Subscribe
Share
Share
Embed
The Transform your Teaching podcast is a service of the Center for Teaching and Learning at Cedarville University in Cedarville, Ohio. Join Dr. Rob McDole and Dr. Jared Pyles as they seek to inspire higher education faculty to adopt innovative teaching and learning practices.
I think one thing I'm excited about is how can we equip institutions with the ability to build much more context aware agents that can support students in their learning in the way that the institution wants these agents to to do so.
Narrator:This is the Transform Your Teaching podcast. The Transform Your Teaching Podcast is a service of the Center for Teaching and Learning at Cedarville University in Cedarville, Ohio.
Ryan:Hello, and welcome to this episode of Transform Your Teaching. In today's episode, Dr. Rob McDole and Dr. Jared Piles continue our series on AI Literacy by chatting with Chris Cummings, senior manager at Grammarly for Education. Thanks for joining us.
Rob:Today is pretty special. We have with us Chris Cummings, and I'm gonna let Chris tell us why he's so special. But before he does that, let me just say, I have used the products that he represents significantly probably over the past, I'm gonna say eight years, eight to nine years. So I'm I'm gonna leave it there, and I'm going to, ask Mr. Cummings if he would please introduce himself.
Chris Cummings:Sure. Thanks. Very high praise in the intro, so I appreciate that. And please don't call me Mr. Cummings because I was gonna mention in my intro that my first job out of college, I was a high school English teacher. And that was the only time that that I felt comfortable being called Mr C.ummings because I needed some as a 22 year old, I needed some delineation between myself and the 18 year old I was I was teaching.
Chris Cummings:But it's great to meet you. I I'm thrilled to be here. I met Ryan, one of your colleagues, at a conference and have listened to some of your podcasts. I have my roots in education. I started teaching right out of college.
Chris Cummings:I taught English and theology at a Catholic boys school out in California, and then after that moved to the EdTech world, spent a lot of time building EdTech products, and had been bouncing around there for a number of decades at this point, which is depressing to say. And for about a year just over a year and a half, I have been I was I was hired at Grammarly. We are now superhuman. And I work with our team that manages the partnerships with some of the largest institutions we support across the globe. We work with well over 3,000 institutions.
Chris Cummings:And beyond that, we have tens of millions of daily active student, faculty, and staff users of our product. And we help align some of the products that you mentioned, Rob, have have been meaningful to you with some of the challenges that that our institutional partners are having across the world. And, certainly, things have been pretty wild over the last three years with with the rise of of AI in education.
Rob:Today, we're we're continuing on with this series AI literacy, and we thought you would be someone who could speak to this in terms of what you're seeing in the industry. Obviously, I am the products that you have are built on top of some form of AI. And to be clear, I'm not going to assume that it's generative AI, although Grammarly probably is some form of that. But I think the topic we've been exploring is is the literacy. What does it mean to be AI literate?
Rob:What are you seeing from your perspective? How would you define it? And, what are some of the most successful institutions out there doing, along these lines as well?
Chris Cummings:It's a good question. I think there's an assumption when that question is asked that there's a very simple answer. And we hear things from employers like we need students to graduate with AI literacy and AI skills. And the reality is I don't think there's a simple list of, well, you just check these boxes and build this boot camp, and then you will have a bunch of people ready for industry. And even if you did that in six months that might look radically different.
Chris Cummings:In a year, it will definitely look different. In five years, who knows what it will look like. And and I think the challenge for all of us right now is how do we instill in individuals and in the students that we work with and across the institutions we work with, the beauty, the chaos, and the importance of being human in this moment. And what does that mean? And what is your role?
Chris Cummings:And how do you navigate that? And how do you not surrender your capabilities, your creativity, your intellect to incredibly powerful tools, but how do you use them to amplify who you are, to make yourself better? In the workplace, they talk a lot about efficiency. And efficiency is fine, but you you also think back to the Industrial Revolution and you think of the the impact that that had on humans where there was tremendous efficiency gains and some really awful things that happened to a large class of people that that found themselves becoming a cog in the machine. And and I think we're dealing with a similar moment where given the power of these tools, the greatest thing we can be doing right now is is helping our students recognize the power that they have as as humans and how they can maintain their their kind of human voice in in this world and landscape.
Jared:So you're the first person --so we've asked that question a lot. You're the first person to take such a philosophical answer. Like, that's the first and I'm sure Rob is, like, grinning from ear to ear at that because Rob's all about the philosophy, and I'm more in the "alright, so, practically speaking, let's talk at my level" kind of a person. But that that that's so well put together.
Jared:But my follow-up to that is, okay. So if we ...so then how do we know? How do we know that the students we're preparing are prepared for an AI literate world if it's just a philosophical outcropping of the technology?
Chris Cummings:Very good question. I think a lot of the answer comes down to the intentional, tight, deliberate integration into experiences to test, to push the boundaries, to create, to do good things, to do things that maybe you shouldn't do into the curriculum. So Jared, you're an instructional designer. We as as AI has kind of evolved in in higher education, you know, there was a rush to build institutional policy. And I think that people thought that the policy was sufficient, and I think that the policy is confusing to students, wildly insufficient in terms of giving direction on how students should behave.
Chris Cummings:And it also put a lot of the onus on the student to figure out what they're supposed to do in the learning experience. And and as I've been talking with institutions, I I think there's a lot of alignment around the fact that, like, a an institutional policy is not enough. A course level policy is not necessarily enough. We really need to think assignment by assignment and learning moment by learning moment. How do we integrate these tools into the experience?
Chris Cummings:Where do we intentionally choose not to because of x, y, or z? Where do we intentionally ask students to try things maybe they shouldn't and reflect on it so they can learn and see? So I think some of the going back to your question, Jared, the the way you start to measure it is is almost by the exposure and integration into the learning experience that you are providing these students. I I if you were to follow-up your question by asking me what is the what is the specific assessment or test that we would do, I haven't thought about it that much to say, like, you just have to do this and then test it by doing x. But I I think the starting point is a thorough look at the entire curriculum in all departments, in all fields, because this technology and this power is is impacting and has an has an effect on all of these fields to to identify where you can build opportunities for students to experiment.
Chris Cummings:You can build opportunities for students and instructors to reflect on their experience using these tools and and give students lots and lots of exposure and build a lot of models for them about how they should behave with and interact with with AI.
Jared:The assumption then is that there's no more opting out of using the tool. It's not like, you know, you you come from the ed tech space. You created I consumed as a as a k twelve teacher. Sounds like what you're saying is it's not necessarily should we use it, but where and how.
Chris Cummings:Yeah. And I think it's less about fully opting out or fully opting in. And and, again, I think that that was an initial jump that many folks took. I'm gonna dive in and go all in on AI, and I'm super excited about this. Or I am a naysayer.
Chris Cummings:I do not believe in this. I'm going to find ways to eradicate any use of AI or any mention of AI from my course. And I think there are flaws to to both approaches. I I think the folks who are diving in strong, the issue there is is is potentially not knowing what this will look like in a year, in five years, and and needing to reevaluate some of the early decisions that were made based on assumptions around how this tool would impact humanity. And the and the naysayers, I I I think that the tool is here, and you don't necessarily need to change everything about your comfort area, how you teach, where you feel comfortable, but but finding ways to to integrate the tool or spots to integrate the tool into the curriculum given that your students are are absolutely using these tools and and experimenting with these tools regardless of of your comfort level.
Rob:So it made me think of something that I've heard here recently from an Anthropic employee who was giving a talk at a a major international conference here, like, literally within the past few weeks.
Rob:And something she said, really jumped out at me, and that was if Anthropic closed or OpenAI decided to close, she said this wheel would just keep spinning, meaning AI development, the push towards it. And she was presenting a position not too dissimilar from yours. And one that I've presented as well, Jared and I have talked about it. We've heard guests talk about it, this middle road.
Rob:We're not huge fanboys, but we're also not, like, sticking our heads in the sand saying, oh, we just can't use it. So trying to find this middle ground, but one of the things that she said was it doesn't matter what OpenAI or Anthropic does at this point if they were to jump off the wheel. In other words, they just shut their doors. It wouldn't stop a thing. And so it kinda just underlines what you were saying just a few minutes ago.
Rob:Okay? So in terms of this, though, I think one of the big questions I've also heard, and it's been, you know, going around in my mind is, who does this tool serve? How do we use it? We talk about it. What are you all doing in the industry?
Rob:Because I'm I'm assuming since we're having this conversation, Chris, that you guys are probably behind doors having conversations in your own companies because your your product depends heavily upon AI. So what are the stranding points there, and and what's surfacing?
Chris Cummings:Yeah. I think there are two areas that we are really excited to talk with institutions about, and one of them is how do we leverage the infrastructure that Grammarly has built that that provides proactive and ubiquitous support? When we're starting to talk about with with some of our product development and some of the discussions we're having with institutions is we have this infrastructure that allows us to push these suggestions to people and just show up and be aware and be helpful wherever you are without you having to do anything. Why are we just delivering your high school grammar teacher to you in those moments? Why can't we leverage additional data across the institution to push proactive support to reduce friction for users and allow them to to really be more human.
Chris Cummings:And it doesn't have to be anything terribly sophisticated. You could think about what what probably happens hundreds of times, depending on the student population, hundreds to thousands of times over the course of a term where a student starts to type a note to their instructor or their TA and says, when is the second term paper due? And this is probably information that is known and documented somewhere. It's a minor annoyance to an instructor to get five of these per assignment. It's potentially a detrimental moment for a student who has poor time management skills or or needs support there.
Chris Cummings:If they don't get an answer for three days and then the term paper is due the next day, that's very problematic for the student's potential of success. So why couldn't we provide a Grammarly underline in that moment as the student is typing and say, hey. According to your syllabus, your second term paper is due next Thursday. And the email doesn't need to be sent. The student has the information they need.
Chris Cummings:It's again, it's not terribly sophisticated, and it seems like a rather meaningless moment in the experience, but those those little moments can be really, really important. Mhmm. And we have infrastructure that can potentially help institutions alleviate some of that friction and make those moments better globally for faculty, students, and staff. So I think that's one area that that institutions are really excited about is how do we how do we proactively push kind of ubiquitous support to our community, but also give them some degree of agency. The second area that I think is is really important in in education and that that folks are talking about is in the area of the complete mystery or black box around what students are doing with AI.
Chris Cummings:And it's it's not meant in a way to be a a surveillance state or enhanced detection or an assumption that students are are are cheating. Because because I think when you start the discussion with the assumption that students are cheating, there's already an erosion of trust, and and you're in a bad spot. It's more around how can we provide students with AI support that is kind of form fitting to the learning experience they're in. And when a student goes to a generic LLM and maybe copies their assignment prompt and pastes it into the LLM, What the LLM does not have, unless the student is is really building a a highly sophisticated prompt with a ton of context, the LLM does not have any information about why the student put it there, what the student's doing, what course they're in, who their instructor is, what the expectations are for its behavior. And it just spits out a very confident sounding, you know, output
Rob:Mhmm.
Chris Cummings:That the student probably then takes and copies and maybe edits. And and think one thing I'm excited about is how can we equip institutions with the ability to build much more context aware agents that can support students in their learning in the way that the institution wants these agents to to do so. And, additionally, if we have the student leveraging these tools, it's hopefully a better experience for them because it's working with a tool that knows exactly what the goals are and what the context is of the particular learning experience. But we're also collecting information for how students are using these tools and able to illuminate some of that mysterious area right now to the institution, to faculty, to staff around how students are using AI in the learning experience. And beyond it just being a model grounded in academic integrity concerns, I I think and, Jared, you may agree or disagree with me.
Chris Cummings:I I think it's really a model that that could be grounded in continuous improvement of the learner experience. Because if we have a ton of insight into what students are doing, where they're struggling, what's not working well. We can take that information, and we can use LLMs to help us modify the learner experience to to make it better for the next cohort.
Jared:So I I do wanna go back. You mentioned humans in the loop. That's something we've heard a lot about. And you've mentioned, agency, you know, tying I love the idea of it being something that like like a Grammarly like tool that shows up and answers questions informed by the syllabus or whatever so that before the student can even fire off the email, the answer comes up to the question about when the next term paper is due. I'm wondering what other human skills are going to become emphasized, really the spotlight been shown on it as these students enter into this AI enabled workplace?
Jared:What is it something that they really need to hold on to besides the agency idea?
Chris Cummings:I'm glad you asked, and this will this will be a bit more practical. But this was a really interesting one, and it I heard it said at a conference, I that the most critical skill for humanity moving forward will be clarity in written communication. And that's because we interact with these tools using plain, you know, English or your native tongue. And I think we've all had the experience where we put in a prompt, we get an output, and we kinda go, why did it give me that as the output? And then you go back and look at your prompt, you go, oh, like, I wasn't really clear at all in terms of what I wanted, how I wanted this to to be.
Chris Cummings:And and that in and of itself is a really powerful learning moment about, you know, how important that clarity of communication is. And and so I do think that that that is a really important skill that we can instill in our learners today is is how to be precise and clear and give context in communication. You think about if you are a manager giving kind of a a directive or making an ask of of one of your direct reports. If you're even even if you're on Zoom, you can kind of see head nods or or get some signal based on body language of this person is ready to go or they have no idea what I'm talking about. We don't get any of those visual cues from these tools.
Chris Cummings:Like, we and they just come back with an output saying, you know, with tremendous confidence and optimism, I did it. You know? And I think we've also all had the experience of of looking at output from an LLM and saying, yeah. But that doesn't look right. Where did you get that?
Chris Cummings:And the LLM immediately responds, oh, that's right. Yeah. I completely messed that up. You know? But I I think giving students lots of experience like that to to see the value of clarity in their direction and in their written communication.
Chris Cummings:And and also those moments where where they have those experiences with LLMs and and and recognize their limitations and recognize how important their voice and their mind is in the in the communication with with the tool.
Rob:Well, so far, what I've heard, just to sum up our conversation here a little bit in terms of bringing it back to the AI literacy frame. One, the human needs to be in the center. They need to be able to actually own and make decisions with what the AI is doing. And then two, we need to provide ways to build trust so that we're not separating trust within. Because if you don't have trust between individuals, you can't have free exchange of anything.
Rob:And our economy, our society is built on trust, some sort of trust.
Chris Cummings:The one other thing that I think is really important for and this is a really cool example of of how one institution is using custom agent builders in in an innovative way, I think. I think we need to constantly be mindful of and think about ways that we can nudge or prompt human to human interaction. So if we build an AI tutor for a biology course, what language can we put in the underlying prompt so that if a student gets to a certain point, the the tutor says, hey. You should probably talk to your instructor about this. And they have office hours next Wednesday, and you can even click this link and sign up for twenty minutes.
Rob:I like that.
Chris Cummings:I I think that these these tools can be so good that that that's one of the things that I always think about is is and I have three young kids, and I think about it a lot is the importance and the value that I place on human to human interaction and how much is learned and how much of a challenge it can be. Asking one of my kids to talk to their counselor about something can be really challenging for them, but I think it's a really important thing for them to do as opposed to me sending a letter on their behalf to their counselor. So as we think about building agents, building tools to support learning, how can we build into the underlying model opportunities to to nudge people to interact with each other.
Jared:It's really good.
Rob:So now we got three major things that probably need to be at least a part of the AI literacy conversation, trust building, person to person interaction, engagement. It's kinda funny saying that, but at the same time, I think that it should be. Like, if you're gonna have human which goes probably it's probably a function of the human in the loop. Maybe it's a subset to that, but I think it's, something definitely worth thinking about in terms of, like, if you're trying to build an AI literacy framework, those I think those things most certainly should should be a part.
Chris Cummings:Yeah. And and clarity clarity in written communication.
Rob:Mhmm. Yes.
Chris Cummings:Like, really, really important.
Jared:Well, Chris, we do really appreciate your time and giving us plenty of stuff to think about. The idea of a human in the loop is such an incredible, important part of this whole process that sometimes we forget about. So
Rob:Yes. Thank you.
Ryan:Thanks for listening to this episode of Transform Your Teaching. If you have any questions or comments about our chat with Chris Cummings, feel free to send us an email at ctlpodcast@cedarville.edu. You can also connect with us on LinkedIn. And don't forget to check out our blog at cedarville.edu/focusblog. Thanks for listening.