(SPEECH) LISA MEDNICK TAKAMI: Takami, special project director at CAEP TAP at NOCE. And I'm joined this morning by Chandni Ajanel, Jaspinder Uppal, Diana Martinez, and Ute Maschke. And I am now going to turn it over to Dulce Delgadillo. (DESCRIPTION) Presentation. (SPEECH) DULCE DELGADILLO: Hi, everyone. Happy Tuesday. Dulce Delgadillo. I'm the director of research at NOCE. I'm proud to be part of the CC TAP team here at NOCE. Happy to see all of you. I do just want to take some time and acknowledge our state leadership at both the California Department of Education and the California Community Colleges Chancellor's Office. They are the engines behind this work and supporting both sides of the house in being able to build out and best support our adult learners. In addition, I also want to recognize our partners at Sacramento County Office of Education. They are our very esteemed colleagues over on that side that have been doing this work for many years. And we are happy to continue to do this work in partnership with them, to be able to best support the field as well. All right. And yes, in the welcome, I see that you are already following instructions. So if in the chat, you could just take some time and please introduce yourself and introduce your affiliation, either your consortia or your college or your institution or your school, however you want to show your affiliation. So I'll go ahead and hand it over to Chandni. (DESCRIPTION) Slide: Housekeeping. (SPEECH) PRESENTER: Hi, everyone. Just wanted to go over a few housekeeping details. First, I'll let you know that this meeting is being recorded. This recording and the PowerPoint you're viewing right now will be released on the Cal Adult Ed website following its remediation. And we ask that you please fill out the survey at the end. We really value your feedback, and we use it to improve our webinars. (DESCRIPTION) Slide: Gracious Space. (SPEECH) Additionally, we just want to give a gentle reminder that we want to cultivate a gracious space for lively, respectful, and professional discussions. Thank you. (DESCRIPTION) Slide: Objectives. (SPEECH) LISA MEDNICK TAKAMI: Thank you, Chandni. Well, good morning again everyone. We have a wonderful program planned for you this morning. And as always, we appreciated those who filled out the pre-webinar survey on whether you're using AI already in the course of CAEP work, or what you might want to use AI for. And during the Q&A session, we will give an opportunity for those who not only have questions, but we're hoping from those colleagues who are already using AI to share among colleagues how you are using it. So I'll just go over our objectives for this morning. We are really looking to create a space where we're providing practical applications of AI for both instructional use in person or virtual, as well as task automation. Demonstrate the use of AI as a complementary tool for researching and resolving technical assistance questions, and really following technological trends and the use of AI in adult education. (DESCRIPTION) Slide: Agenda. (SPEECH) Our agenda for this morning features two wonderful speakers from NOCE. First, Professor Afraim Sedrak, who is faculty for business education and computer information systems and technology here at North Orange Continuing Education. And Afraim will be talking about AI and the way it works, AI bias and the way it happens, and ways to mitigate bias in AI. And our own Diana Martinez from the CAEP TAP at NOCE team, who joined us for part one, will also be talking about using ChatGPT this time for technical assistance requests. We will have an opportunity for questions and discussion. And then our closing activities. So with that, we welcome Professor Afraim Sedrak. (DESCRIPTION) Slide: The Era of AI. (SPEECH) And I will stop sharing so that he can share. AFRAIM SEDRAK: Hello. Good morning, everyone. And thank you for having me. Let me share my screen. (DESCRIPTION) A man with short dark hair and a beard wears a black shirt. He sits in front of a virtual background depicting a modern office. (SPEECH) Please let me know if you see my screen. (DESCRIPTION) Presentation: Artificial intelligence. What is it and how does it work? (SPEECH) LISA MEDNICK TAKAMI: Yes, we can see it, Afraim. Thank you. AFRAIM SEDRAK: OK. Very good. OK. So as you mentioned, the goal for at least this part of the presentation is to discuss how artificial intelligence work. And this is something very important for anyone who's living in this era right now, to be able to have an understanding of what's inside the box. Because most of us, we are using the AI models in different ways, but it's important for us to understand how it's working from inside. It's kind of technical a little bit. I will try to simplify it to the best that we can so we can have this understanding. So let's move on. These are the things that we will cover. So this is the era of artificial intelligence. Now, let's ask ourselves this question, what is artificial intelligence? Most of the people, they either do not have clear understanding of artificial intelligence or maybe a misunderstanding of artificial intelligence. So the simplest manner to identify this-- artificial intelligence is a computer software, a computer program. But this computer program has been evolved in a way that allows the computer to act like humans in so many different manners. And this is where we go to the next slide. (DESCRIPTION) Slide. (SPEECH) Now, we will see that the artificial intelligence is the big umbrella. It's a branch under the computer science knowledge and department. So the artificial intelligence, it's a software that makes it possible for machines to learn from experience, adjust to new inputs, and perform human-like tasks. And when we look into this definition, we will see that there are some new aspects that we usually didn't see with the normal software. So this machine learning, or actually artificial intelligence, learn from experience. Previous softwares that have been developed prior to the AI era, they cannot learn from experience. And also, this AI system can adjust to new inputs. When we say new inputs, we mean that inputs that the AI has never seen before. And still, it can adjust to those. And by the end, it can perform human-like tasks, tasks only humans can do. So for example, drawing a picture, creating a musical note, things like that. So these are tasks that we know that requires creativity. And usually, humans are the only one who's doing that. The next branch under the artificial intelligence is machine learning. And this is a very profound part of the artificial intelligence system. It's practically allowing the computer systems that uses artificial intelligence to learn data on its own, without we feeding in this data to the computer. So we are not feeding the data into the computer anymore. The system learn from the data that we're feeding into it, and creates new rules or new algorithms out of that. So that's the machine learning. And if you think about it, if you think about a baby or a toddler, they learn from what's surrounding them. And the machine learning is something similar to that. That allows the artificial intelligence system to learn from the environment, from what's surrounding them. Then we go to the deep learning. So the deep learning is a type of machine learning. It's like a branch under the machine learning that trains the computer to perform human-like tasks, such as recognizing speeches, identifying images, and making predictions. So you can think of it as the application part of the machine learning. And then we have the neural language processing, which most of the people confuse this as AI. So when we say ChatGPT or most of the AI models that we deal with, it's mainly a natural language processing. We call it NLP. So the natural language processing is what you see when you deal with, for example, ChatGPT or Copilot. Those are artificial intelligence models that allow the computer to interact in a natural language. It has a lot of inputs inside it. And it can deal with you in a way that most of the people, they might not be able to differentiate whether I'm chatting with a human being or a computer system. And this is where the NLP came to play. And then the last one that we will cover is about computer vision. And this computer vision part of the artificial intelligence system is what allows the AI system to be able to interpret visually an object that it can see, mainly a picture. Or if it's a camera, this camera can actually interpret what are the objects that this camera can see. So these are different branches within the artificial intelligence. And as we see, artificial intelligence is actually a very big umbrella that covers so many things. So to have just an understanding, this is just touch base on what we have under the AI technology. (DESCRIPTION) Slide. (SPEECH) Part of reflecting for AI. Most of us are using AI already, and most of us have been using it for a very long time. But probably, we didn't know that we were using AI. So for example, we have smartphones. If your smartphones can unlock using your face recognition, guess what? This is AI. When you are using voice-enabled assistant, that's another model for AI. Photo enhancement, that's another model for AI. Some of us, we have photo albums to the point that you can select one individual in one picture, and you can ask your phone to list all the pictures that this individual exists in. This is something that works with AI. So you have AI already, and you've been using it for quite a long time, but maybe we didn't recognize it as AI. Also, in social media. You have image filter. You have recommended contents, that it's being pushed to you, depending on your interest. And this interest could be something that you've been talking about, a message that you've been texting, a search that you've been doing in the past. And then you see most of the feeds coming to cover the interest that you have. This is how AI is reflected in social media. E-commerce for the personalized ads. And this is, I believe, obvious. But by this time right now, for all of us, that whenever you search for something, especially to buy, you see a lot of advertisements about that specific thing. So this is something that is using AI. Also, for navigation and autopilot. So if you're using navigation in your car, specifically the new autonomous driving cars, the car is being driven using a lot of technologies that involves AI. So it can trace the traffic condition, the traffic routes. The autopilot feature on its own is a major advancement for AI. Also, the weather forecast. The weather forecast depends a lot on AI analysis. As we have seen in the previous slide, it can predict patterns. And this is what the weather forecast do, using the AI to allow us to predict for, maybe, seven or 10 days in the future. This is using this feature of artificial intelligence. For health care and diagnostics as well. It helps a lot-- the physicians and doctors-- to diagnose in so many ways. So still, you are using AI even if you don't know it. As personal gadgets, you have blood pressure, EKG, sleeping pattern, high sounds. Like maybe in your watch, you see an alert that says that your environment has a very high sound and this is dangerous for your ears. This is happening using artificial intelligence. Motion tracking as well. So some camera system and surveillance system, they can track, for example, someone walking without a mask. Someone was walking and then start running. Someone was walking, carrying a bag, and then they left this bag. And then they continue walking. So maybe in airports or things like that. So this is a model that uses artificial intelligence, the vision part of it, to identify specific pattern and give us warning about them. So think about it from the other perspective. If we don't have that AI system, how many human individual you would need to give us this amount of monitoring or this level of accuracy in monitoring to the point of-- expectation, we can say that it's impossible. But when we incorporate AI, now we can have a very close, accurate level of monitoring things. For the financial application as well. If you have received any alert from your bank account, for example, telling you that there was an action or a transaction using your credit card and they think that this is fraudulent activity, this is actually happening through AI. It analyze everything and try to catch the patterns that doesn't match your regular activities. And then it warns you about it. If that's you, OK, no harm is done. If that's not you, so being able to catch it using AI. So these are already things that we've been using within the AI system in our daily life. Now, what is the AI in general? This picture is not just an artistic picture. It's a picture to reflect something important inside the human brain. The human brain has two main components when it comes to thinking and decision-making, neurons and synapses. So neurons are the glowing parts that you see in the human brain. And we have a lot of them in our human brains. And then we have the synapses. Those are the lines that connects those neurons together. This is the human brain. So building the artificial intelligence was something that is built according to the human brain and how it works. This is how we came to the artificial intelligence scientific part of it. (DESCRIPTION) Slide. (SPEECH) So let's see. You know what? Before we do that-- OK, now let's see this part, neural network. In this picture, I'm trying to simplify it heavily. So don't take it literally. It's just a simplification of what's the neural network and how it works. So the neural network represents of input layer. This is the part that you see, those blue circles, and then hidden layers which represent the neurons and how it's thinking in decision-making. And then we have the output layer, which give us the decision. So I'm trying to explain it to see how, for example, the visual AI part work. So let's think about this. We have a picture of a pet. And we want the AI system to identify this picture and tell us what kind of animal is that. So we will assume that we will put two rules. One of them is, this pet has, for example, four legs and it has a tail. So the first input layer, it will detect whether we have four legs or not. And the second one will detect if this animal has a tail or not. And from there, these neurons, they fire through the synapses. These are the lines to the next layer that examine an additional things to make sure that this is what we want. And those on their own, they fire as well. And they give us the output layer, which tell us, OK, this is a cat, because the condition that we're looking for has been matched. So it has four legs and a tail. But it's not as simple as this. Because if you think about it with this picture with a crocodile, do we have an animal that has four legs and the tail? Yes. So if this picture is feeded into the artificial intelligence system, according to those two neurons or two new input layers, it will be difficult to distinguish between the cat and the crocodile because we have little number of sensors, if it's OK to say that, or the input points that we can test for. (DESCRIPTION) An elephant image. (SPEECH) Think about it. If it's an elephant, it would be the same. If it's a dog, it would be the same. So now, when you see the full picture, you see that we have to add so many additional layers in order for us to be able to test whether this is a cat or not. (DESCRIPTION) Slide. (SPEECH) This takes us to the next one. So this slide is more of what it's looking like in the neural network. So when we have a cat, this cat has the four legs and the tail. And we have so many input points to discuss for so many things. The shape of the ears, the shape of the nose, the shape of the eyes, the color, the softness of the fur, things like this. So we have so many things that is being tested in these neurons. And every one of those nodes-- this is what we call them in the computer science. And every one of those nodes, it tests for something, and then it fires the results to every node in the following layer. And usually, all the layers that we have in between, they are hidden. So you don't see them. So every one of those nodes, it fires to every one of the nodes in the following layer, with the output of what it thinks that the question could be answered. And on its turn, the hidden layer 1 do the same, examine for other things, and then go to layer 2. And then layer 2 examine for other things, and then go to layer 3 and so on, until we get to the output layer. And now, don't be mistaken. This is so much simplified. This is not how it is in real life. It's way more complicated than that. I'm just simplifying it for the sake of understanding. So when it comes to the actual neural network, we are not talking about 10 or maybe 20 nodes. We are talking about so many of them. And when it comes to the hidden layers, we have so many of them as well. So we're not talking about three or five or even a hundred. We have way more complicated than that. And then we get the output layer. So this is helping us to do something like this. Now, I have this picture. Would this be interpreted as a cat? If we think about all the consideration that we've been thinking about to identify this animal as a cat, we would see that this would match this one as well. So we expect the output for those two to be a cat. But how about this? (DESCRIPTION) Upside down cat. (SPEECH) Now, we see that this is completely different. Yet the vision using the AI system will be able to still identify this one as a cat. And this is how complicated the system is, because it tests for so many aspects, more than what we just covered in our discussion right now, to the point that the head is upside down. You don't see any tail. You don't see any legs, or even the shape of the ears is not clear. But still, the AI system can identify this as a cat. So to give us better understanding how this is happening, we see that in the artificial intelligence system, we feed data to the system. And this is what we see on the left-hand side here. And specifically, to be accurate, it's millions or billions of records of data that is being fed into the artificial intelligence system. And we allow the artificial intelligence system to process these inputs. And then it gives us an output. So this processing is happening through very complicated algorithms that we build into the AI system. And those algorithms, when they are processing the data that we feed, it give us the label for what we are looking for. So we have so many items or so many styles of AI systems. One of them, as we call it, is guided learning, which we give the data and we give the label. So in this way, we teach the system how to identify them. And after so many attempts, now the system, like a small child, started to recognize things. Some systems as well, we don't give it the label. We just give it penalties inside the algorithms. When it's not correctly finding the correct animal, it gets a penalty. So it knows that the answer, for example, from this node, is not accurate. Then it lowers the weight, and it tries to get to the correct answer as well. So this is how the process is happening-- for example, the visual part of the AI-- behind the scene. Now, when we think from this perspective, could there be any bias created by the system, or could there be any wrong output that comes from the system? Definitely yes. So for example, if we don't feed huge amount of data to the system, now the system didn't learn everything. It learns just what we feed to the system. So that could create some bias. When we feed the system and try to train the AI system on limited amount of data, that's actually limiting what the system can deal with in the future. Also, the algorithms that we use to judge the data that is being processed to give us the label. Those algorithms, those are the things that we build. If we build them in a way that it's inclusive and include all the expected outputs, then the system will be comprehensive. But if it's poor algorithms, then we expect the data that will come out of that will not be that accurate. And at the end, we need to feed the label. When we have the guided learning model, we need to feed the correct label. So it's complicated to the point that here, all of those images are cats. But the label for the last one, the third one that we see on the bottom, it should be cat, upside down, head only. Something similar to that. So we give the AI system a more descriptive information to allow the system to give us more accurate recognition using the visual system. So this is how it works. Now, this part is to help us to see where are we from AI perspective. (DESCRIPTION) Slide: Humans from AI Perspective. (SPEECH) I will use an upside down triangle of authority. So the first part would be the user. Who is the user? The user is someone who is aware of the AI and uses the AI intentionally to better do the task, or do the tasks more effectively. So that would be a user. Comes into the next level, operator. This is someone who is not only aware about the AI system, who is not only using them to do the task more effectively. This is someone who is actually automating tasks, thinking of any tasks that I do or others are doing. And I take those manual tasks and put them in an automated way where I utilize the AI power of thinking, AI power of processing, depending on which module that you are using. And those people who falls under this layer, they have, for sure, more control than those who falls under the users layer. And then we have the developer layer. And this is the maximum level of power when it comes concerning the AI. And this is literally those who create the AI models, or maybe a little bit of those who train the AI models. So those, for sure, they have the upper hand in everything. Now, it's time for every one of us to ask ourselves, which layer I'm in right now. So just have maybe 10 seconds and think of which layer you are in. And unfortunately, I don't wish for any of you to be in the fourth one. But the fourth one is not a layer. It's outside of the control system. We call it consumer. So if you are not a developer, you might be an operator. If you're not an operator, you might be a user. But if you are none of these, then the rest of the human beings fall under the category of consumers, which have no control over the AI. Probably, they have no awareness of the AI, but they are using it. So when you unlock your phone with your picture, you're using AI. But if you are not aware of this and you're not using it to make your tests more effectively, then we fall under the consumer section. And from the consumer perspective, those are the people who need to worry about whether they are keeping up with the advancements of technology or not. And this is where our role as an educator and as an educational institution come to play a very important role, at least to put our students under the users level, maybe into the operator level. And if it could be possible, this will be under the developer level, which will give us, as a nation, a high level of advancements as well. (DESCRIPTION) Slide: A must have AI tools. (SPEECH) These are just for references, because I was talking in the previous slide about being a user or more. So if I'm not aware about the AI, then I'm consumer. But if I'm aware about it and I'm using it to do my task in a better way, in a more effective way, then I'm moving more from consumer to be a user. So now, if I'm using Microsoft Copilot or ChatGPT to do things in a better way, that's a good thing. To write new things, I can use Writesonic, or still ChatGPT and Copilot. To generate art, I can use Midjourney. So these are tools-- or if it's OK to say it in scientific manner, this is our models-- that could be used to do these activities. So if I want to generate art, I can use Midjourney. If I want to generate code, Replit for programmers. If I want to generate videos, I can use Synthesia. If I want to generate music, I can use the Soundraw. If I want to create PowerPoint, SlidesAI. Edit picture, Remini. Summarize notes, Wordtune. And so on. So what do you think of this slide? My perspective is, this should be your starting note that you keep for yourself. This is your toolbox. And this toolbox will grow and will continue to grow, I want to say, on a weekly basis. You will figure out new tools, new models that come from AI, and you can adapt those tools into your library. So the more tools you have and the more tools that you are educated about, and you are using them, the more effective you're going to be in doing your tasks. And think about it this way. If someone is unaware of all what we're talking about, and they are competing with someone who is equipped with an arsenal full of these tools, which one do you expect is going to survive and flourish and be better? And this is what we do. As an educational institute, we should equip our students with these tools so they are not obsolete the moment they graduate. If they don't have any knowledge about AI, they're kind of obsolete the moment they graduate. And then they need to study again. (DESCRIPTION) Slide. (SPEECH) Then those two are images that are generated using AI. It's just to reflect on what do you think about this, someone who doesn't know, doesn't utilize AI at all and someone who is using AI. So I leave this to your imagination, and think about it. And if you have any question, I believe that would be at the end of the presentation. Thank you all. (DESCRIPTION) Images: A man with wavy hair and a beard wears an apron and carries wooden planks in a woodworking setting. An older man sits on a wooden path in a desert, watching a businessman ascend into a futuristic spaceship near a city. Slide: The Era of AI. What is artificial intelligence? Flips through the slides quickly. (SPEECH) LISA MEDNICK TAKAMI: If things are coming to your mind in terms of what Afraim presented-- thank you so much, Afraim. I didn't realize I was muted. Please do put them in the chat, and we'll just move ourselves along here and introduce you to Diana Martinez. We'll be doing her session as soon as I can get to us. And here we are. Afraim, again, thank you so much. And now, I'll pass it over to Diana. DIANA MARTINEZ: Great. Thank you. And Lisa, could you go back a slide, please? LISA MEDNICK TAKAMI: Yes, I certainly can. Sorry about that. DIANA MARTINEZ: Perfect. Thank you. So hi, everyone. Again, my name is Diana. And thank you for tuning in. As a brief review from our previous webinar, these are the keynotes we want you to take away. It's completely fine if you didn't attend your last webinar. Again, this is something we're trying to build up to possibly create a long-term series. But throughout the series, we will continuously emphasize that AI is enhancing, not replacing, human expertise. In our case, since we'll be referring to technical assistance requests or questions, AI can empower self-service for simple queries. It will empower, not replace, humans to solve TA request. And AI can reduce turnaround time for technical assistance requests and assistance. Majority of TA requests received concern allowable expenses and fiscal matters. And AI can just help speed this process. Next slide, please. (DESCRIPTION) Slide: AI Platform Overview. (SPEECH) Thank you. So during our last webinar, we discussed Perplexity and the features it offers to help solve technical assistance requests. Due to continuous mergers and acquisitions, OpenAI has become the frontline platform for AI. And that's why today, we'll be discussing ChatGPT. It's a product developed by OpenAI, and it's arguably the most accessible. There's so many tutorials and outside resources for ChatGPT. So we really believe it's valuable for everyone to at least be familiar with it. And for our demo later today, we'll be showcasing the free version of ChatGPT. The premium version offers faster response times and early access to new features. And the free version will be sufficient for general inquiries and tasks. Next slide, please. (DESCRIPTION) Slide. (SPEECH) Thank you. So the first step, regardless of the platform you use, is to gather necessary files to serve as resources for CAEP TAP at NOCE. We primarily reference the CAEP fiscal guidance documents, such as the fiscal management guide, program guidance, and key memos like AB1491. We mainly receive fiscal technical assistance requests. So these documents are our most used resources. Again, if anyone needs access to these documents, they're all on the CAEP website, which is right over here. And you'll be able to access it later after these slides have been remediated. However, it's equally important for your consortium members to incorporate their own local guidance and resources, especially when looking at your request that involve more local involvement. And using your specific materials ensures that the responses are tailored to your specific context. We at CC TAP, or CAEP TAP at NOCE, use the short form template file as a documentation file. Again, the responses and formatting from ChatGPT vary on what exactly your request is. And we always include this document for formatting purposes. And once the documents are uploaded, organizing and labeling chats is important for easy reference later. We'll show you in the demo how to upload these documents to ensure that they're well-organized within the platform. Next slide, please. (DESCRIPTION) Slide: Handling a TA request. (SPEECH) So a common request that we receive is, X College is asking about the indirect rate for CAEP funds. I, the requester, understand that it's the lesser of 5% of their district approved rate. They said they don't have an approved indirect cost rate by the state. The list I have only shows district schools and COE. If they don't have a state approved rate, can they take the 5%? We mainly receive requests like these at the beginning of the school year, semester, or quarter. And we'll be using this TA request as an example to show in our demo. And again, these slides will be shared after they have been remediated. But it'll contain a brief step-by-step guide to follow if you want to try this out in the future. And again, CAEP TAP is here to support you. So if you have any questions regarding the demo or the use of AI to help support TA request, please contact us. Next slide, please. (DESCRIPTION) Slide. (SPEECH) So again, live demo. I'll go over the step-by-step briefly. Next slide, please. (DESCRIPTION) Slide. (SPEECH) So again, uploading documents to ChatGPT. These documents are all accessible on the CAEP website. It's best to upload documents directly and renaming the chat to make it easy to reference later. Next slide, please. (DESCRIPTION) Slide. (SPEECH) And create a new chat and reference stored memory. It's going to be specific. Reference relevant resources. And the more specific and resource-oriented your prompt is, the more accurate and actionable the response will be. Next slide, please. (DESCRIPTION) Slide: Review. (SPEECH) And again, just review for accuracy. This is something you should do with all AI inputs. And for our case, because we know how difficult it is to manage these large documents, we always recommend you to come to CAEP TAP to support you afterwards. If something doesn't look accurate, or if you just need additional guidance or confirmation of the validity of the response, we're here to support you. So I'll go ahead and share my screen now. (DESCRIPTION) Closes slides. Diana Martinez has long dark hair and speaks in a video call with a blue virtual background displaying the NOCE logo. Text: NOCE, NORTH ORANGE CONTINUING EDUCATION (SPEECH) OK. I hope everyone can see. (DESCRIPTION) Screenshare. (SPEECH) So after logging in and creating your own account, you can go ahead and create a new chat. And the first step would be to upload your documents. As you see here, I have the short form request template to help with formatting. And then we have the CAEP fiscal management guide, which is updated. And again, the most recent updates will be on the CAEP website. So if there are any changes or any new memos, please check out the CAEP website. And CAEP AB1491 guidance memo and the AB104 program guidance. So here I said, hi, ChatGPT. Please use these documents as a resource for future queries I will send later. As of now, please do not summarize any of these documents. Simply store their information. And if you do just upload these documents, it will store the memory. The only problem is that it may give you a huge summary of each of the documents, or give you some response that you may not need or want. And it'll just take up more memory. As you can see here, my memory is full for ChatGPT. I've been using it a lot for research and the responses. These documents have been uploaded and stored for future reference. And then after uploading these documents, it'll generate a relevant but random chat name. And here, you can just rename it. I chose "CAEP Files" just because I found it to be more relevant in our case. And then I will go ahead and input the request that we'll be discussing today. So creating a new chat. And here, I put in, hi, ChatGPT. Please resolve the following TA request referencing our previous chat containing important resources, CAEP Files. This is the name of the chat. I have worked on here. And then follow the short form request template for formatting, which can, again, be found in CAEP Files. And then here, I copy and pasted the request. (DESCRIPTION) Hovers over a paperclip icon. (SPEECH) You can see here that you can attach more files. But because I'm using the free version, I have a limit as to how many files I can upload. And just for time's sake, I uploaded them ahead of time. And then you could also search the web. However, you do have a limit as to how many times you can use this feature in the free version of the platform. So I'll go ahead and send this. (DESCRIPTION) ChatGPT output. (SPEECH) So following the short form request template, it shows me here the requester name and institution. For the example, we used X College. And then it just really repeats the request. X College is inquiring about the allowable indirect cost rate for CAEP funds. Again, background. The college is aware that the allowable indirect cost rate is the lesser of the 5% or their district approved rate. And then here, we have the next steps. Confirm with the college that they have explored any possibility of obtaining a state-approved indirect cost rate. And then provide clarification on the CAEP fiscal management guide regarding the 5%. And then here, it does offer a solution. However, AI isn't perfect, and it mostly just reference and summarize what's already existing in the CAEP fiscal management guide. If you need more guidance or if this just doesn't make sense, you can go ahead and ask it to explain it in more digestible terms, or explain it more step by step. But again, you can go and come to us at CAEP TAP to help support you with this request. And AI is constantly changing. We are looking into possibly creating a GPT, which will help us with focusing on just CAEP resources, and making easier to share these files and resources. We can share this chat. However, it's not always accessible, especially with the free version. And I believe that is all. Thank you. I'll stop sharing for now. And then I'll go ahead and recap the benefits briefly. (DESCRIPTION) Opens slides again. Text: Review the response for your accuracy. Slide: benefits recap. (SPEECH) Again, benefits recap. AI tools boost efficiency for both CAEP field and TAP teams, making processes faster and more streamlined. They simplify task management, keeping everything organized and accessible. So sorry about that. I'll go ahead and-- LISA MEDNICK TAKAMI: I think Diana might have had a tech issue on her end. So I'll just finish up on the benefits. So thank you so much, Diana. And so as she was talking about increased efficiency for CAEP field and TAP teams, simplified task management, on-demand access to CAEP resources, which I thought was really, really evident in the demo that she did today, and staying current with technological trends. And so we're really in a brave new world, from what Afraim presented and from what Diana demoed. It is exciting. And it's not perfect, like all technologies. And I want to thank both of our speakers. And I'm going to hand it over to Dulce. Thank you. DULCE DELGADILLO: All right. Great. So again, thank you so much to both Diana and Afraim for lending their expertise to our community. I learned a lot about AI. I think just learning the nuts and bolts behind that, and just breaking it down to what it's actually doing in the background, and then how I could potentially use it. We are a part of consortium as well. And so we are definitely always dealing with what is allowable, what's not allowable, how can we be much more efficient, and how can we partner. So we're definitely going to be using some of those skills. And that list that Afraim shared out, very helpful. Thank you very much. So we're going to go ahead and transition over to the next slide, which I believe is our Q&A. (DESCRIPTION) A video presentation displays a slide overview. Slide: Q and A. An illustration shows five people engaging in conversation with speech bubbles above. (SPEECH) There we go. All right. Perfect. So we're going to go ahead and open up the floor or the chat, or go ahead and unmute yourself. Raise your hand. Very informal. Please let us know. Or if you also just want to share how you are using AI as well. I am using it in all parts of my life, professional and personal, I have to say. I just put in an agenda for a meeting on how to summarize some of the deliverables that we're trying to push out through the Office of Research. So I've definitely been using that. Definitely use it for my emails. How can I maybe streamline it or synthesize it? Any questions? Any questions from our audience today? Or any comments or just feedback? DAN: To be honest, this is my name. It's Dan. I'm so unaware of what's going on that I don't even know if I'd know where to begin. DULCE DELGADILLO: In terms of just being able to interact with it? DAN: --for ChatGPT. Do this, do that. But I mean, I don't even know how to sign up for ChatGPT. DULCE DELGADILLO: All right. OK. So maybe we can put the link. DAN: And then once I've signed up, what am I going to do with it? DULCE DELGADILLO: You can click on it. And in fact, you can go ahead. And I know for Perplexity, you don't even need an account. You can literally just go in and just ask it questions. And in fact, I actually have ChatGPT on my phone, and I talk to it. So I don't even type it in, or I don't even use my phone. I actually used the voice and I specifically ask it questions. And I say, walk me through this, or simplify it, or explain it to my eight-year-old who's going to need to understand ratios soon. DIANA MARTINEZ: For ChatGPT, you can actually use it without having to log in. You'll just be unable to add any documents or add the "search the web" feature. You could also just copy and paste an email or any short document and ask it to summarize it, or do any task you need it to do. To log in or sign up, you can just use a Gmail account, or I believe they also allow Microsoft Outlook as a login. Yeah. Google, Microsoft, and Apple. Or you could just create a new account using any other email address that isn't listed on there. DULCE DELGADILLO: Great. Thank you. Diana, if you don't mind just putting in the ChatGPT website in the chat. All right. Yeah, the ChatGPT website in the chat. Thank you. Yes! All right. And I see Janice. Oh, sorry. I see Janice putting in here, saying, I hope our statewide leadership can advocate for a special pricing plan for education for ChatGPT Pro. Yes, that would be great if we can definitely-- what we're noticing is that for a lot of these, there's tiers, right? Just like in everything else, right? We live in the age of subscriptions. And so really just looking at what best serves you, how you can explore it. Afraim, go ahead. AFRAIM SEDRAK: OK, a couple of comments. First, if we want to think about artificial intelligence, we can think of it as if we have a very knowledgeable expert in every field of knowledge that we can refer to when we have a question. So we simply think about it this way. If you want to draw a picture, you have Picasso beside you. Just give it the command. I want to draw a picture of so and so, and it will generate it for you. If I want to understand a very complicated scientific topic, I can ask the AI, and it can simplify it and explain it for me. So this is one thing. On the other hand, the different models of AI that we can utilize, even for free. In Microsoft Windows right now, there is something called Copilot. It's available. It's already in Windows. It's an icon, usually will be on the taskbar. Simply, you click on this icon, and you ask any question. You give it any task to do. And you count it as if you have a very knowledgeable, efficient personal assistant dedicated for your service 24/7. As simple as that. Thank you. DULCE DELGADILLO: Thank you. Yes, definitely a new age. We are finishing up the first quarter, right? First 25 years here. And so we are definitely going into this new age of technology. All right. Before I forget, I definitely do not want to forget the resources that we have available to all of you through CC TAP. One of them is our CC TAP Listserv. So when webinars or resources from our colleagues at SCOE or our colleagues at WestEd or our colleagues at CASAS are also providing some professional development out in the field, keep an eye out. You can go ahead and subscribe. You do not need to be part of a community college in order to access this. This is an open access. We just host it through the Chancellor's Office Listserv pieces. Oh, I got a couple of comments here. In addition to the Listserv, you can also email tap@CalAdultEd.org. That email goes straight to our colleagues at SCOE that will be able to also forward again. We provide technical assistance across both sectors in the Chancellor's Office, K-through-12 and all K partners, in collaboration with our partners at SCOE. Next slide. Really quick. (DESCRIPTION) Slide. QR code. (SPEECH) Also, we do follow the model of trainers training trainers, right? So we have experts right in the field, in our own networks, in our own circles. So we encourage you. If you are just fascinated with CAEP fiscal reporting, come and talk to us. If you love data and love to do CAEP evaluation, come and talk to us. We were able to find some AI experts, and we were able to talk to us. So we definitely encourage you. If you have an expertise or a niche in somewhere that would benefit our CAEP partners, please reach out to us. That is the QR code. We are gathering a list of experts so that we can all help each other and create the best pathways for our students. Next slide. (DESCRIPTION) Slide: Thank you. (SPEECH) All right. So really quick. I do just, again, want to thank our speakers, Afraim and Diana. Thank you for lending your expertise. Chad has placed our end of our feedback survey. We conduct these after every webinar. We look at it. We discuss it. We take it back to our next webinar. And we really do use it to improve your experience, and help all of this information be as useful for you in doing your work. And with that, I just want to thank all of you for all the work that you are doing. And I hope that you have a great rest of your day. You can also go ahead and scan that QR code if you want to provide some feedback as well, or click on the link. We got some great feedback here on these pieces. If you have any other comments, either put it in the chat or leave it on our survey, and we are happy to look at it. Thank you so much again for spending your morning with us this morning. LISA MEDNICK TAKAMI: Thank you so much, Dulce. Because we are ahead of schedule, which is just amazing for us, I wanted to share. First, I want to give an opportunity if anybody else has a brave question or comment. So we do have time for that. And then I will share some of the way in which our colleagues are already using AI that they had put into our survey. So do we have any other questions for Diana, Afraim, or any of us? Or any comments about the uses of AI? Afraim, please. AFRAIM SEDRAK: Yes, I see a question in the chat. It says, will there be any training on the privacy, security risks, and tools that keep users safe utilizing AI? There are simple practices that we can even start doing right now. So for example, if I have an email that I want to summarize, or I have a document that I want to think about or even give a response to, naturally, people will copy the email and post it into any artificial intelligence tool and ask this artificial intelligence tool to summarize it, or even to generate response to it. But the way that we can simply protect our privacy is, before I copy and paste, I can copy and paste the email into a Word document, and then remove any identifiable information. So for example, if I have names, I can remove these names. I can even put fictitious names or leave it without names. And then give it to the AI. So you will not be worried about your privacy at this point. For sure, there will be more advanced tools that allow the system to be deployed into your computer. So it's not running outside of your system, which keep more privacy. But even with the available tools that you have in your hand right now, you can simply remove any identifiable information before you give it to the AI. So you are assured that your privacy is not being compromised. LISA MEDNICK TAKAMI: Thank you, Afraim. That's very helpful as a tool. Do we have any other questions from the chat, or would anyone else like to come off of mic to ask a question or make a comment? (DESCRIPTION) Emails and names of the speakers. (SPEECH) OK. Well, I'll go ahead and share some of the responses from our pre-webinar survey in terms of how our colleagues are already using AI. One respondent indicated that they are creating forms, asking for advice on how to handle interpersonal communications. Another colleague indicated "as a substitute for Google searches," which is what I commented. And I am a new user. I find it so satisfying that it's much faster than a Google search. And it gives embedded links with the sources of information. Another colleague commented, including tasks for students to do with AI in their assignments to improve their AI literacy, and use it in their vocational field. And this is something that our colleague Yarik Yano had discussed in the part one of webinars, preparing our students. And Afraim also mentioned this in his presentation today, preparing our students to be able to use AI in the workforce. Another colleague indicated creating guides, books, and SOPs. I'm guessing that's scope of programs. Another colleague commented, I'm learning to use AI when editing documents. Another colleague indicated writing projects, letters, summarizing articles, researching topics. And finally, a colleague indicated lesson planning. So from there, we have a really nice range of how our CAEP practitioners and colleagues are using AI already. And we will take that information, along with anything that is put in today's survey, as we consider a continuation of this series on AI. So with that, if there are no other questions or comments, I think we will give you back-- oh, Afraim has got something. Go right ahead, Afraim. AFRAIM SEDRAK: OK. Since we have five more minutes, I would like to do an experiment for everyone if you want to test inside your brain how the visual part of AI work. Allow me to share my screen for a second. (DESCRIPTION) Closes slides and opens other slides. (SPEECH) So, everyone, do you see the picture on the screen right now? Just simply list in your mind what are the items or the objects that you see in this picture. I will give you a very simple simulation for how the visual part of AI work. So I will give you maybe 30 seconds. Just try to think. With your eyes, see what are the objects that you see in this picture. (DESCRIPTION) Image: A close-up of a rough beige brick wall. (SPEECH) 15 seconds. How many objects do you see? OK. So I will call it now that this is the time. So let's think about it this way. When I do this for students, when I try to explain the idea of visual recognition using AI, I give them this picture and I ask them, what do you see in this picture? So most of them, they say, OK, I see maybe some bricks. And that's it. That's all what we see. So for everyone else, would you please interact or raise your hand or write in the chat if you see anything else in this picture? I'm looking into the chatting. OK. One say a red shadow. Mouse. Bricks and rocks. Lines. A rock between bricks. No one get it so far. So let me tell you. I will tell you what it is and try to find it. In this picture, there is a cigar. Can you see it? Yeah, just pay closer attention. There is a cigar. Can you see it? OK. If you cannot see it, this is the visual recognition part of the AI before it's being trained. It's looking into an image, but it cannot identify what's the object inside this image. Now, when we complete that training, something like this happen. And please, pay close attention to the picture. (DESCRIPTION) A red rectangle highlights it. (SPEECH) Do you see that? Do you see the cigar? Now it's surrounded with this red rectangle. Do you see the cigar right now? Do you see it? LISA MEDNICK TAKAMI: Yeah. AFRAIM SEDRAK: Can you recognize the cigar right now? OK. Now, your brain has been trained to see the cigar. And I can tell you, from this point forward, you can never unsee it. So if I remove the rectangle, you will still be able to see it. So this is how the AI visual part is being trained. Now, when it see any image or any life scene, it cannot identify the objects. And when it's being trained, now it can identify it. And later on, it can never unsee it. And this is how, for example, it helps in medical diagnostics for patients. It's amazing. And the possibilities is limitless. Thank you for your time. LISA MEDNICK TAKAMI: Thank you, Afraim. Wishing everyone a wonderful rest of your Tuesday. Thank you for joining us. Please do complete the survey. And just our gratitude to both Diana and Afraim for sharing their respective areas of expertise, and to the entire CAEP community for joining us this morning. Have a wonderful day.