In early 2023, Jeff Maggioncalda, CEO of Coursera, started developing the EdTech firm’s strategy for incorporating generative AI into their offerings. He asked his teams to focus on value to the firm and cost of implementation, and they identified four key projects: powering translations and modifying content format and delivery, personalized coaching, an automatic course-building tool, and building out new GenAI-related academic content.
By early 2024, the firm had made significant progress in bringing these capabilities to market, but GenAI was evolving quickly and Coursera needed to continuously improve its offerings. While the firm had been an early mover, competitors were adapting fast.
Was Coursera taking full advantage of the opportunities presented by the technology? What more could it do to remain competitive? Harvard Business School professor Suraj Srinivasan discusses those questions in the case, “Coursera’s Foray Into Gen AI.”
BRIAN KENNY: Since ChatGPT appeared on the scene back in the fall of 2022, newsfeeds have been flooded with claims that generative AI is going to change the world, so if you’re suffering from a bit of GenAI fatigue, I get it, but even if it’s not readily apparent to you in your day-to-day work, AI is steadily increasing its presence in a lot of workplaces. According to a CompTIA study, 75% of the Fortune 500 are either implementing or seriously considering exploring AI programs. Adoption spans almost every sector, but it’s very prominent in a few, including education. A recent survey of 10,000 executive learners found that half the respondents expect to return to the classroom to learn about AI within the next five years. Educators, including Harvard are eager to meet the demand, perhaps with a little help from GenAI itself. Today on Cold Call, we welcome Professor Suraj Srinivasan to discuss his case, “Coursera’s Foray into GenAI.” I’m your host, Brian Kenny and you’re listening to Cold Call on the HBR podcast network. Suraj Srinivasan examines the institutions of corporate governance in the US and internationally, and I think you’re a “three-peat” visitor to Cold Call, Suraj, Welcome back.
SURAJ SRINIVASAN: Who’s keeping count, but yes, it’s fantastic
BRIAN KENNY: It’s great to have you back on the show.
BRIAN KENNY: You suggested this case to be on the show, and I was really glad that you did because there’s such an appetite for anything having to do with GenAI, people really want to hear about it. I think people want relatable examples about how it’s being used and how it’s being implemented, particularly in the workplace because we’re all trying to figure out how do you embrace this technology, what should you stay away from. I’d like to know why you decided to write about this and how it relates to your work because it doesn’t seem to fit neatly into the kind of things that you research.
SURAJ SRINIVASAN: Thank you for having me, Brian, this is a favorite place to share my research and my case writing. A lot of my current work has been in the areas of data science and AI. This case illustrates the range of opportunities and challenges that comes with this technology. That really is the central issue in this case is, how does generative AI create opportunities and threats or challenges for the business model for this company, Coursera, which has been around for several years, prior to this technology, a new technology comes in, it’s going to dramatically impact the business model. The central issue in this case is, how does it impact the business model, how would you, if you were the CEO of the company, think about both the set of opportunities that get created and the set of threats and challenges that get created?
BRIAN KENNY: Yeah. You’ve been studying this for a while and you’re teaching about it, and you heard my introduction, I teased a little bit about GenAI fatigue, but I’m wondering, just your opinion about whether or not is this overblown, all the stuff we’re hearing about GenAI or is it really going to be the kind of game changer that people say it is?
SURAJ SRINIVASAN: I think the latter, that it is certainly already proven to be a game changer in so many ways. You and I and people who use ChatGPT or similar tools in our day-to-day lives, see where it can make an impact, but if you take those individual experiences of ours and put it into the scale where it can start affecting organizational methods of working, the kinds of things that can happen inside businesses, that can create a fairly dramatic effect. GenAI, even more than traditional AI or some of the mobile and the cloud technologies that preceded it has already shown its impact much more than some of these prior technologies have, so in that sense, the impact of GenAI, what you’re seeing, what you’re reading about, what you’re hearing, the issues that you talked about, is reflecting that this technology seems to be potentially even more impactful than some of the precursors to this that we have seen in the last 20 years or so.
BRIAN KENNY: All right, so it’s incumbent on all of us to learn more about it, I guess so?
SURAJ SRINIVASAN: Absolutely, not just learn about it, think about it in an organizational context, but very importantly, incorporate it into our own personal and professional lives in what we do, because that’s how, I think, that kind of comes through in the case as well. We have to know what it can do personally, hands-on, to know what it can do organizationally.
BRIAN KENNY: Yeah, that’s great. For those of our listeners who don’t know about Coursera, can you tell us a little bit about them, what they do and how they evolved?
SURAJ SRINIVASAN: Coursera is an ed tech, an education technology company, so essentially they have a large menu of courses. I think around 4,000 or so. You can sign up for those courses, these are all delivered virtually. Coursera provides the platform for these courses. The courses themselves are created by the likes of Harvard and University of Michigan and Berkeley and so on. These are professors at all these various universities, and in some cases, private companies like Google and Microsoft and others who might be creating courses. And then students, executives can go and sign up for these courses. It started with the massive online courses that the MOOC phenomenon now or a decade or older, that’s where the whole idea started for companies like Coursera where students can go and sign up for a course, you can take the free version of the course, almost all courses are available for free, or you can get the paid version, the paid tier which offers fewer additional features like certification on top of the content itself, which mostly is available to the free version. That’s the business model, students pay for when they get certified, and then they do a lot of work with enterprise customers, training and L&D, learning and development work for businesses and others. A large customer group for them is universities in the US and around the world that might have their students take courses or portions of courses through Coursera for course credit at that college or university.
BRIAN KENNY: Okay, great. The CEO, I’m going to butcher his name I’m afraid, but Jeff Maggioncalda-
SURAJ SRINIVASAN: Maggioncalda, yes.
BRIAN KENNY: Maggioncalda. Great. Tell us a little bit about him. What’s his background?
SURAJ SRINIVASAN: Jeff, I got to know him well writing this case, he’s a very charismatic, enterprising person. He studied at Stanford, undergraduate and then his MBA. His first job right out of his MBA was to start a company along with the Stanford professor, and it was a financial technology business that he started, it’s called Financial Engines, that he ran for many, many years and then moved to Coursera in 2017.
BRIAN KENNY: There was an expression that he used that I was really intrigued by, the idea of conscious competence. Can you talk a little bit about that?
SURAJ SRINIVASAN: Jeff is a very intellectual CEO. He’s a doer as Financial Engines and Coursera has well shown, but he is also very thoughtful, most CEOs are, but I was struck by Jeff’s thoughtfulness in terms of how he thinks about things. This idea of conscious competence, the way Jeff described it is when things work, to know why they worked, and when they didn’t work, to understand why they didn’t work, essentially, as a way of developing your own theory of organizational change or organizational strategy execution. If you try to understand what caused a good outcome or a bad outcome, then over time you develop enough data points to understand why something works and why something doesn’t work. Essentially, it is developing a theory of execution by consciously trying to understand the reason why something happens, or why something works inside a business or it doesn’t. He’s thoughtful enough that he’s coined his own way of framing it.
BRIAN KENNY: It’s a great term! Yeah.
SURAJ SRINIVASAN: Exactly.
BRIAN KENNY: He couples that up with situational awareness, that’s where it relates back to his ability to look at something like GenAI and see, wow, there’s something here, we need to be paying attention to this. That seemed to crystallize for him around COVID. That was, I think, a turning point for him as he thought about what they needed to do at Coursera. Can you talk a little bit about that?
SURAJ SRINIVASAN: For them, much as in GenAI, which we’ll get to in a second, COVID would be a threat as well as a way to contribute at that point in time. A threat because we don’t know what in the early stages of COVID, what does it mean for learning, when schools and colleges and universities are going to be forced to shut down, what will it mean for an educational content platform like Coursera? I think very quickly they realized that when everybody’s going home, the way they can contribute is now magnified by making Coursera available for learners around the world. That’s what I think he meant by situational awareness on how companies and CEOs should respond quickly to the moment when you see something dramatic and big potentially going to happen so that you’re not behind the curve but you’re ahead of the curve. Now that you started the podcast with talking about late fall of November of 2022 when ChatGPT became the thing-
BRIAN KENNY: Yeah, overnight.
SURAJ SRINIVASAN: Overnight, literally. Not that, that GPT-3 and two and so on, and the idea of transformers and all had been around for almost five years at that point, with people who are deep into that technology, but as it exploded into public consciousness, for a CEO of a company like Coursera, it becomes a moment of situational awareness. What does this mean for us? What does this mean for my business? What are the challenges? What are the opportunities this is going to put in front of us?
BRIAN KENNY: What was Jeff’s initial take on it when he first saw GenAI? Did he see it as an opportunity or as a threat?
SURAJ SRINIVASAN: I think both, and that’s what actually makes this case. If I go back to the thing we were talking about a few minutes ago, that’s what makes this case a wonderful example of the potential and the challenges that come with GenAI. I think the company, Jeff and the rest of the team at Coursera quickly recognize that this could be a game changer, both in terms of what the technology allows them to do and the risks and challenges that it might create.
BRIAN KENNY: Yeah, and we’ve heard a lot of people make comparisons to the dawn of the internet when businesses were trying to figure out is this something that we should be paying attention to? Is it not? In many ways, kind of taking a step forward and feeling around in the darkness to see what are we going to encounter here. It feels like this is the same thing, but only this feels accelerated. This feels like it’s happening faster.
SURAJ SRINIVASAN: Absolutely. If I may take a moment now to just lay out the arc of what all is possible, the framework if you may, of what is possible with GenAI, and that’s the framework around which I built my course and how Coursera helped lay out that framework. As you pointed out just now, this tool, generative AI as seen through ChatGPT or anything else, can have a pretty big impact on individual productivity, on doing things better and faster as individuals. If I scale that in an organizational context, if everybody in an organization starts doing it, starts doing better and faster, what will the impact of that be? There’s going to be tremendous productivity and efficiency gain in an organization. Coursera, for instance, quickly realized, and it’s now being seen widely, that software engineering, coding, writing code is something that GenAI tools are very good at, or as an assistant, as a copilot for your programmer. And one of the things they did was to create a paired programmer, essentially a software tool, GenAI based tool that can help Coursera. If you think about Coursera’s business, the content is created by the Michigans and Harvards of the world. They provide the platform, so it’s a highly tech..
BRIAN KENNY: Yeah, it’s a tech play for the most part.
SURAJ SRINIVASAN: Yeah, exactly, and so a lot of their employees are software developers, software engineers, so that becomes a tool for doing things better and faster. Now, these are things human beings can do, you and software engineers can do, but if you can do things 20, 30% better or 20, 30% faster, that’s roughly the kind of estimates we are seeing across businesses. What does it mean to my business model if I can start doing things better and faster, but at scale? Now, there are a couple of examples from Coursera where, to illustrate this point, Coursera is primarily or was primarily an English language product. The content was all mostly in English. Now, if you’re limited by English language, then you have a certain audience base, you might translate, but it would cost them about $10,000 to translate from one course from English to another language, so you wouldn’t do it for the entire portfolio of 4,000 courses…
BRIAN KENNY: It takes a long time too.
SURAJ SRINIVASAN: It takes a long time, exactly. It takes a long time and it costs a lot of money. What this technology now allows is translation to give you a dramatic scale of impact, what would cost them $10,000 in several months to translate could be done for a cost of $10 in just a matter of a few days. I won’t get into the details, but the quality of translation can also, using AI, be dramatically improved. This is at scale, a course will be in several hours, and then now you’re translating it from let’s say English to Spanish. You translate that. The cool thing about how these technologies work is that you can translate from English to Spanish and then translate back from Spanish to English, compare English and English, and then you know how good the translation is.
BRIAN KENNY: Wow.
SURAJ SRINIVASAN: If you see the quality of what you got back is similar to what you put in, you know the Spanish is correct as well. That’s just a small, interesting anecdote. I get fascinated by how quick this loop could be.
BRIAN KENNY: No, that’s great.
SURAJ SRINIVASAN: What they ended up doing in a matter of eight months or so is translate all their 4,000 courses into 25 languages. Imagine, one course to one language cost that much, but now it is beyond just productivity improvement, this is a business model strategic change that from being primarily a course where people would interact with you in English, now you can sell this product around the world, you can go to Malaysia, you can go to Brazil, you can expand the top of your funnel, which means the students who are coming in…
BRIAN KENNY: You are creating value now, you’re not just adding productivity.
SURAJ SRINIVASAN: Exactly. If your assets, what creates value for you is courses in English, now, the value of those courses just exploded because now you can essentially help students around the world in any language, learn from them. This is now a strategic change in the business model where and how you can sell. You could become a global company. Your markets are now around the world. What does it mean for your sales force? What does it mean for your go-to market, and so on? By the way, this is still something that humans can do, translate..
BRIAN KENNY: Of course.
SURAJ SRINIVASAN: But it’s now being done at a dramatically lower cost. We started with productivity improvement at individual level, productivity improvement at an organizational level, strategic change that comes from… This is still productivity improvement translation, but at a scale where it means your business model can change. The last piece of this, that Coursera illustrates in the course that we covered was things that actually humans can’t do at the scale and cost that would be needed. Let me give you an example. Coursera has a product, one of the holy grails in education is customizing the learning for a learner. The fascinating thing about this technology as applied to Coursera and tech businesses is this oxymoron, which is, customization at scale. I call it oxymoron because customization is like personalizing and then doing it at scale.
BRIAN KENNY: Of course.
SURAJ SRINIVASAN: Now, that’s not something that a human can replicate because you can’t customize at scale. The last piece of this GenAI spectrum, the framework of what is possible in the technology is really this innovation that comes from the features of this technology, which is not just doing things better and faster, but doing things which were not possible before. One of the things that I illustrated with this case is this product they have called Coach, which essentially is a synthetic tutor that will live along with the code, follow the learner along, and help somebody learn. Even in the period that I’ve been following them, this product has been improving and they are learning as they improve their user interface with the learner, but the essential concept of this is, if a synthetic tutor can be there on the learning platform trying to help you learn, then how great would that be for a student that is trying to learn about it? One of the things that I learned from them is that students are more willing to ask naive questions. The things that students will sometimes say, this may be a stupid question, and I say, there’s no question that’s ever stupid if it helps your learning. Students are typically very reluctant to ask those questions of an instructor, of a live human instructor, on the other side, they’re much more willing to ask these simple questions, which actually holds back a lot of learning, to a AI tutor because nobody’s judging you from the other side.
BRIAN KENNY: It makes perfect sense.
SURAJ SRINIVASAN: Really, what I wanted to illustrate, you asked me, is this technology going to be dramatic? This is a range of things that is possible, which is individual level productivity that can scale into organizational productivity, that can scale into strategic choices or strategic opportunities that can come your way in terms of business model impact, and then things that you couldn’t have done before at all, like innovation, inventing new ways of actually engaging with your customers.
BRIAN KENNY: Yeah, and recognizing what those ideas are is part of the challenge. They set out to do this, but they did so with some guiding principles, and I know that we’ve worked hard to think about principles at Harvard Business School for how we can use AI in a safe, productive way. Can you describe some of the things that Coursera thought about?
SURAJ SRINIVASAN: As you can imagine, as you alluded to, there’s a range of ethical issues, responsible AI type issues, the risk that comes with both the technology and how it is applied. Again, that was an important part of my course. For their part, they listed out five key principles that they would want to apply as they adopted GenAI through the business, and much of these will resonate with you and with other businesses. The first was, this should be a positive impact. I guess, it goes without saying, but it should help learners, educators, and society at large in terms of what the technology would do. The second was safety and security. The AI system should prioritize safety, protect the security and privacy of the users and their data. The third was fairness. AI systems should empower everyone and be fair. The operative word being everyone and operate without bias through the algorithms that the gen AI is using. The fourth was transparency, that the company should be transparent about both the capabilities and the limitations of what is possible so that the users, both the learner users, the students who are coming there, as well as the educators. If a faculty, if a professor somewhere has a course they’re providing through Coursera, they should know how Coursera is using this technology vis-a-vis them, because you could think about one of the risks that the company faces, which is a benefit as well as a cost for tech more generally, is that the cost of content creation is dramatically falling because of GenAI. What does it mean for a business that essentially creates value through content that they get or that they curate and deliver to large numbers? If the cost of content creation falls, why does it make sense for me or any professor at Harvard or Berkeley or any other school to create content that they will then try to monetize through Coursera? If the cost of content creation falls close to zero, then who’s going to pay for it? Why would I be creating content? That’s a big, big issue. The last one was on accountability, that they wanted to be themselves, held accountable for the performance of these AI systems, the quality of these AI systems and the impact. What this would mean in terms of execution was that they would commit to monitor and evaluate the effectiveness of these technologies, any unintended consequences that might be coming along, so this wouldn’t be a case of, move fast and break things and then we figure out, here you would have to move fast but very carefully so that you are keeping an eye on performance, risk, bias, quality along with rolling this content out.
BRIAN KENNY: The case also speaks to the fact that they had to upskill their own community. We’re facing this. I think every organization is trying to think about, gosh, how do we learn about this? How do we get everybody up to speed? They approached it in an interesting way. Can you describe what they did?
SURAJ SRINIVASAN: Well, few things. I mean, imagine you’re the CEO of a company like this. You have to first learn it yourself.
BRIAN KENNY: Right.
SURAJ SRINIVASAN: Jeff started essentially being at the front of the technology, starting to adopt it themselves. They started insisting that their organization, people who are in Coursera start adopting various elements of generative AI, getting accounts in ChatGPT for instance, so that everybody starts understanding the potential of this. That’s what I keep telling my students and executives that go through our programs and companies that I work with, you’re not going to know what this is. This is like using Google, unless you use it hands-on, whether it is creating a PowerPoint slide or creating your own avatar and having it do something, unless you do it yourself, it’s hard to comprehend what it means for the organization. Among other things, this is one of the things that they started doing. So one of the things they also did was actually create a course, which I think this is probably the only course that Coursera itself is the creator for, because otherwise they are essentially a platform for others’ courses.
BRIAN KENNY: Right.
SURAJ SRINIVASAN: Jeff created a course for CEOs, he called it Generative AI for CEOs, bringing to the course what he learned or how he uses it as a CEO, but that’s important, I think, for anybody listening to this, as you and I have been doing and many of our colleagues here, to start becoming hands-on users of the technology even beyond the basics of just use ChatGPT’s user interface, start doing slightly more sophisticated things with it so that you can actually see what…
BRIAN KENNY: It’s kind of fun once you start, and you just need to be curious, I think, and once you start playing with it, you get some really interesting things.
SURAJ SRINIVASAN: And you understand limitations, as much as the, I think I might have focused too much on the benefits or the opportunities that come with it, every one of these opportunities comes with a flip side. Technology is moving fast, so it creates a lot of basic questions about how quickly should I move if the technology is moving this fast, is it actually better for me to wait and watch, to see where things end up?
BRIAN KENNY: I was going to ask you about that. Is it worth it to be the first mover in this or are you better off just hanging back a little bit and letting somebody else make the mistake?
SURAJ SRINIVASAN: This is a really important question. Here’s one way, it’s a complicated issue, it’s not a simple way of thinking about it. What you have to consider is, where might this technology keep heading, if you’re an entrepreneur, that might be a risk of, where is the technology headed, and what I’m trying to do, if it’s going to become so much easier for the model builders, the open AIs and Metas, Googles of the world to incorporate into the model itself, then it takes away the reason for my business to exist. For a larger enterprise like Coursera or let’s say a large bank or any other bigger company which is trying to embed these technologies into their existing products or trying to think about new products, here’s one consideration, one important consideration. If speed of customer adoption, if customers adopt, and how you can build on that, either through network effects or just through customer familiarity is an important factor, then you need to be ahead, then you need to embed technology into your existing products. Customer adoption will either bring more customers to you or you will learn more about how customers are adopting, which will help you move and scale faster. In fact, in Coursera’s case, that’s the view that Jeff took is, let’s focus on today’s value addition, let’s not worry about long-term sustainability of this technology because things are going to move fast, we are going to learn, customers or learners are going to learn, and so it helps to move faster to be the earlier adopter in that case. If, instead, your product or service can lose its distinctiveness when the next version of the technology comes, when there is a risk of getting quickly upstaged by where the technology is headed, then it might make sense to watch a little bit about where things are going. There are other kinds of use cases, where risk of hallucination or providing wrong information to customers outweighs the benefits, and so if that risk is harder to mitigate, then you’re better off waiting and watching and not have customers pay the price for your speed. Let’s not move so quickly here because we still don’t know how we are going to manage the risk here.
BRIAN KENNY: Something tells me we might see some of those examples somewhere down the line, that become Harvard Business School cases and we can talk about them on this show.
SURAJ SRINIVASAN: Absolutely. I’m sure there’ll be no dearth of that, both on the opportunity side and the risk management side. This story is still being told. That’s the reason I’m excited about it, in fact, the reason I created the course. Like you said, Coursera’s example, we learned by doing, we learn by understanding what’s happening around the world and how this technology is being deployed and we are doing that at HBS in so many ways. That was the reason I did this course and I really, really enjoyed this case a lot, writing it and teaching it.
BRIAN KENNY: It’s been a great conversation as it always is when you’re on the show, Suraj. One last question for you, which is, if you want our listeners to remember one thing about the Coursera case, what would it be?
SURAJ SRINIVASAN: That’s a great question, Brian. It’s so many things and I’m going to just pick…
BRIAN KENNY: I’m limiting you to one.
SURAJ SRINIVASAN: One, I know. I think the most important thing that I learned from this, and which I hope will be a takeaway from this case is that this technology for all the pros and cons, the risks and rewards that come with it is highly, highly impactful. This company, this example, gives us a model of how to understand the potential as well as the limitations and how to roll it out inside an organization. I think there are lessons from that, for us as individuals, on how should I learn this technology, what should I be learning? There are lessons here as business students, as executives on how do I take something that is nascent, that is new, that is novel, and understand it quickly enough, probably not wait to master it, which is the first mover versus second mover question, but start thinking about, start applying it in meaningful ways that can have an impact on us, on our customers, in some cases, the largest society as well. You will face, we will face, as this technology grows, many moments which are pretty impactful. It is incumbent on us to get on top of the technology and start learning both the merits and the drawbacks.
BRIAN KENNY: Suraj, I’m going to share a secret with you and our listeners, which is that I used ChatGPT to help me write the introduction to this very podcast.
SURAJ SRINIVASAN: There you go. As you read in the case, the project Genesis, which is what Jeff called their GenAI initiatives, came from ChatGPT. He asked what should we call it, and I guess it gave him the suggestion
BRIAN KENNY: There you go. Thank you for joining me on Cold Call.
SURAJ SRINIVASAN: My pleasure. This was so enjoyable and I’m glad you enjoyed the case.
BRIAN KENNY: If you enjoy Cold Call, you might like our other podcasts, After Hours, Climate Rising, Deep Purpose, IdeaCast, Managing the Future of Work, Skydeck, Think Big Buy Small, and Women at Work, find them on Apple, Spotify, or wherever you listen. And if you could take a minute to rate and review us, we’d be grateful. If you have any suggestions or just want to say hello, email us at [email protected]. Thanks again for joining us, I’m your host Brian Kenny, and you’ve been listening to Cold Call, an official podcast of Harvard Business School and part of the HBR Podcast Network.
