As the debate over the role and future of Massive Open Online Courses (MOOCs) gathers steam, Andrew Ng, associate professor of computer science at Stanford University and cofounder of Coursera, shared his views on how MOOCs can optimize and personalize education experiences at a seminar co-hosted by the Stanford Chinese Faculty and Family Club and the China 2.0 initiative at Stanford Graduate School of Business. (Watch the seminar video here.)

Ng offered up plenty of food for thought about the potential implications of the “massive data” collected from MOOCs. Coursera, which started by offering two Stanford courses online, is an education company partnering with 76 universities and organizations worldwide to offer 550 free online courses. It is currently the largest MOOC platform, according to Ng, with six million students and 108 partners.

Speaking to more than 120 attendees, Ng claimed that Coursera collected more educational data in 2013 than the entire academic field of education throughout the history of mankind. “The volume of data to unleash is unprecedented,” he noted. Researchers and educators seeking to innovate in education will be able to apply even the most basic data analytic techniques to produce a multitude of research papers and generate innovative ideas to advance education.

In a crowded field with other firms and organizations such as NovoEd, OpenEdX, Udacity, and Udemy offering MOOCs, Coursera’s students tend to be professionals seeking continuing education, with more than 75% of students already having earned a Bachelor’s degree at a minimum. By analyzing rich data collected from tens of thousands of MOOC students, instructors are receiving almost instantaneous feedback to help them improve their teaching methods and to communicate with students more effectively. Ng gave an example of how Coursera sends more “user-friendly” weekly emails to increase actions from students. “If you are reminded about things you have done well, and [we] ask you to engage in the next class, you are more likely to take action,” Ng said.

With auto-graded quizzes and peer-grading built into courses, students receive feedback much quicker in MOOCs than having to wait for professors or graduate students to return their graded work. And feedback is received at both ends. MOOCs have dramatically increased the pace of feedback for those doing the teaching. Stanford faculty members receive feedback from students annually. “Once a year I get the opportunity to improve my class… This is an incredibly slow rate of learning for me,” Ng commented. However, data is collected from students continuously on MOOC platforms – when a student finishes a lecture, speeds up videos, submits a survey, or engages in social media – instructors receive a wealth of informative resources to “rapidly improve their classes,” argued Ng.

Running A/B tests is another way to improve learning interfaces. Ng shared an ongoing test carried out by his company in determining the importance of featuring an instructor’s face in the lower corner during lectures versus hearing merely a voice over a teaching presentation. Although staff at Coursera was split over what they thought would help students more, it turned out that students without the video of the instructor found it much harder to concentrate.

Ng notes that production resources and the lack of a revenue stream are currently two major bottlenecks for MOOCs, but he is proud to be feeding Coursera’s test and analysis results back into its course development and design to create even more value-optimized courses. Coursera also makes the company’s raw educational data available to partners and other universities. He attributed an optimization mindset to his Google days, where he was the founder of Google’s large-scale deep learning project. “Your ability to get the data to optimize makes something much more valuable, and we hope to do that with education,” said Ng. “We are in very early stages of figuring out what technologies should look like for MOOCs. There is a serious trade-off between cost of content production and how engaging the content is. We continue to see other transformations.”