Reinventing the Lecture
Stanford University is blending the online and in-class experiences
Steven Cherry: Hi, this is Steven Cherry for IEEE Spectrum’s “Techwise Conversations.”
Lawrence Summers, the controversial former president of Harvard University, recently had an op-ed piece in the New York Times in which he made a telling comparison. He wrote,
There was a time when professors had to prepare materials for their students. Then it became clear that it would be a better system if textbooks were written by just a few of the most able: faculty members would be freed up and materials would be improved, as competition drove up textbook quality.
Similarly [he continues], it makes sense for students to watch video of the clearest calculus teacher or the most lucid analyst of the Revolutionary War rather than having thousands of separate efforts.
Back in 2002, the faculty at MIT first put up 50 online courses in what came to be called OpenCourseWare. In 2010, they passed the 2000 mark.
Last year, these efforts began to coalesce when Stanford University created three online computer science courses that used a format similar to the Khan Academy videos. There was a ton of excitement when they were first announced—here at Spectrum, one of our bloggers wrote about it with the headline “You (YOU!) Can Take Stanford's 'Intro to AI' Course Next Quarter, For Free” with the second you in capital letters and an exclamation mark. Three hundred thousand students were equally excited enough to sign up.
Daphne Koller is one of the two leaders of Stanford’s online efforts. She’s a professor in the Stanford artificial intelligence lab and wrote about the Stanford experience in her own op-ed in the New York Times last month. She’s my guest today, by phone from Palo Alto, California.
Daphne, welcome to the podcast.
Daphne Koller: Thank you, Steven.
Steven Cherry: Maybe first you could just describe the format of the Stanford online courses.
Daphne Koller: So the format we’ve chosen to adopt is a format where the primary content, the primary mechanism for conveying the content to students, which has traditionally been the lecture where the professor stands and just talks at the students—we’ve taken that component and moved it to an online format which in fact we believe is more engaging to the students than just sitting in the classroom listening to the instructor speak. So the video is edited to correspond to relatively short chunks, kind of Khan Academy–style, maybe 10 to 20 minutes depending on the topic, and interspersed in the video stream are “in-video quizzes,” as we call them, where the video pauses and the student is supposed to engage with the material by answering a short question. This is very valuable experience for the student because they can tell whether they’re tracking the material or not, and it’s actually more engaging than sitting in the classroom because in the classroom, usually only one student raises their hand and answers the question, and most of the students don’t get a chance to think about that before it happens. So in addition to that, we have a lot of automated assessments that happen at the end of these modules, where students get to test their understanding of material and demonstrate mastery, and that really enhances their comprehension and retention of the material. We’ve developed auto-graded assessments for both multiple-choice but also programming assignments as well, where students can try out a short program, see if it gives the right output, and they’re getting immediate feedback from the system as to whether they’re getting things or not. Now at the same time that we’ve moved all that content out of the classroom, we’ve freed up this valuable classroom time for much more meaningful interaction between the faculty member and the student. So they can work together on solving problems, together with each other or together with the instructor. You could do case studies, you could bring in guest lecturers that can motivate the use of the material—for example, in practical applications downstream—so as to keep things motivating. And there are so many studies that demonstrate the benefits of these active learning strategies to student engagement and retention of material. And so you’re really getting the best of both worlds; you’re enhancing both the conveyal of content as well as the opening up room for more active engagement.
Steven Cherry: Those assessments and exercises also tell the student whether, like, he really knows enough to race ahead through a course or to sort of slow down and spend more time.
Daphne Koller: That’s exactly right, and that feedback, that instant feedback to the student, is so important because when you think about how things are handled now, you teach the material one week, the problem set goes out the next week, it’s due the week after that, takes a couple weeks to grade—by the time the student gets the feedback, the material is so long gone that they don’t really have a sense of whether they really understood or what they didn’t understand, what they need to go back and relearn, so this instant feedback is such a critical component to student learning. And it allows a student to take a different path through the material. They suddenly realize they’re missing some background—they can go and watch some extra modules, they can go back and rewatch the stuff that they missed the first time. And I think that really provides a much more in-depth learning experience.
Steven Cherry: So you called this—in your op-ed, you called it a “blended approach.” I’m curious about one thing, and I’ll just say by way of personal disclosure, I teach creative writing and essay writing as an adjunct instructor at the College of New Rochelle here in New York. Do you think this “blended approach” would work in a writing class?
Daphne Koller: So okay, I don’t really know about a writing class—that is something that I hadn’t thought about. But we’ve spent some time engaging with instructors of humanities classes of other types, and we think there’s some really exciting ways in which the online component that precedes the in-class component can really be used to ensure that not only do students have the basic background to participate in a meaningful in-class discussion but also to prepare their minds for the kinds of questions that the professor wants to raise as part of the discussion. And so it’s not just a matter of, did you do the readings? but really, are you thinking about the right things? And so I don’t know about writing—I think that’s an interesting question that I would have to think about—but I do think that this blended model, as opposed to a purely online model, could really be useful not just in the computer sciences and the natural sciences and engineering.
Steven Cherry: Now, I guess this can work outside of the college setting as well. In your op-ed article, you describe a seventh-grade remedial math class in northern California that adopted this blended approach.
Daphne Koller: Right. That is Khan Academy, and Salman Khan deserves [a] tremendous amount of credit for taking his online content and thinking about how to use it not just as a purely separate experience for high school and middle school students but also as something that enhances their classroom interaction. All my knowledge of this comes from reading media reports—I don’t have any particular insight into this from a personal perspective—but what I’ve read about sounds incredibly impressive, and I’m a great admirer of Salman Khan’s.
Steven Cherry: Yeah, it sounds like they have some data that shows that it really worked.
Daphne Koller: Oh yeah, the data seems really compelling. And it seems compelling in the remedial classes as well as the more advanced students that really want to just race ahead with the material.
Steven Cherry: So it seems to me that there’s going to be—maybe there already is—some resistance in some pockets of academia to this. There was a recent article at the website Higher Ed, and it had the headline “Online Grows, Doubts Persist,” and it cited a study by the Babson Survey Research Group and the College Board that found that a minority of faculty is, quote, “concerned that the quality of online instruction is not equal to that of courses delivered face to face.” Have you seen any of that resistance yourself at Stanford?
Daphne Koller: Well, I mean I haven’t done a comprehensive survey of faculty members to try and see how many of them would be enthusiastic about this or not; I would expect that as with pretty much anything you’re going to find people who think it’s a great idea and people who don’t. We’ve had a lot of enthusiasm about adopting this technology among our colleagues, and people are really excited about this as a direction.
Steven Cherry: Daphne, the MIT experience seems to be that online courses run $10,000 to $15,000 apiece, and presumably the blended approach would be more in effect if you count instructor time. I don’t know if that saves money compared to a regular classroom of today, but it sure isn’t cheap. Do you have any understanding yet of what this is going to cost in terms of time and money?
Daphne Koller: I think that if a system is well designed, one can bring the costs down considerably these days by utilizing, for example, cheaper streaming technologies. I can’t say as to whether MIT has done that or not, so I think that it requires careful design to keep the costs down. But I think that also once you produce a class at this quality of production values, you can now reduce, for example, the on-site efforts for instruction. So—for example, my TAs right now are spending a lot of time designing auto-graded assessment, both quizzes as well as programming assignments. That is really a significant investment of effort, and I should be very grateful to them, they’re doing a spectacular job. But once they do that, they don’t need to grade any more—at least, most of the content will not need to be manually graded. So they will not have to put in that effort, which is the part of the job that TAs like the least, and even more importantly, downstream generations of TAs will not have to do any of that because it’ll all be there already. And so there’s an initial investment of effort up front, but I think that investment amortizes over multiple course offerings. The same thing happens with the instructor: You teach your class on video once, you edit it, you make sure it has high production values. These interactive sessions, you don’t want to do that three times a week because that just increases the load on the students because they have to watch the videos and come to class three times a week. So what I’ve done is, I’ve had one longer class session just once a week where we have all this interactive engagement, and the rest of the time they just do the videos. And so I don’t think that ultimately it’s going to be—as you amortize this over multiple course offerings and you factor in the savings, if you factor in instructor and TA time, that it’s going to be a loss.
Steven Cherry: So a video’s watched once together as a larger group? Or are they always just watched individually?
Daphne Koller: Well, we don’t mandate that; we certainly don’t ask them to come to class to watch the videos. We’ve seen some students who prefer to watch in their dorm room at two o’clock in the morning, and others that get together in small groups and watch together and discuss, and I think it’s great when students do that.
Steven Cherry: Daphne, on a personal note, you have a pretty remarkable background as a student as well as a teacher.
Daphne Koller: Yes, that’s true, and I think that my background as a student is one of the motivating factors for me in trying to get education to be more flexible and more personalized. As a student, I was somewhat bored in school and really would have liked to move ahead much faster, and ultimately I found myself in a setting that actually allowed that because I ended up in a high school that fortunately was right across the street from a leading university, literally across the street. And so I managed to work out an arrangement where instead of taking high school classes, I would be able to go and take college classes instead, and so I managed to do two years of my college education in parallel with my high school education.
Steven Cherry: So if you hadn’t been so fortunate, you would have really desperately needed a program just like this one.
Daphne Koller: Yes, I really would have, and it would have been so amazing to have this. It would have been amazing to have this even in the circumstances that I was in because up until I got into high school, I was—to be honest—very bored in school, and it would have been amazing to have an opportunity to have something like Khan Academy that allowed me to move ahead.
Steven Cherry: Very good. Well, you know the Stanford artificial intelligence lab where you work was the longtime home of computer scientist and pioneer John McCarthy, who just died last year. He was a really iconoclastic thinker and an enormous innovator all his life—I think he would be really proud of what you’re doing.
Daphne Koller: I would hope so. I have the greatest respect for John McCarthy as a visionary of the field, and I would hope that if he were here, he would be proud of our achievements, although I would say that at the moment there isn’t really very much AI in the system as it currently stands. But I think there’s a lot of promise there.
Steven Cherry: Very good. Well, thanks for joining us today.
Daphne Koller: Oh, thank you very much.
Steven Cherry: We’ve been speaking with Stanford computer science professor Daphne Koller about some of the biggest changes to come to the academy since the Socratic dialogue gave way to the auditorium lecture. For IEEE Spectrum’s “Techwise Conversations,” I’m Steven Cherry.
Announcer: “Techwise Conversations” is sponsored by National Instruments.
This interview was recorded 26 January 2012.
Segment producer: Barbara Finkelstein; audio engineer: Francesco Ferorelli
Follow us on Twitter @techwisepodcast
NOTE: Transcripts are created for the convenience of our readers and listeners and may not perfectly match their associated interviews and narratives. The authoritative record of IEEE Spectrum’s audio programming is the audio version.