Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Intentional Teaching, a show about teaching in higher education

Intentional Teaching, a show about teaching in higher education artwork

AI Learning Assistants at Scale with Kavitta Ghai and George Hanshaw

Nov 3, 202641 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Share your thoughts about this episode as a text message. Consider the AI learning assistant, with access to all the materials from a particular course and programmed to serve as a tutor for students in that course. What happens when you embed this kind of AI technology in every single course offered at a university? Today on the podcast, I talk with George Hanshaw, director of learning solutions at Azusa Pacific University Global, where they did just that. I also talk with Kavitta Ghai, co-founder and CEO of Nectir, the company that partnered with George and his team to roll out AI learning assistants at scale. George and Kavitta talk about the faculty and student experience of this technology, why it's a particularly good fit for asynchronous online courses, and why students choose to talk to an institutionally provided AI tutor instead of their good friend ChatGPT. Episode Resources George Hanshaw on LinkedIn Kavitta Ghai on LinkedIn Nectir Three papers by George Hanshaw and colleagues on the Spark experiment: “Exploring the Effectiveness of AI Course Assistants on the Student Learning Experience,” “Exploring Barriers to AI Course Assistant Adoption,” and “An Experimental Study on the Effect of AI Course Assistants on Student Grade Outcomes.” “Artificial Intelligence, Cognitive Offloading, and Implications for Education” by Jason Lodge and Leslie Loble Support the show Podcast Links: Order The Norton Guide to AI-Aware Teaching by Annette Vee, Marc Watkins, and Derek Bruff. Intentional Teaching is sponsored by UPCEA, the online and professional education association. Subscribe to the Intentional Teaching newsletter: https://derekbruff.ck.page/subscribe Support Intentional Teaching on Patreon: https://www.patreon.com/intentionalteaching Find me on LinkedIn and Bluesky. See my website for my "Agile Learning" blog and information about having me speak at your campus or conference.

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

Derek BruffWelcome to Intentional Teaching, a podcast aimed at educators to help them develop foundational teaching skills and explore new ideas in teaching. I'm your host, Derek Bruff. Over the last couple of years, I've put together a couple of episodes of this podcast devoted to the notion of an AI-powered learning assistant. This is the idea that you can design a custom AI chatbot that has access to materials from a particular course and can serve as some kind of tutor for students in that course. I've known of a number of faculty here and there either building or using these AI learning assistants, but it wasn't until earlier this year that I learned about a university that rolled out AI learning assistance to all of its courses.

Derek BruffThat university, Azusa Pacific University Global, is entirely online and its courses are entirely asynchronous. I was very curious how system-wide AI learning assistance would work in that context. Today on the podcast, I talked with George Hanshaw, Director of Learning Solutions at Azusa Pacific University Global about this very ambitious AI project. Did I mention that they went from a small pilot to university-wide rollout in just four weeks? Yeah, that's what they did.

Derek BruffI also talked with Kavitta Ghai, co-founder and CEO of Nectir, the company that George and his team partnered with to provide AI learning assistance enterprise-wide. Nectir has recently signed a contract with the California Community College System, offering its AI learning assistant platform to all 116 community colleges and potentially over 2 million students. It's clear that this use of AI to support student learning has a lot of energy behind it, but does it work? In our conversation, George and Kavitta talk about both the faculty and student experience of this technology, why it's a particularly good fit for asynchronous online courses, and why students choose to use this institutionally provided AI tool instead of their old friend ChatGPT.

Derek BruffKavita and George, thank you so much for being on the Intentional Teaching Podcast. I'm very excited to have you on today and to hear a little bit of your stories. Thanks for being here.

Kavitta GhaiThanks for having us, Derek.

George HanshawYeah, thank you.

Derek BruffAbsolutely. I have recently crossed the 100 episode mark in my podcast. And as part of a little bit of change up, I've started asking a new standard opening question. So my question is this can you share a moment where each of you found joy in your work in the past few weeks? Kavita, I'll start with you.

Kavitta GhaiOkay, sounds good. I love this question. I think being a founder and CEO, and especially having started my company when I was an undergrad student at UC Santa Barbara, I never imagined that I'd be in this position eight years later. And so everything that happens at work is a joy for me to just be doing this is incredible. But when I think back to the last couple of weeks, you know, we're we're in that summer phase where teachers are getting ready for the school year. And so a lot of the work that we do at Nectir is training teachers on AI literacy, not just so that they can deliver it to their students, but so that they understand how to safely use this technology, how to best engage with it.

Kavitta GhaiAnd we can share best practices with them that we've learned from, you know, this California statewide deployment that we've done. And there was this uh faculty member at one of the California community colleges who came to one of our sessions last week and told us, I hate AI. I never have used it before. I don't want to use it with my students, I don't want them to use it. I don't see there being a use case for it in the classroom. She's very adamant about her stance on AI. And we said, okay, I hear you. I totally understand.

Kavitta GhaiThere is a lot about AI that we should be wary of and a lot that you can probably learn from and bring into your classroom. So let us share that with you and we'll see how you feel afterward. And after just a one-hour training session with our customer success team, she walked away saying, Oh, I see it. I see the vision. I get it, why we need AI in the classroom. She was like, I can see so many ways to use this with my students, so many problems that I've been having that I can solve with it.

Kavitta GhaiAnd a week later, this happened yesterday, we just got the message in the Slack channel. She built six AI assistants for her class in one week. And she's ready to go. She's like, I've been telling all of my fellow faculty how much I love it. This thing is amazing. My students are gonna get so much out of it. And to have seen how quickly we were able to change her mind by, you know, not persuading or convincing her that she should give it a chance or try it, but by actually showing her how well this has worked for other faculty and students, how much we're seeing not just the GPA increase, but students' motivation to learn actually increase.

Kavitta GhaiAnd then see her change her tune about it and build six AI assistants a week later. I it's just one of the coolest feelings ever. It's exactly why I do what I do. And those moments are so fulfilling to me. So that was a very easy answer for me.

Derek BruffYeah, yeah. Empowering her, giving her a sense of agency over this.

Kavitta GhaiYeah, exactly.

Derek BruffWhat about what about you, George? Where have you found joy in your work lately?

George HanshawWell, that that's amazing what just happened, uh, Kavita, with your story. That's awesome.

Kavitta GhaiThank you.

George HanshawAnd I'm lucky enough that I get to teach. I teach at a couple different universities, and and I'm the director for digital learning solutions where we implemented Nectir university-wide. It's it's not an option, it is in there. But as a faculty member, man, I I think of it instead of what brings me joy, I think. I think what brings me joy is bringing the joy. Let's bring it out and let's get it because I just get such a sense of fulfillment, joy, and any other adjective you want to use, seeing uh students kind of grow.

George HanshawThey grow from personal interactions with me, with interactions with their peers, interactions with the uh assistants that we have set up. So it's just kind of amazing. And one person lately came and she told me, said, You know, I am not a good academic writer. And I was like, Great, neither was I. Because I came into academia later in life, right? I say, so I totally get it. And I was able to help her, uh, kind of tutor a little bit on things to look at.

George HanshawAnd then we have um one of our nectar assistants is a writing assistant in our writing hub. And I said, Hey, when you write something up, put it here and it'll give you good quality feedback on how to adjust. It's not going to rewrite the paper for you. And she wrote me back just saying how amazing that process was to go with it to a writing hub and get that feedback when she needed it, not when George or anybody else was available.

George HanshawSo that sort of stuff happens a lot. And so a lot of a lot of joys happening. Yeah, yeah.

Derek BruffEmpowering the students, right? Giving them growth. That's powerful. Well, George, I want to talk about these assistants. Um, so if I'm a faculty member at your institution right now, um it sounds like I'm gonna have some AI-powered assistants, whether I want them or not, um, but what does that look like in practice? Like, how do I how would I set up the assistants for my course? Um Kavitta, you said this faculty member built six of them last week, right? Can you give me an example of what that might look like in practice at a course level for an instructor?

George HanshawYes, absolutely. And our faculty, if they want a specific, uh, we call it Spark. If they want a specific Spark, that's what we call our assistant built, they come to a digital learning solutions team and we build it together. Okay. So it does exactly what they want. So we have a slightly different model than some, but at the moment that you come into class, you have Spark available, students have Spark available week one, okay. Uh, just right there.

George HanshawAnd Sparks train on every single thing in the classroom, which is cool. So you can, and it's Socratic, so we talked about Socratic assistance before. Uh, they come in and they say, Hey, what's my first assignment? I'll say, Well, your first assignment is X, Y, Z. How much do you already know about this? And then it starts to engage in a conversation. So it's actually really cool and it's preset. So faculty just come into their classroom and Spark is there.

George HanshawAnd we've also set it up for active learning. So we have different assistants in multiple places within the uh course because we're we're 100% online, right?

Derek BruffYeah, and that's the game change. Yeah, as it I don't I I don't know this, but like do you have synchronous elements to some of your online courses, or is it is it all asynchronous?

George HanshawIt's a hundred percent asynchronous. Okay. And what's kind of amazing, so initially when when online became a thing, right? There was the studies that said no significant difference. So online and face-to-face were had similar outcomes. But now, as we're able to add in active learning uh with our AI assistants and have that 24-7 interaction being able to happen, we're finding that there is a difference now and is swaying towards the online classes.

George HanshawSo that's forcing um you know changes to be happening in the face-to-face environment as well. Okay.

Derek BruffSo when you say active learning in an asynchronous online class, what does that look like? Like what is what is an AI assistant doing if it's helping foster active learning?

George HanshawIsn't that cool, right? So, like we we literally have a think pair share in some discussions. Okay, yeah. Here's the prompt, student thinks about it, engages in a conversation with the AI assistant, and then shares out what they've learned to the entire team. So they're actually engaging in that conversation that helps that learning stick. Because we we we're big, we like to call it sticky learning. We like that learning to stick so students can recall what they learned and use it in creative ways to solve these very complex issues that people are facing today.

George HanshawThey're more comp issues are more complex today than they were just a few years ago. So we want to make sure that they have that opportunity to engage in those very nuanced and gray discussions.

Derek BruffThank you. Uh Kavita, what would you add to that? Like to paint a picture of kind of what what the tool looks like in practice, and then like I say, then we'll kind of step back a bit.

Kavitta GhaiYeah. So our team at Nectir developed Nectir AI in early or late 2023, early 2024. And we did it in private beta with Stanford GSB, Boston University's Questrum School of Business, and LAPU. And so picked really three distinct types of campuses and said, talk to us about AI. What scares you? What's not working? What are you afraid your students or teachers are going to do with it? And also, what are the problems that you're facing that you really think deep down AI might be able to solve this?

Kavitta GhaiAnd, you know, they everything from we already have copilot, but it's not scalable. It doesn't allow us to customize it the way we want, or we don't want to be beholden to a specific model provider. We want to have access to all these models, but it can't be sending this data back as training data to the model providers. So we had to take all these different circumstances and build something that we call AI infrastructure. So Nectir helps schools adopt and implement AI safely. So it gives them access to all of these frontier models and then allows them to really easily create custom AI assistance and agents that are trained or fine-tuned on their specific course and campus data.

Kavitta GhaiSo when a student is going to ChatGPT and asking about class, it's going to give them a generic answer. Maybe they're uploading the assignment with it, but that's really the context that it has. It's going to pull whatever best information it has on those topics. But as we know, especially once you get to college-level courses, the whole point of going to college is you have these experts that are leading the class that have deeper knowledge than you're going to be able to just go find easily on Google or even ChatGPT. And they have a specific pedagogy. The way that they teach and deliver this information is very specific, faculty by faculty.

Kavitta GhaiAnd so where we think AI is best used in the classroom is when a teacher is able to set guardrails to tell the AI assistant here's all of the material of my class, everything from the textbook to the assignments to the rubrics, and here's how I teach my class. So this is the structure, this is what we learn week by week. Here's even the tone and intonation that I use with my students. I want you to reflect that as closely as possible in every conversation you have because I have a very specific way of teaching and a very specific set of information that I need my students to know for the assignments and the tests.

Kavitta GhaiAnd so with Nectir, we allow every faculty member to really easily go in and figure out how to customize that and build it within 15 minutes and then embed it right inside of their LMS class page so that their students aren't going out to a whole new tool. It's part of that existing workflow they're already doing on a daily basis. So you want to put that 24-7 support system, meeting the students where they are, not adding more work to their plate. If anything, it should be a place where they can go ask more questions than they've ever asked before because it's the kind of questions that you would be afraid to raise your hand and ask in front of everyone else.

Kavitta GhaiBut you can ask as many of those as you want to your AI assistant. And it's going to give you an answer that's grounded in both the course material and the guardrails and parameters that your teacher is comfortable with. So the student knows I'm engaging with AI that has been made safe for me.

Derek BruffAnd George, I want to turn to you because I everything Kavitta said makes a lot of sense to me as a if I put my faculty hat on, I understood most of that. And I'm like, okay, yeah, we need to, we need to protect student privacy, right? Um, and if I especially put if I put my administrator hat on to understand kind of the real value of having systems that are properly vetted and we have you know institutional contracts with with vendors, right? Like there's a lot of protection that comes from that. I'm curious, in your view, why do students use these tools?

Derek BruffEspecially if ChatGPT is right there on their phone and they're used to talking with chat all the time, as as as the kids and my wife say. Why would they use Spark if they if if Chat GPT is just sitting there?

George HanshawWell, you know, we actually asked them, uh, and we got some very interesting answers to that. And first is it's right there. So the access is there, they're in their classroom, there's Spark. So it's easily accessible. That's the first thing. And then because it's trained on the content in the class, so every single Spark is trained on that class, so they get specific responses about it and specific interactions about it. Uh, so those are the two main things because we we looked actually when we initially rolled it out, there was a little bit of students.

George HanshawUh, can I use this? Right? There's a little bit of that confusion. So we had to have like almost an internal marketing campaign. Say, hey, this is here, it's safe as Kavitta says, uh, and it's meant to help you learn. So by that internal little internal marketing campaign, um, it it helped, and students use it. And we found that the ones who use uh Spark actually, even in their like uh post-course surveys, they write more concretely and uh less emotionally about it.

George HanshawSo they're they're actually getting more from the content than folks who don't use it. So it's it's pretty cool to kind of look at those nuanced differences. So if they use like ChatGPT or something, they're gonna get an answer and they're not going to be able to write more concretely about it. They're just gonna their critical thinking skills are gonna shrink while we're here to help their critical thinking skills grow.

Derek BruffYeah. Because part of it is that Chat GPT will not act in that Socratic manner typically. It's gonna answer all your questions and try to answer two more you haven't even thought about, right? And it's gonna do too much of the thinking for you. Where it sounds like the Nectir tool has some safeguards built in to have it act as more of that Socratic tutor so that it's engaging in actually useful learning interactions with students. Um yeah, that I mean that's the goal, right?

George HanshawBecause we've seen studies on the studies on cognitive decline, and that's yeah, of course. If if I asked Kavitta to do all my homework, you know, I'd have cognitive decline too. Right, right. Because I wouldn't be doing anything. So that's a real important distinction to make. We see um uh students getting better when we use it well. So that's that's a very important thing to do.

Derek BruffBecause you want a tool that's gonna actually do the educational work, and ChatGPT won't do that, but then you have to find ways to motivate students to use that tool as opposed to ChatGPT, which is kind of convenient in other ways. Um what um I I am curious, Kabita, you said that you can get you can get an agent built in about 15 minutes. Um, I have talked to some faculty who have been kind of working this space for a while, working with other tools to try to build custom AI chatbots that do kind of pedagogically purposeful things. And usually it's more like four hours of work to kind of have something that works uh to as a starting point, right?

Derek BruffThat that you're you're drafting these system messages to these custom chatbots, and you're having to overcome a lot of the design that's already in in the AI chatbots. So I guess I'm curious about like what does it take before the faculty members get involved to make these chatbots work in useful ways? And also how much flexibility is there for faculty to kind of do something that's very unique and personalized to the kinds of teaching interactions they want to they want to have.

Kavitta GhaiYeah. So you can create a custom AI assistant or agent and place it at a section level, a course level, a department level, or even campus-wide. So a campus-wide assistant could be something like a FAFSA advisor. It's trained on all your financial aid documents, all the deadlines, it can review FAFSA paperwork for a student before they submit it. And then it can even say, hey, based on this paperwork, you also would qualify for these scholarships. You should think about applying for them. And so you can think about how wherever a school needs to bridge a gap of support, whether that is for their internal team or for the students or faculty, that's where they can place a Nectir AI assistant.

Kavitta GhaiAnd it can be custom to anything that they want. So it's not like you're going in there and there's like 20 you can choose from. You're gonna go in, you can give it an avatar, a name, you can personify it, like they call it Spark. And then it can be, like I said, put at any level within the LMS on the campus website, wherever you want it. And then once you give it a name and an avatar and a little description, then you're gonna go into the knowledge base, and that's where you can upload anything from a lecture video to a YouTube video, a URL, or you can even just press bring my entire Canvas class page in here.

Kavitta GhaiAnd it'll auto-scrape the whole Canvas class page, add it into the brain of the AI assistant, never shares it back with the model provider. And then you can even decide if you want your assistant to have access to an LLM. If you decide I actually only want it to be trained on my course documents and not have any LLM access at all, you can set that as well. And you can always go back and change those settings as you want. And then once you get to the prompt section, you go to our prompt library and we have hundreds of prompts that you can get started with. So pretty much every use case that we've tested and verified works, we'll put it into the prompt library.

Kavitta GhaiSo we're not asking people to start from scratch. And that's also where our customer success team comes in. We sit with every single faculty member on every campus we go to and we help them build their assistance, help them figure out how to implement it in their course specifically. So for some of them, it's building a course assistant, which is kind of like a teaching assistant for their students. But then a lot of folks, like that teacher who built six of them, they get really specialized. Like in MBA courses, we see oftentimes there's a case study partner. And so they're saying, you're actually gonna play devil's advocate to whatever is the opposite stance that the student is supposed to have.

Kavitta GhaiYou're gonna poke holes into all of their arguments. And then at the end of the conversation, you're gonna give the student feedback on how it went, what they could do better based on this rubric. And so teachers can get really specific about here's exactly where I want to use AI. Doesn't have to be across the entire class, or it could be 10 different assistants for 10 different things that you want to do in class. So we wanted to build something that teachers can grow with over time as they get more comfortable with it. But yeah, to build that first course assistant takes about 15 minutes.

Kavitta GhaiEven if you don't want to drag and drop everything, great, it'll just auto-upload it from your Canvas page and then embed it right in there for your students to use.

Derek BruffWow. Okay, okay. And some of the hard work that my colleagues have done of drafting a prompt, testing, redrafting the prompt, testing more, giving it to students. The students break the bot, they have to revise the prompt. Your team has done some of that already for this kind of library of prompts so that you have prompts that work well with your agents.

Kavitta GhaiExactly. And faculty can even come to us and say, Hey, I have this prompt. Can you make it better? Can you refine this for me? And so a lot of it is, you know, when I tell people this, they're like, well, that's a lot of human labor, a lot of really detailed specialized work person by person. But yeah, that's what it takes. We're at a whole new paradigm of technology. We have to go the slow way and teach people how to use this properly because every single one of us is going to be impacted by AI, if not already. And so it is our responsibility as the people who are building these technologies to make Sure, that we're not just throwing it out there and saying good luck, have fun, but we're actually going and making sure that it is being used correctly every single time because this is a very, very important use case.

Kavitta GhaiThis we cannot afford for students to not have AI in the right guardrailed way that they need it. It has to be done right off the bat.

Derek BruffYeah. Yeah. I think about this is a bad analogy, but right. Like, what if there was no PowerPoint and all of a sudden PowerPoint showed up, and every faculty member in the universe had to learn how to use PowerPoint? Um that would be a more solvable problem than where we are now with AI.

Kavitta GhaiExactly. Yes.

Derek BruffRight. Um, George, can you share maybe a couple more examples of how some of your faculty colleagues are making use of Spark, of Nectir?

George HanshawWell, I would actually like to do that by piggybacking off something that Kav said about feedback. Because using AI, it's actually elevating the role of the faculty member because it's giving them more time to build relationships and give better, better, more detailed feedback to students, right? So students grow as well. And one way that we did that was we made several Spark assistants, and we actually did a study on it that's up for publication right now. And you can drag and drop an assignment right in there because it's trained on what the assignment should be, the rubrics, you put in your own personality in there as a faculty member, and then it gives detailed feedback, hey, in this section, and it'll write it out in different colored fonts, whatever you want, to the student, and the so the instructor can get all of that, they dunk in the assignments, they get that feedback, they read through it, and then they make their adjustments as well.

George HanshawAnd now I can leave a quick video um feedback announcement kind of talking about the uh feedback that was given. So now I've actually increased my human touch because I didn't have to spend that hour uh going through all of these different parts and pieces within that assignment. So I can scan, give my personal feedback, which is much more contextual than what we're getting from the assistants, right? So that's elevating my role as a faculty member.

Derek BruffI think that's interesting because I feel like the notion that a student would have access to an AI assistant, it's trained on my course materials, it's designed to help me in this particular course do well. Um I can talk to my AI friend anytime I want to, 24-7. Why would I talk to my faculty member anymore? Like, do you do are there concerns about kind of disrupting that relationship now that that the the bot is handling a lot of interactions that might have gone to the faculty member in the past?

George HanshawWell, uh here's what we found. Okay, because yes, people were worried about that, right? And here's what we found there's a lot of students who normally wouldn't come to Dr. George, right? They just wouldn't because they're like, I'm not gonna ask a dumb question. So they can't, right? I mean, not that I would react in that way, but that's a thought. I've had that thought myself just the other day, actually, where I was not gonna ask that question. Uh but when they have a chance to practice asking these questions, guess what?

George HanshawThat builds up their efficacy so they can now come and ask me a question. We have a great interaction. So they're actually coming to me even more prepared with questions. So I have to make sure that I'm on the top of my game as well to ensure that I'm giving them the best uh feedback, the best interactions possible. So we're at what we're actually seeing. So the fear was man, what do what why do we even need faculty? Well, we need faculty even more now to deliver the contextual pieces, relationship building pieces, and then the the AI structure actually helps people feel more comfortable coming to the prof, which is cool.

Derek BruffYeah, yeah. That's it. I have a colleague at UVA who's been doing a lot with agents, and she's she finds the same, the same. She was actually on the podcast a while back, Jing Jing Li. Really? And she finds that um she was worried about this, and she she she reported recently to me her office hours are just as busy. The types of questions are different. Yeah, she's doing more career mentoring now in office hours than like answering concept questions, right? Because the students are getting some of that from their AI bots, and so the quality of the interactions, yeah, yeah.

Kavitta GhaiYeah, yeah, we hear that across the board for so many schools that we're at. We keep hearing that feedback of different questions that I'm getting, but almost more human interaction. Because I think what it's doing, and what we've seen in a lot of the anecdotal studies we've done, it's reducing the stress and anxiety of students when they're in class itself because they have so much time outside of class to ask all the questions and feel as prepared as they need. Whereas before you were kind of stuck either waiting to get into office hours that once a week, or you were waiting to get an email response from your teacher or TA, and you kind of just had to like twiddle your thumbs in the meantime.

Kavitta GhaiBut now you actually have this resource where at two in the morning, which is when you're studying, no one else is awake, but the AI is there and you can ask the dumb question over and over and over. And this was very much done purposefully, the way that we thought about this. For me, I'm autistic and I have ADHD. So I built this company because no classroom that I've ever been in in my life has felt comfortable for me or my brain to be in. And I was the kid in the back row my whole life, never raised my hand, knew that it was gonna be a stupid question that came out of my mouth.

Kavitta GhaiAnd I always felt like I'm just not built for this. Like everyone else can probably figure this out. It's it's a me problem. And it took me being in college and seeing so many other people say the same thing, people who I thought were so intelligent feel the exact same way. And I realized maybe it's the structure of the classroom itself and it's not us. Maybe even if all of our brains learn differently and think differently, there still should be a way that all of us can have an accessible way of getting the help that we need.

Kavitta GhaiEspecially if I'm paying five to six figures for this education, I feel like I should be able to ask all the questions that I want and get the answers when I need them and not have to wait two days for it. So I think it's what we're doing here with AI is leveling the playing field so that now all the kids in the back three rows, they're asking just as many questions, if not more, than the kids who sit in the front row. And everyone, regardless of whether English is your first language or you're shy, or you're a single parent who's working a full-time job and studying at night, everyone gets the same access to really high quality support that meets them where they are and can answer whatever questions they specifically have, whether it's about the course itself, the deadlines coming up, the career path that they might be on.

Kavitta GhaiEveryone should be able to get what they specifically need out of every class that they take.

Derek BruffWow. Um you are you are a mission-driven uh CEO. I like that. Thank you. I wanna I want to go back in time just a little bit um and kind of talk about a couple of kind of moments here because uh I think I heard you, Kavitta, say you started this company in 2018, which was well before any of us had heard about Chat GPT.

Kavitta GhaiYeah.

Derek BruffAnd I'm getting a sense of kind of why you started the company, but can you say a little bit more about like what did Nectir look like before Generative AI started to change everything? And how did how did that kind of pivot your company's mission?

Kavitta GhaiYeah. So my co-founder Jordan and I started Nectir when we were undergrad students at UC Santa Barbara. So he was a sophomore, I was a junior, and we never thought it was gonna turn into what it is today, let alone a company even then. It was really just born out of this frustration of God, I don't want to go to class tomorrow. And I love college, but I hate school. And we would always be the last two sitting on the couch and convincing each other, you drop out first and then I'll do it.

Kavitta GhaiNo, you drop out first and then I'll do it. And finally, one of these days, Jordan asked me, he was like, Okay, hypothetical question. What would it take for you to be excited to go to class tomorrow? What would it have to look like for you to be genuinely in a good mood about going to class and learning and doing all the school stuff? And it led to this seven-hour conversation we're at UCSVs who are walking on the beach as we're doing it. It was beautiful. And finally realized that the core problem you're trying to solve for was creating some way for students to have 24-7 support.

Kavitta GhaiWherever they are, doesn't matter what the question is, whether they need help with financial aid paperwork or they need help with figuring out what next class they should take, or how to study for the midterm tomorrow. There should be a way to get your questions answered way faster than we were at that point. And so we approached UCSB and we just started asking questions. Like, who can tell us why every tool that is being given to us in the classroom feels like it was built by someone who hasn't seen a classroom for the last decade? And we got sent to, we didn't know what it was at the time, the instructional design department.

Derek BruffOkay.

Kavitta GhaiAnd we sat down with the instructional designers and we asked them point blank, why does it feel like the classroom sucks? And if we're paying you $40,000 each, why is it that you can't give us better technology? And we totally expected to get yelled at for this.

Derek BruffStop asking awkward questions, kids.

Kavitta GhaiExactly. And instead, they were like, no, we agree. We totally agree with you. We also wish the technology was better. But they explained how in order to adopt and scale any tool across a campus, it has to be FERPA compliant. We have to be able to afford it, it takes a year to adopt it. We have to make sure every faculty member knows how to use it because they're the gatekeepers of it, has to integrate with the LMS, all of these things. But they said there's just not enough people building this technology in a way that schools can actually adopt at scale. And they had the exact same view of the problem as we did.

Kavitta GhaiThey had something that they called time poverty. Time poverty was UCSB's measure of how long it took for a student's question to get answered. And that was anything from going into office hours to talking to your professor or an advisor, and it took on average two days to get a question answered. And they said we have 14,000 faculty, staff, and admin and 26,000 students. And it's 14,000 people is still not enough to give every student the help that they need. It just isn't. And so they said, we don't know, there is no solution for how we scale the help that every student needs to every single one of them.

Kavitta GhaiAnd so that's what we started to solve for before anything else. We said there's a million problems to solve in this 400-year-old classroom model that we still have. But the most important one that would move the needle the most is if every student got the support they needed 24-7 and no one fell through the cracks for any reason. And so pre-AI, the best thing we could think of was to build a system where students could ask and answer each other's questions. And there was just a better way to communicate than email because that was the only thing that we had. And email was just not working.

Kavitta GhaiEvery company had figured out that Slack was a better way of communicating than email. So that's where we started. We built Slack for schools. And it was this campus-wide communication hub. Every time a student registered for a class, they would get added into their Nectir class channel. The TA and professor would be in there. And so they could answer a question once and everyone would see it. They wouldn't get the same question a dozen times in their inbox. And it became this habit for students to start asking questions in the Slack chat or the Nectir chat to each other before and after class.

Kavitta GhaiSo very naturally, Nectir chat evolved into Nectir AI when we realized this is great, but AI can answer those questions even faster and with more accuracy and in a more personalized way than another student can. And so Nectir Chat actually still exists today. And at some point, we'd love to combine them and have this human collaboration meets AI. I think that would be really cool. But our main focus and what we're solving for is providing 24-7 personalized support to every single student wherever they need it in their learning journey.

Derek BruffYeah. So, George, I'm going to go back to you. Because did I read correctly that you went from a pilot of like single-digit number of classes using it to all classes?

George HanshawOh, yeah.

Derek BruffLike you rolled it out.

George HanshawYeah.

Derek BruffWow.

George HanshawIn four weeks. Well, we actually did it in two, but the classes didn't start in four weeks. Yeah. Well, you know, like, hey, two weeks.

Derek BruffLet's go. You know, uh wow. Okay. So it's available in all classes, but it is it still something that faculty have to decide to turn on and start that you know 15-minute process to get going, or is it is it kind of waiting there for students with or without faculty input?

George HanshawYou know, that was an institutional decision to it is on and it is using, it is being used uh for the mere fact that if you don't teach your students and give them safe AI and teach them how to use AI in a way that helps our critical thinking, they're gonna be using something like chat claude and they're gonna give you answers. Okay, right. So we we found that we were actually doing our students a disservice not to have it in the classroom.

George HanshawAnd we started working, right? You're actually harming your students if you don't have something like that in your classroom because they're getting it elsewhere.

Derek BruffRight. This is, I guess, a response to what I hear from especially asynchronous online instructors all the time, which is my discussion boards are now trash, it's just full of AI slop that my students have commissioned. Not 100%, right? But that's that's that's the concern that I hear often is that the students are just phoning it in. And it sounds like your your approach to kind of it's the fighting fire with fire, right? Like we're gonna give students a better option than that. They lean into that.

George HanshawI'll practice my vulnerability right now. We were in an off-site meeting, all the faculty, myself, we were there, and this was three years ago now, I want to say. And this was after we installed the assistants, right? And everything was working really well. We were just beginning to explore other ways to use them, things like that. And then I walk into this meeting and they said, George, you're in a rut. And I was like, I'm in a rut. I don't like ruts. What's the rut? Right and they and what specifically in our history course, one of the uh our full-time faculty said, When I go into class, it's Claude talking to Chat GPT, man.

George HanshawAnd I said, Well, we need to do something about that, right? Because we don't want that. So we installed different, you know, think pair shares. One of the things where they would students go in, and specifically in that history course, they talk with documents, they they have um antagonistic conversations with uh people from history, right? Where they have to take the other side, like what Kavitta was talking about earlier. And by doing that, it transformed that class in a huge way.

George HanshawWe see that through the post-survey results, through what the instructor's saying, because I said, hey, you know, are we still having Claude talking at Chat GPT, or are we actually having conversations today? Right? And they're they're actually having conversations. So we then took that and we're scaling it out to everything. So it's how you build your discussions that causes that. Because you know what, quite frankly, um I would not blame a student for using Chat GPT or Claude for some of the stupid discussion questions, yeah.

George HanshawRight?

Derek BruffYeah, yeah. If it feels like busy work, it's really demotivating.

George HanshawYeah, right. How to demotivate a student in 30 seconds. Right.

Kavitta GhaiTotally.

Derek BruffWell, when I think about especially some of the challenges that online education is facing right now, like it does seem like there needs to be a very different model going forward. Uh, thank you for being on the show and uh spending a little time with us today. This has been really, really fascinating. I'm really glad to hear how this is playing out in some real classrooms out there. So thank thanks thanks for coming on.

Kavitta GhaiThanks for having us, Derek. This is awesome.

Derek BruffThat was George Hanshaw, Director of Learning Solutions at Azusa Pacific University Global, and Kavitta Ghai, co-founder and CEO of Nectir. Thanks to both of them for coming on the show and pulling back the curtain on this use of AI to support student learning. There is a literature review of sorts published earlier in 2026 by Australian educators Jason Lodge and Leslie Loble. I've been quoting it regularly in my faculty workshops on AI where teaching. After surveying the literature on AI and cognitive offloading, there's not a ton of research yet in this space, but enough to see some trends, they argue that students' default uses of AI and learning often, quote, short-circuit the cognitive effort required for sustainable deep learning, end quote.

Derek BruffBut, Lodge and Loble go on to say, quote, the good news is that research also suggests that these harmful effects can be counteracted through purposeful teaching and learning strategies and effective design of AI education technology, end quote. Usually on this podcast, I focus on the teaching and learning strategies side of that bit of good news, but I was glad today to get a sense of how thoughtfully designed educational technologies can support better student engagement with AI. Intentional Teaching is sponsored by UPCEA, the Online and Professional Education Association.

Derek BruffIn the show notes, you'll find a link to the UPCEA website where you can find out about their research, networking opportunities, and professional development offerings. This episode of Intentional Teaching was produced and edited by me, Derek Bruff. See the show notes for links to my website, the Intentional Teaching newsletter, and my Patreon where you can help support the show for just a few bucks a month. If you found this or any episode of Intentional Teaching useful, would you consider sharing it with a colleague? That would mean a lot. As always, thanks for listening.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Teaching Civic Skills with Rachel Piontak

    Nov 17, 202644 minAudio

    Share your thoughts about this episode as a text message. Partnering with over a hundred college presidents, the non-profit Citizens & Scholars launched a Faculty Institute a few years ago. It’s an 18-month professional...

    Credit the people on this episode

  2. AI-Aware Computer Science Education with Briana Morrison, Greg Nelson, and Elijah Rivera

    Oct 20, 20261 hr 21 minAudio

    Share your thoughts about this episode as a text message. Before there was ChatGPT, there were generative AI coding tools. This means computer science, as a discipline, has had a little more time to adapt to AI than...

    Credit the people on this episode

  3. Light Bulb Moments from 100 Episodes

    Oct 14, 202658 minAudio

    Share your thoughts about this episode as a text message. Welcome to the 100th episode of Intentional Teaching! I am very excited to reach this milestone, and I am deeply grateful to all my guests and all my listeners...

    Credit the people on this episode

  4. Curricular Responses to AI with Deborah Barnett

    Sep 22, 202642 minAudio

    Share your thoughts about this episode as a text message. Deborah Barnett is assistant dean for education at the University of California, Berkeley, School of Public Health, where she has been talking with her faculty...

    Credit the people on this episode

  5. Durable Skills in the Humanities with Chelsea Brislin, Shawna Felkins, and Jennifer Osterhage

    Sep 11, 202650 minAudio

    Share your thoughts about this episode as a text message. How can humanities courses and majors help students develop the kinds of durable skills--communication, collaboration, adaptability, and more--that employers...

    Credit the people on this episode