Chief Content Officer, Coursera
Coursera's Marni Baker Stein on the Global Impact of AI on Education | Trending in Ed
Get ready to dive deep into the future of education! Mike Palmer is LIVE from the ASU+GSV conference with Marni Baker Stein, ...
About Marni Stein
Marni Baker Stein, Chief Content Officer at Coursera, has been discussing the impact of generative AI on education and the workforce in several podcast appearances and conferences. She stated that a Penn study predicts 49% of job tasks will be impacted by generative AI in the next three years, but only about 5% of employers have an executable plan to prepare workers. She noted that generative AI and large language models are the fastest growing skill set of 2024, and that Coursera saw enrollment in this content quadruple from 2023 to 2024. Stein also highlighted a gender gap in AI learning, saying that 72% of participants in the field are men globally, and attributed this to women being underrepresented in the tech industry and having less time due to caretaking responsibilities. She said Coursera is preparing so that women have more access to technical credentials. Stein has described Coursera's use of AI tools, including an AI-powered assistant called Coursera Coach that she said increases engagement among women, learners without undergraduate degrees, and those early in their careers. She also discussed AI-driven translation of courses into 26 languages, which she said has led to over three million learners accessing translated courses and completing them 25% faster. Stein stated that Coursera does not see itself as disrupting traditional higher education but as a complement, offering universities skills-based content. She cited research showing 94% of leaders believe micro-credentials can strengthen student outcomes, and noted that Coursera had over four million enrollments in entry-level certifications in 2023, a 25% year-over-year increase.
Transcript (27 segments)
M
Mike Palmer0:12
Welcome to Trending in Education. Mike Palmer here. I'm joined today in person by Marni Baker Stein, who's the Chief Content Officer for Coursera, a company that we've talked about a bunch on the podcast. It's great to see you. Welcome to the podcast.
M
Marni Stein0:27
Thank you. It's great to be here. And we're both out here for the ASU GSB conference, which is a big deal for folks in the future of education.
M
Mike Palmer0:40
We always start by getting to know you a little bit better. Can you share with us your origin story? How you got to this point in your career?
M
Marni Stein0:46
My origin story, I would say, actually began right here in sunny California. I was at UCSB in the mid-90s when the first web browsers came out, like Mosaic. And UCSB Extension jumped on that to start their first online programs. I happened to be there at that time, and we developed a webmaster program. This is Santa Barbara, serving the central California region and beyond. I thought, my stars, this is amazing. We're offering this course that these folks would have never otherwise been able to take advantage of. This is the future. I was hooked. It became the theme of the rest of my career. I left Santa Barbara to go to Penn, where I got my PhD. I worked while getting my PhD in language learning, pushing the boundaries of online education for teaching language. Then we moved to continuing and professional education at Penn and started Penn's first online low-residency master's program.
M
Mike Palmer2:04
Is it fair for me to call you a pioneer? It sounds like there's some pioneering going on.
M
Marni Stein2:08
There's some pioneering. It didn't feel like pioneering at the time, but it was exactly that. Those early experiences took me to Columbia, where we did similar, and then to the University of Texas system, where we worked on mobile-first, offline-first access initiatives heading into 2010-2013. Based on that work, which was really competency-based and focused on talent pipelines in Texas, I went to Western Governors University, where I was provost for several years. That was an amazing experience. WGU is pointed the spear on so many innovations in equity and access. Through that work, I got to know the former CEO of Coursera, Jeff Maggioncalda. We started talking about the future of learning. AI was just starting to emerge, generative AI was just starting to emerge. I decided to join Coursera, not only to work with the incredible teams on the future of learning, but also to understand what was going on globally in terms of approaches to education policy and how worldwide leaders—government, business, university—were approaching problems at the intersection of education and work, and education and technology.
M
Mike Palmer3:49
The platform that Coursera has, you were talking about 2010 to 2013. 2012 I always talk about as the year of the MOOC, and Coursera is kind of the OG of MOOCs. Folks probably know, but more as a representative of Coursera, that story and how you got to where you are today.
M
Marni Stein4:11
Coursera was founded by two Stanford professors, Andrew Ng and Daphne Koller, whose idea was from an access perspective: how do we take incredibly powerful courses and experts and provide a learning experience around their expertise that could be offered globally? They focused first on data science, then evolved into tech, data, and business. Over time, we brought in more university partners and bright lights. It was data science and some AI from inception. That's how the flywheel of Coursera started. We are now at 168 million registered users globally. We still work with the world's greatest universities and experts, and also with incredible industry partners driving the future of technology—Google, Meta, IBM, Adobe. Between the expertise from these industries and universities, we have an incredible convening of knowledge producers that we can offer to the world. It's exciting to see how it's grown and to be a part of it.
M
Mike Palmer5:39
I always enjoy interviewing folks like yourself who don't run away from talking about AI. By spring of 2025, a lot of people get tired of talking about it, but that doesn't dismiss the fact that it's making a real impact. One of the places where I always viewed it as a direct hit on content development when the generative stuff came out. I worked at Kaplan doing that type of work for many years, and we always wanted to have a lot of the capabilities that were suddenly available to everyone in their pocket. As someone who's thinking about content development for 168 million learners, that's a pretty big constituency. AI comes in, and since you've been at Coursera, it's been a new game in terms of generative capabilities. How are you thinking about connecting that to the content work at Coursera?
M
Marni Stein6:36
Great question. I joined just after December 2022, right after generative AI hit the scene. Pretty much ever since, the content engine at Coursera has been running to keep up with the content evolution in generative AI. The thing about GenAI content—whether it's skills, tool sets, theory, or ethics—is that quarter over quarter, it's evolving and transforming. The shelf life on a lot of this content is fairly short. So there's the effort at making sure we have the right university, industry, and subject matter partnerships in place to keep driving that engine, whether developing new content or optimizing existing content. We have Google, IBM, Amazon, Meta as partners, as well as Vanderbilt and Michigan, to help us stay on the cutting edge of where GenAI is and where it's going. GenAI itself becomes a critical, powerful, groundbreaking tool for the creation and optimization of content. The domains are moving forward because of the disruption in the workplace, and the half-lives of skills are getting shorter, so you have to try to be ahead of that wave. You're connected to both industry, which identifies needs, and higher ed, where the expertise resides—though Google and Meta are starting to develop their own curriculum. Ultimately, the personalization and the level to which you can use generative tools to make it specific to the learner is probably some of the low-hanging fruit we can focus on.
M
Mike Palmer8:47
Where we started—and it's hard to think about because it's only been 18-20 months, but dog years—was really focusing on some of the most critical access problems. One is translations. We have 168 million learners around the world and serve over 7,000 industry, campus, and government partners. Language access was one of the biggest problems to solve, and it was super expensive two or three years ago. But it became clear that machine learning and AI would give us the capability to translate all our content so more folks could benefit. We now have thousands of courses translated into 26 different languages. We're not only translating videos and readings, but also the interactive navigation and coach experiences. It's toggle translation, so if someone is interested in language learning, they can toggle between English and Spanish or English and French. We're seeing three million people have accessed these translated courses, and those learners are not only more engaged but completing courses 25% faster.
It's interesting: those places where things were conceptual, like when you were talking about back at UC Santa Barbara going online to provide access to everyone, now, however many years later, around the globe, you're providing access to millions who wouldn't have otherwise gotten it. Roger Spitz talks about the Black Mirror effect—we tend to think of technology futures as dystopian, and it's easy not to notice that some of these technologies are actually world-changing. You're a global company; the scale of access you're opening up through these new tools is mind-blowing.
M
Marni Stein11:43
The other foundational AI-powered tool we set out to figure out two years ago is that we have millions of learners and great faculty, but the interaction is limited between learners and faculty. We have peer review interactions, but it's hard to scale meaningfully when you're talking millions. It would be ridiculously expensive to try to do that in any meaningful way. So we created Coursera Coach, which sits inside courses and assists learners in lightly personalized ways around navigating the course, understanding its relevance to their career or personal goals, and talking with them about what they're learning, giving alternative ways to engage with content. What's interesting from an access point of view is it's impacting precisely the audiences we've struggled to impact in any learning environment. Women, who are usually busy juggling lots of things and less likely to reach out to faculty or advisors, are 11% more likely in this coach setting to reach out, engage, ask questions, and dive deeper. We're seeing the same for learners without an undergraduate degree and those early in their career. We hypothesize that this lightly personalized, contextualized, non-judgmental space is very attractive for learners who might be intimidated in other environments.
M
Mike Palmer14:02
I've also heard it talked about in a couple different ways. There's a certain level of feeling like the chatbot, the agent, is there for you specifically, especially if it can retain knowledge over sessions and remember stuff about you. There's a mattering crisis right now where people don't feel seen or understood. Even though it's emotionally a little weird to think about our relationship with these agents, it seems like it's going to become increasingly the norm. We'll be thinking about how we deal with other humans, but also a series of agents we engage with and customize over time. How do you see this playing forward?
M
Marni Stein14:56
In terms of this mattering crisis, which has probably gone on throughout human existence, we know how important it is for learners to feel belonging in order to do well in any learning environment. That resonates with me. Our next phase is to create AI agent-driven learning experiences that really do have memory and ultimately personalize that experience with more learner inputs. You're also training the learner for a future workplace where they use a cross-section of LLMs daily. There's a level of executive function and metacognition necessary to get a chatbot to do something for a specific output. From an instructional point of view, memory and personalization allow us to have high-impact pedagogies in a scale environment. We've already developed the Socratic dialogue capability, called Dialogues, which has two-way conversations with learners to deepen understanding, personalize, contextualize, and assess where they are. We've been working with university partners on piloting this and figuring out how to author these environments. They're incredibly engaging for learners and opening doors to pedagogical approaches like roleplay, guided inquiry, and interactive case studies that before only very intimate learning environments could support. Now we can do that at scale in personalized ways. Figuring out how these new learning environments fit into preparing learners for the future of work, where they are the human on a team of other humans and AI agentic teammates, is part of the challenge ahead.
M
Mike Palmer18:00
It's almost a mindset reorientation around a few things: working with humans and AI, but also lifelong learning. Coursera must see this a lot—the idea that higher ed is 18 to 22 or 18 to 25 is shortsighted. We all need access to the skills redefinition and development that's constantly changing throughout our lives. How about on the skills side? What are you seeing there? You're kind of a canary in the coal mine around where new skills are developing and what people are looking for.
M
Marni Stein18:49
Our approach to understanding specifically GenAI, but also other skills in data science, cybersecurity, accounting, you name it, is that we look across almost 40 different career trajectories. We unbundle those into understanding job families, job tasks, and the skills associated with those tasks. That's how we populate our catalogs—so someone can come to Coursera, study cybersecurity, and take a dynamic journey of upskilling, reskilling, and skill refreshing that has integrity based on what employers demand, from external data sources and our enterprise partners. With GenAI specifically, it's important to understand not just what GenAI is, but how it's being deployed and how it's shaping today's job tasks across industries. That's the level we need to understand to provide instruction and skills verification that is useful and meaningful. It becomes tricky because employers themselves are struggling to understand how to deploy AI and how it impacts the structure of their organization and roles. We're in a dynamic learning loop with our employer and university partners. It's fascinating and complex. We're doing a good job, but we're never going to cross the finish line with this.
M
Mike Palmer21:05
It's almost like a different mindset is necessary around dealing with this stuff. We talked about harder technical skills. The other trend emerging is durable skills—relational intelligence, what we're doing in the same physical space right here is something it'll take a while, if ever, for robots to catch up to. How does Coursera think about developing what's been called soft skills? I think we're trying to move to stronger language like power skills or enduring skills. They are still the most important skills. If you look at our crosscutting skills demand across all sectors—tech, data, business, health, teaching, education—the top skills are still communication, critical thinking, the ability to tolerate ambiguity, manage change, and manage teams. These skills are a critical part of what we're doing. What we love about our content creator ecosystem is we have some creators of this new technology and industry partners, along with incredible university partners helping us think through those durable power skills. That makes for a very impactful set of influencers that helps us see this holistically.
We got to figure out who are the humans in the loop and how they fit. I want to get some closing thoughts in a second, but before we do, how do you think about the human in the loop and designing systems that include humans and AI? You mentioned AI ethics before. There's a different level of design thinking that goes into designing systems with AI agents. How are you thinking about that?
M
Marni Stein23:13
We are in close contact with not only our thousands of employer partners, but also learning from demand signals and intelligence from around the globe on how industry is actually deploying AI. In many cases, like Maslow's hierarchy of needs, they're just figuring out how to deploy AI and starting to move into how it influences structure, efficacy, roles, and teams, let alone innovating and transforming. We're fairly early stages. From my perspective, it's important that universities start thinking about integrating this intelligence dynamically into their curricula for undergraduates, and we need more interdisciplinary master's programs in AI to support leaders figuring this out. We have a wonderful long-standing collaboration with the University of Colorado Boulder, and they're developing a really cool master's in AI that's launching, looking at powerful future-forward ways of what the next generation of leaders will need to answer these questions.
M
Mike Palmer25:00
I wasn't sure which way you were going to go. As a parent of a six-year-old, I thought you went to post-secondary, master's, graduate level, but I was also thinking about developing an AI program in K-12 to start getting folks exposed. The rising generation, Gen Alpha, they're the AI generation, AI natives. It's almost too much to think about. You kind of have to start small before you go big.
M
Marni Stein25:30
I really do believe even in K-12, starting with how AI can support some of the biggest access problems in K-12 and really focusing efforts there. My kids went to schools in New York City for kindergarten through third grade that had really big classrooms and were not well-resourced to support the teacher as they tried to support all those young minds. I'm very hopeful that some of the wins we're seeing in how these tools can powerfully personalize learning will be implemented in K-12. This new AI-native generation won't think twice about how AI can help them with their learning and support their journey toward who they're going to become. It'll be so integrated in what we do—that's my pipe dream.
M
Mike Palmer26:33
You have to be optimistic, and as an elder generation, you have to not freak out. That's my general perception. There's a shortage of optimism about the future, and if we're not optimistic, things become self-fulfilling. We've had a wonderful conversation about Coursera, your background, and how you're leaning into new and emerging capabilities. Anything we haven't hit on that you want to touch on before we wrap up?
M
Marni Stein27:04
No, I love where you ended it. I think we all need to be optimistic and think of this new tool that's here to stay. It's not going away. How it fits into everything we do and helps us solve problems will make us not only more efficient, but increase the quality of what we're doing and allow us to innovate faster, better, and in more exciting ways than ever before.
M
Mike Palmer27:30
Amazing stuff here with Marni Baker Stein, the Chief Content Officer of Coursera. Thanks so much for taking your time today.
M
Marni Stein27:36
No problem. Thank you.
M
Mike Palmer27:38
And for our listeners, hopefully you enjoyed what you heard. Subscribe, tell your friends. We'll be back again soon. This is Trending in Education.