Name: Lessons from AI-Native Universities
Uploaded: 2024-12-05T18:57:16.931Z
Duration: 29 min 8 s
Description: Lessons from AI-Native Universities

Transcript for "Lessons from AI-Native Universities": Hi, everyone. Thanks for joining us today. We're gonna start in another minute here to give a little bit more people time to trickle in. So if you can just hold tight, we'll get started soon. Okay. We're just gonna get started now even though we know some people are still filing in. But welcome everyone. Thank you so much for joining our webinar today on lessons from AI Native Universities. I'm Vanessa, and I lead our EDU customer success team here at OpenAI, and I'll be one of your hosts. Joining me today is Keelan, a solutions engineer. He'll introduce himself in just a bit, but let's dive right into what we're gonna cover today. Our goal for today is to help you walk away with an understanding of how AI native universities integrate AI across campus life, transforming education with personalized learning, dynamic interactions, and operational efficiency. Here's what to expect. We'll introduce ChatChipt edu by sharing real world use cases from universities that have already adopted ChatCpt across their campus. We'll then demo some tangible examples on how departments across your campus can use these new ChatCpt models And finally we'll wrap up with specifics on ChatCPT EDU and how we can help with adoption across your university. We are joined by several of our OpenAI colleagues today who will be moderating and answering your questions from the Q and A feature at the bottom of your screen so feel free to ask questions. And finally, we get asked this a lot we will send you the recording afterwards and you're welcome to share that with others at your campus that you think will be inspired by the content today. So let's dig in here. Doctor Connolly from Cambridge describes ChatChippT as the tipping point in AI, a technology that educators can no longer ignore. We see in our data here at OpenAI that the number one thing use Chat gpt for is to learn something. This is a powerful reminder that educators today have a unique opportunity to shape how students utilize AI to achieve their goals and ambitions. In this session, we'll explore how ChatGPT can be responsibly integrated to enhance learning and prepare students for a future where AI really will be essential. So we get asked this a lot. What are we seeing? What are other universities doing? From our conversations with staff and faculty at universities, there are four main themes that have emerged about how Chattopty can really impact education and your campus. So let's dive in. The first one here is personalized learning. Personalized learning has long been a goal in education, and with AI, it's finally here. With ChatGPT, I have a personalized tutor that I can talk to every day. It knows my goals, it helps with my projects, and it can be up to date with the unique skills I've already built. Educators can help enable this experience for all their students by providing them with the ability to engage with classroom content in a way that they can best understand. So what does this look like? Harvard Business School professor Jeffrey Buscang created a custom GPT within their ChatChippet EDU instance that plugs into his Slack and lets students ask questions about their case studies, like which CEO handled layoffs well? Common question. He says his students ask most questions between 12 and 3 AM when a human tutor isn't available, and he then uses those questions to guide his in person classroom discussions. This ChatGPT custom GPT really expands the learning capabilities of the students. We'll talk about custom GPTs a little bit later, but they're a tailored AI module that can be created by anyone within the ChatGPT edu workspace that performs specific tasks or serves particular use cases by integrating unique datasets or instructions. So next, we also see our leading universities have benefit significantly from having their staff use these AI powered services. For example, ChatGPT can automate tasks, like assisting TAs in scheduling, understanding campus policies, answering common student queries, and a crowd favorite, reviewing and generating reports across lots of data. Think about alumni relations departments being able to understand complex donor patterns and habits with a few clicks. We'll show some of this a little bit later. This helps streamline administrative tasks, support decision making, and generally improves operational efficiency. By freeing up time from repetitive admin work, staff can focus on higher impact initiatives. A staff member at the University of Maryland shared that because of ChachiPT, creativity and meaning are at an all time high for me. I can outsource routine work and find much more satisfaction in their job. We wanna support staff who are the knowledge workers of the campus and doing their best work with ChatGPT edu. Next, let's talk about faculty. We see that they are using ChatGPT to innovate in curriculum development following their institution's templates. It allows them to create tailored assignments, design innovative projects, and develop new kinds of learning experiences that weren't possible before. Using AI in the classroom enables more engaging, interactive, and up to date learning for students. We want to emphasize the power of co creating content with AI, where faculty can work with AI to design case studies and simulations, making the learning experience more dynamic and customized to student needs. Next, we'll talk about research. Researchers have huge benefit from G CHPT and they really can speed up their work and enhance outcomes. For example, we see that researchers are using CHPT to generate literature reviews saving time synthesizing existing studies. AI also supports hypothesis development by suggesting potential research directions, provides data driven insights by analyzing large datasets in a very quick way, automating repetitive tasks like data cleaning, and enables advanced knowledge retrieval for faster access to relevant information for your studies. Now I'm excited to pass this over to Keelan. He is gonna show you how some of our customers have actually made these AI benefits come true using live demos focused on research and how educators can design a tailored practicum. Thanks, Vanessa. Hi, all. I am Keelan Shul. I'm on the solution engineering team. Now I wanna start with a research example. Imagine I'm a grad student. I'm working in a biomedical lab that's consistently racing against the clock with a super demanding professor. We're actively working on biomarkers for disease detection using green fluorescent proteins, and I have this massive dataset from UniProt that I need to synthesize and look for some details. So traditionally, I would need to take this and assess each individual line item, put it through a lot of programs to be able to find out what I'm looking for. We can use Chat gbt to really expedite this entire process. So first off, I'm gonna drop this dataset into o one minutei and let it synthesize the data so we can look for some of the trends. Once it's ready to go, I'm gonna start asking a few questions. We can tell it only took about 7 seconds and already produced a pretty comprehensive, structure about the entry of the GFPs that are in here, giving us cross references. So neighboring residuals, the different sites that we wanna pay attention for mutations. If I'm working at different variants and I'm curious about what different trends and how mutations might affect the fluorescent intensity, maybe the color shift or stability, we can ask our Orwan mini model, which is gonna be the reasoning model to think through these different problems. You can see that it only took a couple of seconds that would have taken me days otherwise to look through. Now I can look at the chain of thoughts. This look for the GFP mutations. It look for the different effects. Pretty simple straightforward stuff that I could pull from literature, but maybe what I wanna do now is actually look for new mutations that are not obvious within this dataset. So similar thing, I'm gonna ask chat gbt to look through this dataset, and I wanna look for professional improvements through brightness, for its full stability, and maybe known patterns within here. As this is thinking through, we're gonna see the trends again. So the chain of thought, what ChatBT is doing is pretty unique here. It's gonna pinpoint key areas. It's piecing together key positions, targeting some of the stability, and charting the mutation dynamics. Now I'm having this really nice to easy read chart. I can see the positions 99, 206, 145. These all look like they might be variants that I can use for my study. But let me ask one final query. So I wanna ensure that the variant that I'm gonna choose is not gonna disrupt any of the chromophore formation, and I wanna make sure there's no negative impact on my study. So not only are we gonna look for key trends of positive impacts, but let's also look for ones that we wanna avoid. And I want chat gbt to tell me how and why it made the decision. So that only took 4 seconds to run through. Now again, if I wasn't using chat gbt, this would have taken me probably over a week to pick through each of the different positions, understanding the different formations, and this made it really, really easy to scan through, giving me a high level of detail on a conclusion of how I can preserve the integrity of the key residuals or avoiding specific mutations from my study. I can take each of these findings and quickly share this with other colleagues, and all of this information and work is done in the private and secure version of chat g v two e d u. So making sure that none of my research is gonna leak before I want it to. So now let's discuss a different use case. This is gonna be more of a campus handbook type use case. This is gonna be applicable to educators and students alike, and we speak with a huge set of information buried somewhere on their site, which can be very difficult to query. So we wanted to add this year's maybe student handbook, the student affairs guideline, or even details about financial aid, we can add this to a custom GPT. Now Vanessa spoke about these GPTs earlier, and this is really a domain specific customized version of your chat GPT. This could be like a campus help desk, maybe something connected to your LMS. What we can do from the main chat is actually at mention one of our GPTs for specific information. In this case, I wanna use the student affairs handbook. As you start typing, it'll populate that from the set of workspaces or GPTs that I have access to, And I wanna create a template that's gonna define a set of learning objectives to be able to create a practicum for this next year's plan. It's gonna pull the information from the specific documents that are inside of that GPT and also give reference to that. So you can see that I grabbed it from the 2024 or 25 student handbook. It's gonna give me the practicum plan template with goals, with action plans, timelines, objectives. But I wanna collaborate a little closer with Chat GPT on the output of this. For that, we can use Canvas. Canvas is a great writing tool that allows me to do inline edits of the output that Chat2pT created for me. You can see as it writes on the right hand side, I still have my chat on the left, and I can suggest edits through any part of this. For example, I maybe I wanna add in 5 objectives, not 2, and I wanna give in full examples. So we can highlight this and ask it to create 5. Now Chat gpt and Canvas is gonna go through line by line and it's gonna keep and preserve any of the original output, but you can see here that it's added the 5 objectives, given the activities, detailed examples, and resources needed for this. I can continue going back and forth to iterate on the different versions of my Canvas output here. Everything that we've done is also fully versioned, so if I've made a lot of changes, it's gonna highlight those for me. I can flip back to prior versions to make sure that I'm gonna keep everything in the backside. Now if we wanted to do some more AI function, tailoring into this, we can look at suggested edits. What this will do is go through any grammatical changes, areas that might be vague or should be expanded on, and then highlight them for the end user. So I can see that professional goals and was highlighted here, open this up, and it's gonna consider rephrasing this. If I just hit apply, Chat 2 p will go through here and make the edits necessary. Now once I'm ready to send this out, I wanna add a final polish to that. We can have it go through the entire document, add in headers where it needs to be, bold certain text, and really just kinda clean up at the final bit of text that's ready to distribute this. Once it's ready, I can copy everything out of here or I can just back out of Canvas and continue my conversation within the conversation window. Now remember, all of this is always kept in my conversation history. So at any point in time, I can go back and retrieve this, bring it back up, and I can share all of this with my colleagues. The amount of time saved on writing and ideating with Chat GPT really enables the educators to focus on more strategic initiatives. So I wanna walk through a couple other examples. Let's start with our faculty members. So by understanding the faculty's challenges, we can create these custom GPTs that connect to existing systems such as your LMS for curricular development or enhancement, building GPTs for custom, student engagement, or regarding course material. It goes well beyond the classroom as well. ChatGPT can facilitate career and life guidance to provide advice on resume building, job searching, and interview preparation through role playing exercises. We have heard so many stories about ChatTbt helping students with their own personal wellness. Many students are facing pretty significant life challenges, and this tool can provide access to a wellness resource when speaking with a human just isn't an option. I can personally attest to this. Over the last year, I was using voice mode on my phone a lot, and it gave me an opportunity to work through some pretty heavy PTSD after a major accident that I had. It helped me work through some things when I just wasn't ready to speak with somebody. It was life changing for me. And then as for our staff, chat gpt supports strategic planning and by facilitating brainstorming sessions, product initiatives, or even event planning. A common campus help desk, GPT, helps manage routine queries like password resets or campus navigation. This can replace a typical FAQ document that floats around, provide administrative support, and allow staff to focus on these initiatives. Now I'm gonna pass it back to Vanessa to show you how we can make this come to life with universities around the world. Thanks, Caitlin. Our goal at OpenAI is to ensure that we put this transformative technology in the hands of your faculty, students, and administrators in the most responsible and efficient way possible. We have seen incredible success with the engagements that we have helped drive so far, and here are a few logos of some of the customers and universities that we've been working with. So let's talk about a specific deployment here where some of this is live at Arizona State University. They have taken a proactive approach in integrating ChatGPT. In just one semester, over 250 projects were launched using ChatchePT to improve teaching, research, and even workplace positivity. This demonstrates the potential of AI in education. When embraced fully, it can become a force for systematic improvement, enhancing learning outcomes, and staff experience. Now a lot of people ask, like, what support do we provide? So we work with universities to provide frameworks to help you define your AI strategy, align on success metrics for your institution. We really try to provide templates for top use cases and GPTs, like some of the ones that we've talked about here in this session today. But we can also work with you and your faculty to identify unique GPTs and unique use cases that would drive what's most important for your institution for your faculty, your researchers, your students. We also wanna ensure your campus is enabled on how to use this technology. So we have scalable resources, webinars, and things like this to ensure that everyone on your campus feels AI literate. So next, we wanna show an example of one of these assets. Professors like Richard Waterman at Wharton have built custom GPTs for his stats class, which helps students practice for exams and also understand the real world applications for stats topics that may seem obscure. I remember my stats classes in college, little tough. Dave Tully, a University of Maryland professor, uses ChatGPT in his course AI for environmental good, where students create GPTs to be study buddies, research assistants, or community personas for scenario based learning. Here on the slide is an example of some of the GPT guidance and and ideas that we provide universities that can drive amazing impact for your students. The potential really is massive. Chatt GPT EDU and the partnership with OpenAI has the power to transform your campus into an AI native institution, and we hope you're inspired now to think about what this can mean for your university. I'll now pass it back to Keelan to touch on our security posture and wrap us up. So we build chatgptedu to ensure that AI can support everybody in the academic community. So with access to our flagship models, GPT 4 0 and 0 1 preview, institutions have powerful tools for data analysis, for coding, and for automating tasks. We wanted to make sure that these capabilities are accessible, which is why we offer affordable c pricing as well as higher message limits compared to the free version of chat gbt. But chat with gbt chat and gpt edu is also a very secure environment, and it's been built to be able to share custom GPTs across the university workspaces. This means that each institution can tailor AI to fit their very unique needs. At OpenAI, we take data privacy very seriously. One of our core principles that we will never train on your business data. You own and control all of your data, your inputs and outputs, and you have full control over how long your data is retained. This is both conversation and GPTs. We prioritize security, privacy, and provide administrative controls within the GPT management. We wanna make sure that universities feel comfortable and confident that their data is secure and that any tools they use are gonna meet their standards. Now thank you for joining us today. We hope that we give you some insight on how our educational partners are leveraging ChatGPT. We encourage you to test out some of the examples that we spoke with today. Now the QR code will bring you to our EDU landing page with more detail about the way we work with universities and the a OpenAI forum along with recordings from our recent virtual events, hosting UCLA, Wharton, and Harvard Business School. Feel free to reach out to your account team if you have any questions or wanna learn more about chat gbt edu. And looks like we're at the end of our time. Thank you. We hope to see you soon. If there's any questions left in the chat, we'll try to answer those quickly. And finally, once we close out, you'll get a short survey. So please provide feedback so we can improve these going forward.