Skip to Content

Lessons from the Digital Classroom

Technologists and venture capitalists are betting that the data online learning generates will reshape education.

In four small schools scattered across San Francisco, a data experiment is under way. That is where AltSchool is testing how technology can help teachers maximize their students’ learning.

Founded two years ago by Max ­Ventilla, a data expert and former head of personalization at Google, AltSchool runs schools filled with data-gathering technology.

Information is captured from the moment each student arrives at school and checks in on an attendance app. For part of the day, students work independently, using iPads and Chromebooks, on “playlists” of activities that teachers have selected to match their personal goals. Data about each student’s progress is captured for teachers’ later review. Classrooms are recorded, and teachers can flag important moments by pressing a button, as you might TiVo your favorite television show.

The idea is that all the data from this network of schools will be woven into a smart centralized operating system that teachers will be able to use to design effective and personalized instruction. There is even a recommendation engine built in.

While most schools don’t have the type of technology AltSchool is developing, classrooms are increasingly filled with laptops and other digital teaching aids. This year U.S. elementary, middle, and high schools are expected to spend $4.7 billion on information technology. What is new is that many of the technologies are capturing expansive amounts of data, enough of it to search for meaningful patterns and insight into how students learn. The potential for that to be turned into profit is a big reason investors have increased funding of educational technology startups worldwide, from $1.6 billion in 2013 to $2.4 billion in 2014; they invested over $1 billion more in the first quarter of 2015, much of that in China. What all that data is teaching us about how we learn and whether technology is actually making instruction better are the big questions at the heart of this Business Report.

At the AltSchools, where tuition can exceed $20,000 a year, the goal is to create highly individualized instruction built on a system that can grow to reach a broad scale. Four more AltSchools are opening this fall, including one in Brooklyn, New York, and Ventilla hopes to eventually sell access to the system to other schools, too. AltSchool has raised $133 million from the likes of Facebook founder Mark ­Zuckerberg, venture capitalist John Doerr, the Omidyar Network, and venture firms Andreessen Horowitz and Founders Fund. “What if we tried to create not just great schools we’d like to send our kids to, but an expanding ecosystem?” says ­Ventilla, who started thinking about education when he and his wife began applying to preschool for their daughter in 2012. “What role can technology play to superpower each child and each set of parents and educators?”

Similar experiments are under way in colleges as well. In the seven years since the first “massive open online course,” Connectivism and Connective Knowledge, was taught by two Canadian educators, Stephen Downes and George Siemens, MOOCs have become a source of tremendous amounts of data about students’ behavior. Examination of this data has intensified since 2012, when the three largest platforms for these classes were launched: the Harvard-MIT joint venture edX and two for-profit companies founded by former Stanford professors, Udacity and Coursera. Between the fall of 2012 and the summer of 2014, more than a million people participated in the 68 open online courses on EdX, logging 1.1 billion clicks on the edX servers.

While only a small percentage of students complete any given MOOC, their data is helping educators develop new teaching models that promise to be more effective—such as programs that combine online instruction with one-on-one coaching or support, regular quizzes, and other check-ins on progress.

This approach has been shown in some cases to be more successful than traditional classroom instruction. Arizona State University, for example, offers more than 90 different undergraduate and graduate degrees online, part of a long-term goal as a public university of expanding access to education. The university teaches freshman math to 8,000 students a year. Those needing to catch up to college level are placed in Developmental Math, a class where 50 percent of students have traditionally earned a D or F.

Four years ago ASU combined its online and classroom approaches to Developmental Math, switching to video-based lectures and incorporating an online tool made by a company called Knewton. It analyzes students as they work through online math lessons to understand how they learn best and what they have and have not mastered. Reports on students’ progress, the time they are putting in, and their engagement and success then go to student coaches who reach out by e-mail, text message, or in person. In the first two semesters the school used this approach, the pass rate increased to 75 percent.

Udacity has similarly structured itself around individualized feedback. In its first month offering a “nanodegree” in Android programming, designed with Google, Udacity reported that students had submitted over 2,000 projects, which were then evaluated by a paid network of coding experts around the world. Cofounder Sebastian Thrun says 91 percent of paying students with this kind of coached model finish the course. Though it’s not a perfect contrast, the free robotics MOOC Thrun taught as a Stanford professor had a completion rate of 2 percent.

The data from online instruction offers a new level of feedback for teachers, too. Teachers on the Coursera platform have a dashboard on which they can see exactly when in a video students are most likely to stop watching, what percentage of students are getting an assessment question right the first time, and other metrics. If only 20 of the 200 students taking a quiz got a certain question right, teachers can reëxamine how they taught that point in the video or how the question was worded to see what’s going wrong.

“I taught in a university 18 years and I never got that kind of detailed feedback,” says Daphne Koller, a Stanford engineering professor who cofounded Coursera three years ago.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.