One selling point of MOOCs (massive online open courses) has been that students can access courses from the world’s most famous universities. The assumption—especially in the marketing messages from major providers like Coursera and edX—is that the winners of traditional higher education will also end up the winners in the world of online courses.
But that isn’t always happening.
In fact, three of the 10 most popular courses on Coursera aren’t produced by a college or university at all, but by a company. That company—called Deeplearning.ai—is a unique provider of higher education. It is essentially built on the reputation of its founder, Andrew Ng, who teaches all five of the courses it offers so far.
Ng is seen as one of the leading figures in artificial intelligence, having founded and directed the Google Brain project and served as the chief scientist at the Chinese search giant Baidu, as well as having directed the artificial intelligence laboratory at Stanford University. He also happens to be the co-founder of Coursera itself, and it was his Stanford course on machine learning that helped launch the MOOC craze in the first place.
In fact, Ng’s original Stanford MOOC remains the most popular course offered by Coursera. Since the course began in 2012, it has drawn more than 1.7 million enrollments. (It now runs on demand, so people can sign up anytime.) And his new series of courses through Deeplearning.ai, which kicked off last year, have already exceeded 250,000 signups. Even allowing for the famously low completion rates of MOOCs, it still means that hundreds of thousands of people have sat through lecture videos by Ng.
In other words, Andrew Ng probably teaches more people than anyone else on the planet, putting him in a position to have an unprecedented impact on an emerging field.
For now, offering courses on Coursera is the only thing that Deeplearning.ai does, and it is likely a lucrative business. Each of the five courses lasts between two and four weeks, and students who successfully complete all five get a certificate. (The five-week sequence is what Coursera calls a “Specialization.”) Students who want proof of completion and access to all of the resources of the course—including feedback from teaching assistants—must buy a Coursera membership that costs $49 per month. Typically, Coursera shares half of the revenue from each course with the provider, and a Coursera spokesman, Arunav Sinha, says that Deeplearning.ai has “a pretty standard rev share.”
Kian Katanforoosh is the co-creator for the deep learning courses and is listed as the Head Teaching Assistant for all of them.
“We think that to shape the future, there is a huge need for AI engineers,” he says. “We’re really amazed by the number of people who take the courses.”
Asked whether students are hesitant to put time and money into the company’s courses since it has no accreditation and no affiliation with a college, Katanforoosh said that hasn’t been the case.
“We’re not a university, but the content we teach is very applied and important in the industry,” he says. “I think we have the chance to have a pretty big name in AI and in software-engineering circles.”
Plus, he adds, the Coursera name now carries cache for some employers, even without a university involved. As Katanforoosh put it: “Coursera is a worldwide famous company, and probably more world-famous than many universities.”
In other words, when courses are turned into digital products on platforms like Coursera, universities may find themselves cut out of the picture over time. And in a landscape where reputation trumps accreditation, it’s likely that more education providers will emerge based around a single thinker—call it Star Scholar U.
A similar dynamic played out with another former Stanford professor, Sebastian Thrun. The same year that Coursera was founded, he started a company called Udacity, which today offers most of its courses without any help from colleges. That company is now valued at more than a billion dollars.
‘Hot’ Part of AI
Does the popularity of Deeplearning.ai courses worry the big-name universities who partner with Coursera? After all, colleges still offer the bulk of the company’s courses.
“I think it’s an opportunity for universities to communicate what’s different in taking a course from a university—particularly a research university,” says James DeVaney, associate vice provost for academic innovation at the University of Michigan (which offers two of the top-10 courses on Coursera).
“Michigan and other institutions like it should make sure we’re offering something of a compelling and differentiated value,” he adds. “One of the ways we can do that is in our breadth,” noting that none of the commercial course providers on MOOC platforms can cover the range of subjects a college or university does. And so he argues that colleges should look for opportunities for collaborating across disciplines in their online programs.
Meanwhile, it’s important to note the limits of the MOOC format. Some argue that taking a MOOC is more like reading a textbook than it is like engaging in a semester-long college course.
But it’s more accurate to call these open online courses a new category of learning experience. After all, the Deeplearning.ai courses taught by Ng require students to turn in coding assignments that are auto-graded, and students who pay can ask questions of live instructors. And unlike textbooks, MOOCs give students the chance to buy a certificate of completion to show to employers.
Which leads to the question that has led to widespread interest in MOOCs from the beginning: Can MOOCs (and now MOOC-based certificates) replace a traditional college education?
Carnegie Mellon University recently announced what it claims is the first undergraduate degree in artificial intelligence, making it something of a competitor to Andrew Ng’s Deeplearning.ai certificate program.
Reid Simmons, a research professor of robotics and computer science at Carnegie Mellon, directs its new AI program. In an email interview this week, he laid out how he thinks his program differs from Ng’s:
“We provide a rigorous, comprehensive, immersive learning experience, that is hard to duplicate in an online setting. The curriculum is designed to give students a full exposure to the field of AI, including the mathematical and computer science skills that they will need to utilize and develop AI techniques in the future. Note that the certificate course you indicated in Deep Learning is only a very small (albeit ‘hot’) part of AI. While it is no real substitute for an intensive four-year bachelor’s program, such programs may provide students who already have a suitable background to quickly learn individual aspects of the field. This may be desirable for those who cannot take the time to invest in a comprehensive bachelor’s program.”
His view is that there’s plenty of room for both kinds of options these days.
“The two approaches fill different needs,” he noted. “What’s important to us is to see the world educated in technologies, and CMU cannot possibly handle the load by itself; so we are always happy to hear about additional approaches that supplement our program.”
