We've had amazing chats in the conversation, so two things. One is get involved. So say hello. We usually have a pretty international audience as well. I would say if you're thinking about talking to each other, the chat is the best place. But if you have a question specifically for Nneka or I, including any burning problems of practice that you're thinking about before you actually move on to the school year, please let us know. And just a reminder, just make sure that you have your chat for everyone, not just host and panelists. I see a couple of wonderful people saying hello, but just make sure that everyone can see everything that you're saying.

02:06
Amanda Bickerstaff
And I want to just before we move on, I'm going to come off share. And Anika, I would love you to introduce yourself for those that are not familiar with your work and kind of what's on your mind with the start of school coming. I mean some of school starting already.

02:20
Nneka McGee
Yes. Well, first of all, thank you for having me. Amanda, I'm excited about the work and joining. It's, it's been great. And then after a couple of years of being a part of different opportunities where we've collaborated together, it's great to be in a closer collaboration that's going to benefit, you know, educators and students. That's very exciting work. And being service oriented as the team is, it just feels right at home. So I appreciate that. In terms of my work, I actually, my doctoral research was focused on AI education and its implementation in K12 learning environments and I completed I, you know, tell the story of two weeks before ChatGPT came out is when I received my IRB approval. So it was very interesting seeing kind of pre chatgpt post chatgpt and having to do what they call deviations because it was so explosive.

03:20
Nneka McGee
Since that time I have been privileged, honored and also humbled to be a part of work ranging from the U.S. department of Education, organizations such as Digital Promise, as you mentioned, doing research and writing with institutions such as Stanford Accelerator for Learning. I also do presentations work with organizations as well as we all try to steer the ship of AI and education and its impact on our systems.

03:53
Amanda Bickerstaff
Yeah, I think that one of the things that we think a lot about when we're thinking about our team is that I think that I always gravitate personally to people that I feel like are practitioners at heart and continue to be practitioners at heart, but also those that have a really unique point of view. I think that we are really lucky here that we get to create things that have never been created before. And I think that's something that we're really excited about you joining with us, Naika and for everyone joining, I know that I saw some people that are starting a day, if you're in Charleston, who I do a lot of work with.

04:26
Amanda Bickerstaff
I know today's the first day of school, but for those of you that may be in the on the west coast or at the east coast like I am in New York, you still have a couple weeks left. And so what we decided to do for this first webinar of the new school year is to really focus on what we've seen, work and some of the resources that we have. But we always want to start with evidence. And I think that this is the most important thing to our work is that when we started, I think our first webinar ever Nneka was like June 14th or something crazy of 2023. And we did 25 webinars in 26 weeks. And at that time, everything that were creating was just like, there was almost no evidence space.

05:07
Amanda Bickerstaff
And what's been so positive to see over the last couple of years is how the evidence base has started to get stronger and stronger. And if you're a researcher on the call, we would, you know, we love the research, so keep it coming. But I think that one of the things that we know is that young people are using these tools at an ever increasing rate. And I think that, you know, I really. It's hard to say I like this study, but I do think it's important that it looks at not just high school students, but also elementary school students, because we often hear that elementary school students are not accessing these tools. We have time for AI literacy. We have time for all of this.

05:45
Amanda Bickerstaff
But what we see is that this Common Sense media report that came out a couple weeks ago, a nationally representative sample, more than 1,000 students, they asked them questions about their AI use. I want you to notice just how many of these young people are using AI and not just for schoolwork. So this is something we've been talking about for a long time. But the gr. The dark green is daily. So you have like one out of five students, or in the case of entertainment, one out of four students on average are using this for entertainment. Like, how crazy is that they're doing that not just like on their own, but daily? We see that with schoolwork. You see, that's where you do have a little bit less in 9 to 12, but it's still pretty significant at 74%. And how that ramps up.

06:32
Amanda Bickerstaff
And the way we've always thought about it here at AI for education is the older the student, the more likely they're using general AI for schoolwork. I think by the time Naika, they get to your level at the PhD level, you're probably using it all the time. And so what we have is that we see this kind of happening. But then I think that what we believe very strongly here is that generative AI is going to have an impact on the whole person, the whole child. And so we created an iceberg. I don't know if, Nneka, if you ever saw this, we have an iceberg graphic that's like maybe two and a half years old, where it was like, AI is only for cheating, and then underneath it was like, misinformation, AI companionship.

07:10
Amanda Bickerstaff
And then what you see here is that students are turning to these tools for advice on your health or body, future decision making, practicing conversations, or social skills. Or discussing your feelings or personal problems. And here's the crazy part. Like, that is so much more than. I mean, it's like social media is like somehow had this hybrid crazy baby with Google search and therapists and friends. It really has become this incredibly multifaceted and disruptive force in our young people's lives. And we also know that what's happening is that. And I'm going to stop here and have Nneka you join in because this is something I think that we both find to be staggering and that when students are asked or young people, those high school remember, 94% of them say that they're using general VI for schoolwork.

08:01
Amanda Bickerstaff
Where who is most responsible for teaching me about General AI? Just take a moment to bring this in. That 28% of them say it's themselves. And I think that this is such a crazy statistic that we're essentially saying the most disruptive technology of our time. 200 Scientists and economists came out a couple weeks ago saying we must act now, that the amount of change in the next 10 years will be as significant industry as industrial revolution. And that the more than 30, like almost 30% of young people are saying it's themselves, then also their peers and then the 16% at their school than actually going to general BI itself tech companies. But you see that if you look at this, the majority of what's happening is young people are learning about this on their own.

08:50
Amanda Bickerstaff
So, Nick, I talk a little bit because you've been thinking about literacy for, I would say longer than most people. Definitely from me. And so when you look at this, what does this make you think?

09:02
Nneka McGee
My first thought is that as educators, we need to take this as an opportunity to understand that we have to take on more responsibility for ensuring that teens are getting information, particularly about generative AI and its effects from adults, from people who are informed when we, you know, kind of, if you look at a teen and what they could be exposed to and the risk, yes, there are a lot of promises we all like to see, you know, a cat dancing or, you know, things where it's fun. And that's great. But there are all. There's also the other side that we need to take into account. And teenagers, even, you know, preteens, they may not first have the awareness or have the resources in order to review that information, study that information, absorb that information. As adults, we are in that position.

10:08
Nneka McGee
And particularly as educators, when kids come into the walls of the classroom or in the halls of the school, that is the opportunity for us to take the time and be intentional about making sure that they understand the impact of generative AI in their lives.

10:25
Amanda Bickerstaff
Absolutely. I think that when we look at the next statistic that we have is that and I would say that this student statistic feels a little bit overblown from what we. But that we're tracking ahead where educators have at least had one training on AI literacy. Majority of the educators, which is pretty significant considering how I think that this might be one of the leading industries in terms of some sort of AI literacy worker attempting to do that. But what we see is that 35% of school districts have done some AI training for students. I was just in a room last week, nneka, where did this wonderful kind of futurist convening with NJSA and ClassLink and you have 100 people across leaders from across New Jersey. And I asked who has done AI literacy for your students?

11:18
Amanda Bickerstaff
And less than five people in the room said yes. I think that tends to be what we see is that young people are just really on their own here. But also what we know is that pretty much no professional development for teachers works if it's a one and done. I mean, I think that Nick, you know, more you're a cao. You must have seen all the times you're like, oh, but they had training or like they already have this. And what we see is that the majority of our teachers, whether you're in the room as a leader or teacher, you might recognize this in your own work have had that more than one training or even just like where we know high quality job embedded professional development is the way to go.

11:58
Amanda Bickerstaff
But then I also thought it was really interesting that only 22% of teachers have received training on Genii Risk. And it's one of the reasons why in our framework is that Genai is not just going to be a traditional I need to protect, not put a birth date in or a name. It's so much different. And I think that, I mean, maybe talk a little bit about it, nneka. I mean, you see how important the change management option of this is. But thinking of it at if you were still at a district where you would think about this becoming potentially an issue.

12:31
Nneka McGee
So, you know, in looking at my role, you know, as an executive and then as a chief academic officer, I had started AI initiatives and one of the things that was effective, like you said, is not the one and done, but professional learning over time. So were very intentional about starting with the cohort, ensuring that they learned about AI and then over time, had regular meetings, they set goals, looked at those goals. And I find that not only with generative AI, but with ed tech adoption in general, that approach has been more effective. One of the most disheartening things for me as a cao, and if we have any principals or leaders or even families if responsible for investing in some type of program, then you look at the usage and it's like no one's using it.

13:28
Nneka McGee
A part of that reason and educators can agree is if we don't have the training, the chances of using it are small.

13:36
Amanda Bickerstaff
Absolutely.

13:36
Nneka McGee
With that, like you said, the risk, I think when you know, chatgpt and then other forms of chatbots and generative AI grow in popularity. We thought about the application of it and how to use it and the implications of it in the classroom, but not what are the impacts. And a part of that you discussed earlier, where's the research? Even now there's a lot of studies coming out, but if you look at scale to scale initiative and what they found in their studies is that there aren't a lot of causal studies yet that are available. So with that, now is the time that we're starting to see things such as cognitive offloading. We're starting to see possible impacts to critical thinking. We're starting to see sycophancy and how that can potentially negatively impact a student's experience.

14:37
Nneka McGee
Now that we're starting to see those things, it is essential that we help educators learn more about those topics so that they can be aware and then support our students.

14:50
Amanda Bickerstaff
Absolutely. And I think that it is just so important. We're in talks potentially to work with an organization around a larger study because of some of the work we're doing. But it is anecdotally, I think there is so much to be said about just how disruptive this has become. And we know now that like you said, kids getting higher grades on formative assessments pretty significantly, both in high school and in post secondary. And then when they get to that final exam or that novel application, they're struggling, their grades are going down. And I think that is something that. The idea here is that Genai risks for young people are not just academic integrity or data privacy, but it can be something like preserving voice and preserving cognitive and productive struggle. Those types of pieces are really important. And I think that it is.

15:48
Amanda Bickerstaff
Owen asked us. Owen is a district administrator. I think that asking about fighting for limited time, I think that it is really interesting because when we talk Owen about this work, we talk about it can't be another thing. And so it's 100% better to pick one or two key initiatives that you're already doing, whether it's around teaching and learning or around instructional approaches or professional development, whatever it is, and then find a way in which you can bring AI literacy into that approach and so that it becomes embedded and meaningful.

16:27
Amanda Bickerstaff
So if you have a big push for students having self regulation teachers potentially learning how to use generative AI systems to create, you know, like better guides and better studies and like to give kids feedback in meaningful ways and I think or to like create a gem for a high school student where they can actually have their own like and they actually can have something in which they work together with the right AI literacy to move forward. And I think that to find something that is small and structured and aligned is a lot easier than saying AI can do everything, it can help everything, it can do all your lesson planning, it can do everything, which we know is not the case. Right. These tools are not pedagogical. The teacher is pedagogical. The tool is predictive. Right.

17:14
Amanda Bickerstaff
And so if you can connect those dots, I think can really help. But I think that these types of, when we think about how we build Buy in is that we do this, we show what's happening, we show the research and evidence because that is one of like, it becomes very unavoidable if you're literally looking straight at this is what's happening. So I would suggest another way to do this is we have a. I'm going to put daniel on the spot like I usually do. If you've never been to a webinar, I do like to go off track. But our five questions to ask students can be amended and get data on how much the kids are using at what grade levels and how they're using AI companionship, how they're using these tools for advice.

17:57
Amanda Bickerstaff
And as soon as you do that, there's no oh, but it's, you know, we know what's happening and kids are just doing X. It becomes something like, oh my gosh, this is so much bigger and more important than ever before. And I think that we find that to be probably the biggest lever to change and adoption. So that would be my suggestion. Nneka, do you have anything else that you think around kind of building that buy in around AI literacy?

18:23
Nneka McGee
So one effective tool we use, and I kind of touched on it a bit earlier, were empowering, particularly teachers or other types of staff, whether it be librarians, paraprofessionals. So of course, if you're a digital learning coach and sometimes you're the only one in the district, it might be difficult for you to have the reach that is necessary for AI adoption. However, if you develop a cohort of individuals from different areas, and I sometimes don't want to limit it to desk teachers because you have transportation staff, everyone across the ecosystem is going to be impacted by generative AI in some way and needs that knowledge. So if you develop that cohort, you as the leader in that role, you're disseminating the information that they need to in turn go out. And that limited time that you talked about is so true.

19:16
Nneka McGee
Those professional development calendars are developed far in advance. Everyone's jockeying for that small amount of time. One thing that tends to happen are regular meetings. Whether it's at a school level, whether it's a district, you know, a different district office, those meetings are happening regularly where individuals do have that time. And so maybe this, you know, they get on a calendar for that particular meeting and it's. They're covering something that you've gone over as the leader about AI literacy and then they're disseminating it to their teams and it kind of go, you know, goes in that way and down that path. That has been a very effective tool for ensuring adoption and for more people to take it seriously. Because your reaches further because you've touched more people by having a kind of cohort model.

20:07
Amanda Bickerstaff
Yeah. And we see that with whether it's to train the trainer model, you know, or it is going to be something where it could be like an AI lead for like where each building has an AI lead. And so we're going to shift a little bit into what we think about for this upcoming school year. And this is something that we think a lot about here is that, you know, what we would hope is like the goal for most of us, especially thinking about back to school, is that we would love to see our schools moving towards this approach in terms so having a community wide AI literacy plan. So of course for educators and leaders, but also for students, for families, for board members, for community, like if you have a community based organization working with you, that could have that greater impact.

20:51
Amanda Bickerstaff
We would believe that is. I think it always staggers me how wrong I have been about how long it would take for it to be a national priority for AI literacy. I think, nneka, I have been wrong so much and I continue to be wrong. I think I might be a little naive and optimistic about this, but I think that having that shared understanding, common language does go a long way in terms of safer practices but also the ability to have hard conversations about what's really happening in our schools. The second is guidelines for responsible adoption. We definitely think that the two things that must come hand in hand the community I literacy only makes sense if there are guardrails to keep us safe. And the guardrails only make sense if people have enough AI literacy to understand why they're important.

21:39
Amanda Bickerstaff
And so what we would suggest is having those be like I would rather if I had my magic wand. You have a goal for AI literacy for everyone and a simple set of like one pagers just about these key understandings of around safety, ethics and effective use so that there's something that holds us together. It does not need to be like what we did for Houston or Chicago. It could be a one pager for each group. And I think that can go a long way. And we'll share an example. And then finally I think Nneka, you nailed this especially right now.

22:11
Amanda Bickerstaff
Has anybody heard of techlash where we have you know Utah and Tennessee and you know we have all type Oklahoma like there are so many laws that are being passed around access to technology of any type of in elementary classrooms, potentially even in high schools and middle schools. And I think that is combined with I have heard from many of our partners that we bought a tool, a Jenny Aya tech tool and then all around we have like almost no one's using it or if they are using it's not being used effectively and because of the expense we're no longer going to have it. So we have is that whiplash moment of this is what you're going to focus on and then it's removed.

22:57
Amanda Bickerstaff
And I think that what we want to do instead is flip that in general to more evidence based approach where if you are thinking about bringing in a Genai ed tech tool or a Gen AI foundation model is that you're doing a pilot process first identifying what you want to see in terms of like impact and effectiveness and then do that before you're going to buy this. And I will say right now Genai ed tech is in a little bit of a world of hurt with everything changing with budgets and these pieces. And so they'll most likely be willing to work with you on a plan like this. And I think you'll have more ability to have a say on what training would look like and what access would look like.

23:37
Amanda Bickerstaff
And so I think we would want to have that as an ultimate goal for this year as well. So Here we're going to do it. It's actually interesting about Sarah, your question about the predictive nature is definitely that's an interesting statement around knowing when and when not when to use it safely. So this is what our entire Gen AI literacy framework is about. And I think it's really interesting because we released our C Framework for AI Literacy on April 10. Everybody locked in my head and what we have found through doing our new Genai literacy aligned work is that we think that Genai literacy is as much about informed decision making about Genai than any type of skill.

24:23
Amanda Bickerstaff
It is so much more of a mental model and a decision making framework than an idea of a functional literacy in the sense of I'm going to learn how to prompt or I'm going to learn how to fine tune a model. And I think that's been really interesting. So when you talk about how in the case of Sarah, I might not be able to build a car, but I can drive it and know when it's potentially something is wrong. But there are lots of things built into a car that tell you when things are wrong. Whether it is going to be warning signs or things that no longer work or you start hearing things, you have a lot of input from the car itself.

25:05
Amanda Bickerstaff
I think the issue with generative AI is that there are almost no signals to when something is wrong, inaccurate, sycophantic, meaning pleasing. I think that's why you can't rely on your lack of knowledge, right? You have to have the ability to understand how the tools work so that knowledge of how the tools work enough to be able to understand what could go wrong or what could go right. The ability to apply these key mindsets in terms of that decision making framework and then have that in your head of am I using these tools in safe, ethical and effective ways? And I think that until the tools become better and potentially more safeguarded, it requires a lot on the individual. Probably I would say too much Nneka without change.

25:52
Amanda Bickerstaff
That's why we think tech companies, if anybody's in a tech company, a big foundation model, we really like your tools to be developed in a way that people can understand what's really happening and when not to trust the tools. But I think na, when you think about that decision making framework of Gen literacy, what really comes to your mind.

26:11
Nneka McGee
In terms of looking at, you know, the C framework? And when I see questions like what Sarah asked about the balance, I'm looking at those safe practices in a little bit of those ethical practices. And so it goes back to what you said, understanding the predictive nature. What are those important things? You mentioned one in particular, sycophancy. Not understanding when it's being pleasing. Well, that comes with your safe practices and understanding what type of language, you know, getting the appropriate training. So that when you see that language from an output, it's like in your mind it's like, oh, it's being pleasing, it's trying to influence me. That means that you're going to act in a way that is safer for yourself and for our kids to understand when something is not, you know, when it's. Something is sycophantic.

27:06
Nneka McGee
So not only do you understand, but now you're able to help a student understand as well. So that is where I see that balance in that nature. But also connecting it with the. Just one example of connecting it with the C framework, just like cars is it always going to be 100 safe? It's going to be as safe as we can make it. And so I like to, you know, it's going to take input and also I'm trying to capture that the correct word escapes me now, but I just feel like it's going to take a lot of us together, working not only as individuals but as you mentioned with tech companies and really driving the necessity of. For putting guardrails in place. That is that missing piece when it comes to that safe part of the discussion.

28:04
Amanda Bickerstaff
Well, I think, to the point that I think a lot about is that, you know, if I don't put a 12 year old in a car and say go, that safety isn't just about knowing like how to drive. Like I still wouldn't give a 12 year old who had the same driver's ed course a car as an 18 year old. I also will be honest, I was a terrible driver until I was like 18 because I have attention problems. And so thank goodness my dad owned a car dealership, although he probably is still mad at me. But the idea here is that one way to think about this is that gendervi tools, especially those that are not safeguarded, are not for everyone at all times. And maybe not for age ranges at all. Like it doesn't have to be.

28:50
Amanda Bickerstaff
We're talking about the age appropriateness. But if we think about this and think about safety and think about this work is that I think about this so much and I think that this is. It becomes somewhat controversial. But I would not use generative AI if I do not have expertise to evaluate the outputs or cannot work with an expert to be able to evaluate them. It doesn't mean you can't Vibe code or do other things. But would I ever use a Vibe coded app without any ability to fix that app and then deploy it and try to get people to pay for it? No. Because I'm going to tell you right now, you will find out very quickly how difficult software engineering is and how difficult it is to actually build something that people can use and find value in.

29:35
Amanda Bickerstaff
But I think the thing that I think a lot about, in fact I've been telling this kind of anecdote for almost three years about how an educator came up and said, amanda, use our time to prompt to create this. I've always wanted to be a gardener. I have this $500 budget and like list of things to buy from the, you know, the garden shop. And she's. I was like, do you know enough about gardening to be able to know if this is a good decision? And she looks at me, she said, no. So even though she. And so I was like, maybe talk to a garden or talk to like, go.

30:07
Amanda Bickerstaff
And it's interesting because literally today I shared in our Slack channel that a Taiwanese, I think it was Taiwanese, you know, farmer, industrial farmer, used trash VT to identify how to support his crops and he ended up killing 25 of his crops because the technology does not know anything and gave him the wrong suggestion. And that's like the massive scale of that too. But I think that this is where we have to think about the idea that these tools, even though they can do a lot and they show you that they can do more than they can, it doesn't necessarily mean that is what we should be using them for. And I think that's why we think about the way we put everything together, is that it has to be something that's a cycle.

30:53
Amanda Bickerstaff
The idea is the knowledge, mindsets and practices come together and when you know more about how these tools can make mistakes, can be like sycophantic, can be biased, cannot all these things, is that you build a critical distrust of the tools that are so necessary. Everyone talks about how, you know, we have, you know, we don't know enough about these tools. But what we see is that like, if you start looking like, start thinking about this more deeply, you're going to start to see the holes. And I think that this is where. Here we go. And I agreed with everybody in the chat, the sustainability. They talk about how if this is a decision making process, the decision can be do not use the tool. It is not a reflex, it is not an all in one.

31:42
Amanda Bickerstaff
But I want to share our new resource with everybody. I'm really excited about this because when we wrote our C framework, we got to very close to the end of our time of having to write it and I was like, I really want a misconceptions table. This misconceptions table is genuinely developed based on, I'm not joking, the amount of people we've trained and seeing this happen over and over again. So I'm going to open it up, nneka, and let's look at it together. But as everything that we do here is that it's going to be not just a resource, but it's always meant to be practical. I don't, I mean we're pretty, pretty boring in this. Like, it's always going to be something that has like a frame.

32:24
Amanda Bickerstaff
But what we have here is what misconceptions about generative AI are driving people's use of these tools in ways that could be unsafe, ineffective or unethical. And I think that we see this. So there's two ways that you can think about the misconceptions. One is how Genai works and then one is where it falls short versus where we think it's going to be valuable. And so the case of this here, a confident, polished and thorough Gen AI response is a good sign of accuracy is actually the exact opposite. In fact, there is research that shows that longer outputs have more inaccuracies than shorter outputs on average. And so you actually, the longer it is, the more likely it will have a hallucination or an accuracy.

33:11
Amanda Bickerstaff
That's a crazy thing to tell people because it seems our confirmation bias says, oh, it looks like the thing, it must be good. And actually it's the exact opposite. This is where you can also, things like Genai thinks and reasons like humans can learn from my chats in real time. No learning happens in generative AI systems after they're released. So what I hear from people I taught ChatGPT, that's not how this works. And I think that one of the things that I love nneka is that we've been doing the C Framework presentation where we ask people to go through this list of misconceptions, like school leaders or teachers picking out which ones they think are most important to address. And out of these, 13, eight or nine will be picked by a room. There's not just one.

33:58
Amanda Bickerstaff
It is a, it is a complex piece, but when you start to see it, you start to see these, these, like very strongly held beliefs about gender of AI. Even in this short period of time, it starts to create this space, this distance, this understanding that these tools are going to be significantly less reliable than the way they look like. And I think that, I mean, do you have a favorite, not favorite misconception, what you're the one you think is most important for us to focus on.

34:26
Nneka McGee
Oh, that AI does not think, it does not reason, it is not sentient, that it thinks and reasons like a human. It is a predictive engine, a probability engine. I can't think of the author's name. They call it Mathy Maths. So I think that's the most important thing because it goes back to the Sick of Fancy. Fancy. If you think you know it has a name, it's talking to you. Hello, friend. How are you doing today? How can I help you? Understanding that there is no feeling or reasoning behind any of those outputs goes a long way in the interaction that you have with the tool.

35:13
Amanda Bickerstaff
Absolutely. So this idea of Sick of Fancy, if it agrees with me, it must be, I must be right. If it agrees. The very easy way to do this is to go to ChatGPT, ask for analysis of something, and say, you're wrong. And watch it go, oh, Amanda, you're so right, you're so smart, and I make a joke of that. But this has led to psychosis where if someone is hearing constantly that they're right and everyone else is wrong, this can lead to that, but it also can lead to emotional independence. So if you're thinking about this is where we start moving into things you can bring back. What I would suggest is something as simple as this.

35:51
Amanda Bickerstaff
Bringing this into a staff room, a classroom, and having people identify which of these misconceptions that they hold and then talking about how that could impact their lives. We're thinking about building a little bit of, like a diagnostic too, where Tana and her team has been doing a really good job of creating some of these interactives. But this is something that I think you will see people's brains start to go, oh, my gosh, there's so much more I need to know. But also it's so tangible. Like, it's so tangible. It's like, not crazy.

36:27
Amanda Bickerstaff
And this is why I think we've been so successful is that what you'll notice is that we don't have any tips or tricks, activities on the web, but what we have instead are these examples of places in which you can start to address misconceptions, start to reroute the ways in which we've learned how to use these tools. But I think that this is where it gets really exciting because you can see people's mindsets change in real time once they have these knowledges in place. Then the second resource we have for everybody is an example of where a knowledge base can lead to a different practice. And so Dan's going to drop in. Our newest, our second to newest, our misconceptions is our newest resource. But this is our. This is actually Vic is on the call. Vic, say hi.

37:10
Amanda Bickerstaff
Like, Vic is in the chat already sharing cool stuff. But Vic came to us. And by the way, if anybody ever wants to do this, we've done this with Vera, we've done this now with Vic. If you have a great resource, we always like to make it, you know, to try to help improve it based on what we see. But Vic came to us and said, I have this great resource. And what we did is we worked with him and he's here with us today. You can see his name on the bottom. But what we have here is that if you understand the predictive nature of the tools, then you will understand that AI detection can never work.

37:43
Amanda Bickerstaff
Especially when it's using generative AI to be an AI detector because then what you have is two predictive engines that have no ability to identify right from wrong or digital signatures trying to identify the level of AI. When we think about this is an example of how you can shift someone's mindset as a teacher away from. We can catch AI cheating too. We need to address AI cheating and offloading in a significant and thoughtful way. That includes things like talking to students, building AI literacy, relooking at the way that we assess. And those types of things are going to be way more reliable than turning to an AI detector, which not only makes mistakes, but can be biased against non aviation English speakers, those that love and I don't know, I feel bad. Daniel loves an EM dash.

38:34
Amanda Bickerstaff
He can never use an EM dash ever again. I keep making him remove it on every email because it looks like AI And Nneka, probably you like an EM dash too. But I mean, I think, Naiko, when you think about the work you've done, we just hear it so much. Teacher. There's like an arms race of like teachers using AI detectors, students using AI humanizers, and then like there's no human in that loop. It's just this kind of like, we're going to catch AI cheating, create a culture of distrust, and then kids are going to find a way around it if they're willing to.

39:04
Nneka McGee
Yeah. I think when you are using a tool that to determine if there have been violations of academic integrity or cheating or plagiarism to the extent that it can impact a student negatively, not just in real time, but for years to come. Why would you put yourself in that position as a teacher? Why would you put a student in that position particularly because even with the AI detectors they change. And I don't know if it was something that was mentioned on Slack, but I remember reading somewhere that now one of the detectors kind of tries to scan if you did copy and paste instead of type. I mean someone's going to get her get around that. I like to use the example and this is how critical it is. And you see this.

40:03
Nneka McGee
One of the things that people are doing now is uploading things that they've written years ago and it's showing, oh, it's 75%, it's 100% AI generated when there's no possibility. I used and used to use as an example when I put in a original work, it was 0% AI generated. All I did was ask for the simple how it would change a murder, make a recommendation or things of that nature. Nothing major, minor editing. And then it went from 0% to 13%. Right? Yep. And this is something that we would never. I was like, what if I was using a thesaurus? You know, we have to. Someone talked about balance earlier. We have to have a balance and if we want, we would never stake our claim on something that is that error prone.

41:01
Nneka McGee
And so I think this is such an important resource and I hope we can change minds about the use of these AI detectors because they are so unreliable. And I think this goes a long way in terms of helping build that knowledge base.

41:18
Amanda Bickerstaff
Absolutely. And I just want to show everybody because I think I'm going to go rogue again. Nneka, get ready. You're going to get to deal with this all the time. But I do want to share something that we share in our Safety Workshop that we've been doing. We started doing this with Lightspeed which has been around for 26 years and we don't usually align with tech companies but we just thought it was really interesting to be able to work on safety specifically and they gave us real data about how genai systems around the country. If you're outside the U.S. this is U.S. based and if you were in our Safety workshop, you've seen this before. But what it does, it looks like this is not individuals, this is in the thousands. I just want to be clear.

42:06
Amanda Bickerstaff
But what I want you to notice is that Quillbot is in the top 10 of tools used. That is a AI humanizer that's used to get around AI detection. And then what you have is Turnitin is now a AI detector. Then you've got GPT0. At AI detector zero. GPT is an AI humanizer. They're not the same. They're literally flipped. And one's a humanizer. You see scribbers. And I added Tector. You've got Humanize AI Pro. If you don't take this and see this arms race that is happening, this very destructive arms race of kids turning to AI humanizers and teachers training AI detectors as a way in which this is disconnected and how you can actually start to have better conversations, then I'm not sure what can convince you of that, because it is something that I think we knew it.

42:58
Amanda Bickerstaff
I don't know anecdotally, but to see it like this, to see just how many in the top 20 to 25 tools countrywide are just about cheating or catching cheating like that is like we are hearing stories of young people using AI humanizers on their human generated work because they are afraid of AI detection. So they are either dumbing down or humanizing their own work because of that fear. And so you can't even look at this as just kids doing this to cheat. They're doing this to actually protect potentially the risks associated with getting a failing grade or not being able to check themselves. The last thing, the last anecdote from recently is that this is where it really is going to get interesting is that a young person was accused of AI cheating. And they did.

43:47
Amanda Bickerstaff
They cheated with AI and she got a failing grade. And then what ended up happening is that the other students had also in the classroom used AI and did not get caught. What ended up happening is that the teacher now had this very difficult thing where the young person was of course upset that she got caught, but more upset that why was she caught when everyone else did not get caught and how it created this really difficult experience. And that teacher then spent the rest of that semester dealing with the fallout. And I think that's where we want to see these very different places. And what we would say is that this does not mean, if you're really interested, this does not mean pen and paper only. We're not suggesting that we have to reframe how we assess in this world.

44:37
Amanda Bickerstaff
And by just removing a technology is not going to be Sufficient. It needs to be rethought. Like, this is not a time to go back in time is a time to like, actually look at what we can do going forward and actually creating, like, you know, what's authentic, you know, what's going to get around an AI detector? Authentic assessment, reflection, discussion, collaboration, creativity. That's, you know what, these tools are really bad at that, especially with kids that have poor AI literacy and they don't even know how to use those tools. But this is where you're going to have to really rethink what we're doing. And exactly. Who wants to be. Who came into teaching because they wanted to be an AI cheating detector? Right. That takes all the humanity out of it. Okay, I'm off of this. This is. So we're down.

45:27
Amanda Bickerstaff
Okay, so here we go. We're going to start shifting into a couple other tips and techniques to take back. We find that when we think about the knowledge bases, right? Then these mindsets that are really important, the idea of how do you take the knowledge of how the systems work and it creates that critical distrust and then how do you start to be intentional? So there were some questions that some ideas in the chat about sustainability, like or cognition or creativity or voice. Do I really need to use this tool right now for this thing, or am I using it as a crutch? Or maybe I keep hitting the button over and over again to make it perfect instead of just pulling it out of the system and editing it.

46:13
Amanda Bickerstaff
And not only is that more sustainable, but it also is something in which it's a better. It gets a better use of the tool as the starting place and you are bringing the revision that final piece. If you do decide to use the tools, the best way to use the tools effectively is to come with as much information as possible. So meaning context, exemplars, ideas, so that when you use the tool, you're using it in the best way possible to get the highest quality output. What usually looks like is that it's passive. Naika has an idea, Nikka goes, Nikka gets answer. Nika's done. What we would say instead is that like, wait, I know this is Nneka. She has an idea. She thinks about what's the strategic way to do it.

47:00
Amanda Bickerstaff
And then you make a decision along the way of where it'll be you and where it'll be AI. And maybe that decision is 70, 30, maybe it's 2080, maybe it's zero. And I think that's what we really see as really effective. We also think that staying critical, we know that there's not a generic tool in the world that does not make mistakes, that does not have bias and outputs, that is not sycophantic. I mean, even when we try to make less sycophantic tools, they tend to get. They tend to be less like people like them, less so they get reverted back to that sycophantic approach. We want you to be transparent. I love in the chat. I love how the EM Dash conversation be transparent. Like, you know, in this case, maybe you do need to say, this is my.

47:44
Amanda Bickerstaff
You know, I love an EM Dash, and I'm not letting it go. Or in the case of. What we do is whenever we're using generative AI, it will say on the document, we use generative AI to do X, Y and Z. And we do that in everything that we do because I think it's incredibly important. And if you build transparency from the front, then you can help shift a culture to be able to allow that transparency to not just be assumed as a table stake, but also something that is valued. And the last thing that's to you around, act responsibly. Like, we know I can't. Nick, I'm gonna tell you a story. So I was in Wisconsin, and were doing a session, and two of the guys were two of the leaders.

48:32
Amanda Bickerstaff
One of them made a deep fake of his friend, who was a security guy. So he pulled his image offline in the session, went to Gemini and made him, like, on a horse. And as soon as he did it, he realized he was, like, so out of line. And. But the funny part was is that his friend, the security guy, saw it, so he made a deep fake of him. So he ended up this, like, crazy, like, these adults that were like, this is fun. And then I was in a training last week in New Jersey where another table of our lovely, like, leaders that were all men one, was about to do it. And I'm like, you gotta. You gotta stop, because it's gonna, like, end up being this wildfire thing.

49:12
Amanda Bickerstaff
And so I think that acting responsibly, some of it starts where it's, like, really fun. But what if one of those leaders or teachers or students is not well liked and is suddenly that becomes not a horse, but it becomes something that's, like, a little bit of, like, a bullying aspect. And I know that our, you know, educators are nice people, but we make mistakes, right? Or we don't know what our impact will be. And so just to watch adult.

49:41
Nneka McGee
Do.

49:41
Amanda Bickerstaff
These practices and then realize after, I think that if you look at a 12 year old, 14 year old, an 8 year old, it starts to be very important to have these conversations and then the last. Yeah, I mean, Nneka, do you want to respond?

49:53
Nneka McGee
No, I was agreeing. Yeah, it's so interesting to act responsibly and someone made a point about taking accountability and being responsible for their work. It's essential.

50:07
Amanda Bickerstaff
Yeah. And I think like sometimes we realize why like you need to create spaces in which people can make mistakes in safe ways because as soon as like that person's not going to do that again. And I think that's where you start to like have that and I think that goes to that. Keep learning. The keep learning aspect is I might have used this in a way that wasn't helpful or I used it in a way that was or I need to start trying other things or I'm over relying on it. Right. And I think that's going to be really important. And so you know, I think that's where you might find. So everyone's putting their like their disclosure statements in the chat, which is amazing.

50:46
Amanda Bickerstaff
But you might find the disclosure statement needs to be changed or you need to have rules about when you disclose and when, like how you disclose and that it's not just, you know, one way and that will be based on what happens after you disclose it. Right. You have to be able to be willing to get feedback and listen and change. And so we're going to do a couple of examples of what this could look like. But what we see here is that for us when we think about what that looks like is that we do think that there is a progression in terms of the developmental appropriateness of these tools. And so I think that in this case nneka, like you can maybe walk us through and we're going to hit time so we're going to do this relatively quickly.

51:25
Amanda Bickerstaff
But we'll make sure that this is included in the follow up. Yep. Thank you, daniel.

51:30
Nneka McGee
Oh yeah. I mean when we look at early elementary, I think we can agree there are really no instances where you can see a child directly interacting with the gen system. As you move on into that upper elementary, you will have supervised use with district, I would say district approved tools or family approved for homeschooling. Obviously as you move into middle school you're having more direct scaffolded engagement, particularly for 8th grade when they reach that 13 years of age or older. We're moving past the COPPA guidelines. There's more extended availability and use of AI tools available to them. Obviously, in high school, we can really get to that deep exploration. And this is where we start to see that interaction and how it's going to impact not only their college, but maybe career and the workforce as well.

52:20
Nneka McGee
And then as adults, full independent use and full independent responsibility as well for the outcomes of that use.

52:29
Amanda Bickerstaff
Yeah. And so, for example, I'm going to use Charleston because I know we have at least one person on here is that we like. So Magic student is only going to be available for fifth grade up. And so the district is not having. They have no Approved tools for K4. And we would say that this is. We're pretty conservative, but we wanted to give you a clear but doesn't mean that teachers can't model effective use, can't build awareness, can't talk about if students see this at home in my Snapchat, AI or Alexa as it starts to shift and change. This is where we would like you to start thinking about because I love watching that people like Shelley and others are starting to build these curriculum, especially since you've been around us. We've been in our orbit for a long time, Shelley.

53:17
Amanda Bickerstaff
But this is where you're doing that scaffolding. Even high schoolers. And I would say that adults. One of the cool things about Charleston is that the Magic school for teachers is going to have some of the tools unavailable and they're only going to get this when you actually have done an AI literacy course. So there's no just tool tool. It is now intention. It is connected. And so even though we have this here, that eight. You can't see my finger. Sorry, I was using my finger that 18 enough that adults also need to be brought through this. But you can't just start from here. They have to have that same support. And so this is in our C framework, which daniel will drop right now. And so you can see this. And then also this is our curriculum.

54:03
Amanda Bickerstaff
So we're writing our first ever full curriculum. And Nneka is helping with Tanner. And what we have is that the curriculum is going to be focused on the mindsets. So each lesson will be one of the mindsets. So be intentional, stay critical, be transparent, act responsibly. And what you can see is that what we do is that from second grade up, what we're doing is reinstalling and spiraling what this looks like at developmentally appropriate places. So, Nneka, you want to take us through a bit how that spiral is going to work.

54:33
Nneka McGee
Oh, definitely. So when you look, let's say with lesson one, what happens in lesson one and someone made important points about exposure. So you talked about the example of the K or you know, not actually using it in K through 4. That doesn't mean that awareness isn't necessary. So there you may have an awareness. If you're looking at that foundational knowledge about being intentional about that awareness as you move up into that progression. In terms of what can happen in elementary, maybe a teacher is modeling a tool in terms of being intentional about how generative AI shapes what we understand. In middle school, maybe students start to interact with that tool as they're deciding when and when not to use generative AI. And of course we're looking at that keep learning, which is important, where they're justifying their action.

55:25
Nneka McGee
What you see here with this spiral is the consistency in the approach across the grade level progressions, which is so essential when we're teaching something as important as gen AI literacy. Now that you have that consistency across the grade level approaches, what changes is the depth at which we cover and what the students experiences. Experience during a lesson?

55:51
Amanda Bickerstaff
Absolutely. I just want to make one point as well, is that just as a reminder, this is really about generative AI. And so when we talk about K2, we're talking about chatbots. We're not talking about potential essentially like an Amira learning or another tool that has some evidence that's using machine learning that is predictive in a way that it can be deterministic. You have the right answer, right? You have the prediction. But what we would say that this is really about generative AI. These are chatbots, These are things that cannot be fully reliable. They are things that are sycophantic, all of those risks and also create a relational like experience that a young student is not developmentally appropriate.

56:33
Amanda Bickerstaff
So we would say at this stage, and I feel confident and say that chatbots, generative A chatbots are not developmentally appropriate based on the research and what we have seen and how the tool works for a second grader. And I think that makes us difference. And if you want to send us an email, you can. But we just believe that very strongly. And I think that sometimes we have to be conservative with a new technology especially we've started to see some of these issues not only happen in schools but in adults where we're seeing the same issues. Last couple minutes. We want to leave with a couple of really great pieces here for you to think about. We've got our student use guide which is something we've had. It's crazy. This has been since 2024.

57:15
Amanda Bickerstaff
But this is a really simple way to start building students thinking around when and how they should use those tool. So we definitely say check that out. We have, you can create this, you can feel free to adapt this. We always say feel free to adapt as long as you are referencing our work. I think for adults, especially right now with people saying that when and when not how are we certain we should use these tools? Our every framework is for everyone and it talks about responsibility when and how we should use those tools. We would say that human judgment that needs to be centered and voice is a place in which you probably should not be using general AI even as an adult.

57:53
Amanda Bickerstaff
But what we would say is that the idea of really owning that and this is something that came up a couple times today is having that along the way. And then we're going to wrap with a couple of quick hit. We'll go probably three minutes over. This is a big one. But we're going to go through some practical guidance and if you can't say we know you have to go but we're going to spend two extra minutes, we'll make sure it's in the recording and followed up as a resource. But we're going to go through some of this practical guidance. And so I'll do the First T&AKA if you want to do the second two. So what we say that is within one day is just what are the guidelines and do they exist and what's available.

58:30
Amanda Bickerstaff
They don't exist is a great way to ask for them in a very nice and thoughtful way. And then what we suggest is experiment with at least one teacher facing AI Chatbot. You can use our prompt library and then reflect, remember, keep learning, reflect on your experience. And I would say that one of the things we love about this experiment, try something you haven't tried before. Like try to create a diagram, try to create a vibe coding. Get a little, get a little crazy. You can read our C framework which I think is, it is big but it's very readable and then also create that short term goal. But we would think what's really interesting is you don't, I mean like you can build GEN literacy into a lesson. You can just model how you use generative AI, what you learned or didn't learn.

59:13
Amanda Bickerstaff
You can talk with your colleagues. You can stay up to date with actually having that extensive learning. And if you're a leader what we would suggest if I go the right way is that it's the same thing. Do not have if you have guidelines because we know that some of our leaders do not know what's available, if not advocate for them. But we would suggest is if you. To Nneka's point, if you don't have a school committee, an AI committee or AI leads, create one. We also suggest a C framework. But then start to have a goal for yourself and then think about how you're going to have that for your teachers. So thinking about also how you can build this in. In six months to maybe it's a family night, which Nick is going to talk about in a more a moment.

59:51
Amanda Bickerstaff
You're going to actually look at. We showed you the lightspeed data. Can you look and see what people are actually using so you can have better guidelines to that. And then finally we're going to look at our community and families. So Nick, you want to take us with our families?

01:00:03
Nneka McGee
Sure. Just real quick with students again, sometimes we need to understand particularly because a code of conduct we just, we had great conversations around academic integrity and what the implications for those students can also set goals because it's important to increase your understanding. And then for students, it's also important to reflect once they learn about how these tools can influence them, always go into reflection mode to keep that muscle strong of, you know, the discernment that needs to take place when we're interacting with AI tools.

01:00:37
Amanda Bickerstaff
Absolutely. And I would say also this is where like if you have students, they have responsibility too. And especially if you're willing to do gen AI literacy with them, it changes the chemistry, so to speak, that kids. This is not. Well, we didn't talk about it now becomes if you make these decisions about academic integrity or cognitive offloading, then that's something you're doing like you're doing with intention. And if you do get caught or if it does have an impact, that is something that you're going to have to understand is not going to end in this classroom, but could potentially go with you the rest of your career. Sorry to be down, but I think that real talk right now is going to be very important for especially our high school students.

01:01:20
Nneka McGee
Yes. And then for parents and caregivers. And you see within one day because one of the biggest questions I'm asked and I think you are as well, Amanda's like, what do we do next? So we wanted to include like, you can do this in the call and you can do this. Ask your children about their current understanding of AI, you would be surprised at what you hear. But that can really guide where you need to go again. Within 30 days, some of you is back to school. Some states, for example, like Ohio, every district had to implement an AI policy before the school year began. And so within 30 days, because you may still have districts working through four families register for a Gen AI literacy course.

01:02:03
Nneka McGee
It's important for parents and caregivers to know as well, continue to have those regular conversations because things change, particularly with generative AI tools and how students use them. And so that's important. And because of where we are with generative AI, chances are that your school or institution may have some type of learning opportunity. So it's important to take advantage of that as well.

01:02:30
Amanda Bickerstaff
And we have this great, if you're, you know, a lot of us are practitioners and leaders, we have this great guide that's a parent guide that could be a way for you to engage those per those parents. And then we'll end, I'll take this one of the community, bring people in. This could be adults, this could be seniors, this could be industry. We love bringing people in. And what we say is that you can pull, you can have a big AI literacy day that have a industry panel, could have a opportunity where kids get to try out new tools. You can have something where like, not only do you connect with that, but also I wouldn't under emphasize that a lot of your community members also probably don't have strong AI literacy. So it's an opportunity to do this together.

01:03:09
Amanda Bickerstaff
And so everybody, we did it. This I. This is the Think Naked, the first time we've ever gone over in terms of a webinar. But we do think it's such an important one. We have a lot of great webinars coming up, including with Ames and we're going to do one with out in Colorado with the NSF foundation, the NSF grant that's focused on AI literacy. But we always have these great resources. I want to say thank you to everyone for joining us. If you started school, good luck. We're sending you all the greatest vibes. If you're about to get started or you have a little bit of time left, I hope you sleep well and we just appreciate everybody and thank you, Nneka, for joining and hope everyone has a wonderful rest of your day. Thanks everyone.

  Transcribed by https://fireflies.ai/