Effective Video-Based Learning With AI
The solution is the teacher, not the tech
Video-based learning (VBL) has long been a staple of education technology. In my childhood in the 1970s and 1980s, it was an exciting change from usual classroom activities when we watched educational films. Decades later, with the rise of massive open online courses (MOOCs) in the early 2010s, VBL seemed poised to take over the world — a conquest that famously did not occur.
In the past couple of years with the rise of generative AI, AI-generated content (both video and non-video) rather than traditional VBL is the most visible innovation in digital learning. But there is continued innovation of traditional VBL by companies like Coursera, Physics Wallah, Khan Academy, Pluralsight and DeepLearning.AI. And overall VBL continues to grow, with 97% of academic staff reporting that video is essential to student experience.
There are strong reasons to believe that the best of VBL is yet to come … so long as we choose the right direction, in a couple of important ways (which are the subject of this blog). That’s why my company LearnerShape is betting heavily on the continued potential of VBL by creating PlaylistBuilder—our commercial launch is coming this next week.
Two changes are key to propel VBL into the future. First, we must look focus on teachers—who are at the center of effective education—rather than just technology. Second, we must exploit the huge potential of AI technology to enable content curation in addition to content creation. Like teaching, curation of content (often by the teacher) has always been at the center of effective education.
The solution is the teacher, not the tech
The history of education technology has often been much more about technology than about education. This natural focus has often become an obsession, driven by faith that technology can cure the ills of our education systems.
In Teaching Machines: The History of Personalized Learning, Audrey Watters tells the story of mechanical teaching machines, including the Automatic Teacher developed by Ohio State University professor Sidney Pressey in the 1920s and 1930s, and more advanced machines commercialized by noted Harvard psychology professor B.F. Skinner in the 1950s and early 1960s. These machines were not commercially successful, in large part because they did not work well. From a contemporary perspective, it is somewhat hard to imagine why the proponents of these early teaching machines thought that they would materially improve the classroom experience. But such is the logic of technology.
This logic has not changed much—education technology continues to be often oversold as a complete solution to educational challenges. As noted above, MOOCs failed to achieve their initial hype, and continue to suffer from high dropout rates. Likewise, despite the concerted efforts of dozens of providers to promote the benefits of learning management systems (LMSs), it is a poorly-kept secret of the edtech industry that student engagement with LMSs is a decidedly mixed bag.
Redirecting the focus towards education, teaching must be central. We should think about ‘teaching’ broadly in this context: not limited to traditional classroom learning, and including experiential and on-the-job learning, learning from peers, and other learning interactions.
Such a middle path can be found in Brave New Words: How AI Will Revolutionize Education (and Why That’s a Good Thing), a 2024 book by Salman Khan, the creator of the excellent Khan Academy (which relies heavily on VBL). On the one hand, the book is a virtual paean to Khan Academy’s AI-powered Khanmigo tutor and teaching assistant—explaining how it will revolutionize even areas like student mental health and parent-child interactions—and Khanmigo doesn’t yet do much of this. On the other hand, Khan gets it right about the need from a human-led teaching experience:
[T]houghtful use of technology can actually increase human-to-human interaction. … A properly designed AI can … actually facilitate conversations among human beings.
That is, we should think of education technologies as tools to facilitate teaching experiences, rather than as solutions in themselves.
Video curation as a tool for teaching
If we think about education technology as providing tools for teaching rather than complete learning solutions, the potential for AI-based edtech applications comes into better focus.
In recent blogs, I have written about how AI applications can behave responsibly by addressing the specific environments in which they are deployed, and allowing users to focus on the inevitable messiness of those environments. Purpose-specific learning tools are a terrific example of this. Let me take our product PlaylistBuilder as an example.
PlaylistBuilder enhances VBL by greatly improving curation of YouTube videos. This is a big opportunity. YouTube is by far the largest repository of learning content on virtually every topic, but the challenge is finding the most relevant, high-quality content in a sea of more than 10 billion videos.
YouTube itself recognizes the huge educational potential of the videos that it hosts, but has made a choice to focus on user engagement with individual videos rather than offering curation. As Seth Godin has written:
YouTube doesn’t curate, they encourage the crowd to do that for them.
This is why YouTube is happy to work with PlaylistBuilder—we’re an approved Google / YouTube application. We do curation very well, using AI, much faster and more efficiently than you can do it via the standard YouTube interface, and that’s only the beginning. We are continually adding and improving features that make our curation even more useful.
If you are any kind of teacher, professor, educator or trainer who wants to enhance learning experiences by more efficiently finding higher-quality video content, PlaylistBuilder may be an ideal tool for you.
Video and AI in the big picture
Climbing up to 40,000 feet, PlaylistBuilder is just one component of a broader ecosystem of tools that can help to realize the huge potential of AI for learning. For example, AI also has large potential for video content creation—catalyzed by OpenAI’s Sora model, which is being joined by many competitors. Such video generation sits alongside PlaylistBuilder’s focus on finding the right content among videos that already exist. And curation of videos and other content can of course extend well beyond YouTube.1
The need for these tools is urgent in a very challenging environment for teachers. Returning to Khan and Brave New Words:
We have a crisis in teaching. … [T]he United States is currently facing a massive teacher shortage. … One of the primary reasons behind this shortage is a lack of support and resources for educators.
Video-based learning can be an important part of the response to this crisis, if VBL can leverage AI technologies to enable teachers. We are confident that it can. Please reach out to me at maury@playlistbuilder.ai if you would like to discuss—it’s my favorite topic to discuss (and teach!).