- 16th April 2018
- Posted by: Manolis
Knewton drew heaps of hype and investment by promising to provide artificial-intelligence technology to major textbook companies to make their content more adaptive. Now the company has pivoted, and it is poised to formally announce its own online courseware that will compete head-to-head with those publishing giants.
The move marks a major turn for Knewton, which has raised more than $157 million in venture capital based largely on a promise to provide the high-tech engine inside online textbooks created by publishers and organizations. But in the past few years the company has suffered several setbacks—along with mounting criticism that its founding CEO, Jose Ferreira, overhyped its technology. One tech consultant went so far as to compare Ferreira to a snake-oil salesman.
So the company has hit the reset button. Earlier this year Ferreira stepped aside as CEO, replaced by Brian Kibby, a veteran of major textbook companies including Pearson and McGraw-Hill. And even before the management change, the company had quietly started building a huge library of courses bolted to its adaptive engine. The secret to its swift entry into publishing was OER (open education resources). Rather than hire authors to write textbooks from scratch, the company is now curating open-educational materials already on the internet.
In an interview with EdSurge this month, Kibby called Ferreira “brilliant,” but said that the company’s initial strategy was based on a “flawed” idea. “No [textbook] company would ever allow someone to come in and be the ‘Intel inside,’” said Kibby, noting that major publishers would rather build their own adaptive platforms than rely on an outside vendor. Early on, Pearson had embraced Knewton’s technology and even invested in the company, but this year that publisher decided to pull back on the use of Knewton’s platform.
Kibby also criticized Ferreira’s highly abstract and “complex” public statements about Knewton’s technology (referencing one NPR interview in which the former CEO called Knewton “a robot tutor in the sky”). Kibby said that as CEO, he plans to approach the market “with humility” and work to convince professors that Knewton’s adaptive textbooks will better help them and their students.
Kibby has big claims of his own, however, when it comes to how aggressively Knewton plans to compete with major textbook publishers. “We’re going after the higher-education market in every major course area,” he said, calling Knewton courseware “something that will basically take over the marketplace.”
The company has already built online textbooks for the full range of mathematics curriculum, as well as some economics and chemistry titles. And officials say that professors at 80 colleges are already trying the textbooks in their courses. The company plans an official launch in January, in hopes that more professors will adopt them for the fall.
Michael Feldstein, the edtech consultant who called the company’s previous pitch snake oil, said that “after years of absurd hype from their former CEO, it’s a relief to hear them promote a value proposition that at least makes sense.”
On the other hand, the company has moved into a market that’s saturated—and struggling. “While that’s certainly a reasonable business model,” Feldstein added, “I’m not sure how it justifies the enormous investments that have been made in that company.”
New ‘Course Product’
Textbook providers don’t like to use the word textbook anymore, preferring “courseware” or something that emphasizes interactive features like automated homework systems (which save professors time in scoring assignments and entering grades in the gradebook). On its website, Knewton describes its new online textbooks as “course products.”
One notable feature of Knewton’s approach is the price: Each online textbook costs $44 for two years of access, or $9.95 per month. Just like traditional textbook publishers, the company is working to convince individual professors to adopt the courseware for their students.
Whether that is a good deal depends on how you look at it. Many textbook publishers offer courseware that costs hundreds of dollars to each student. Yet, professors can build their own OER textbook for free, and companies such as Lumen Learning sell courseware products based on OER (with homework-delivery features and other tools) for less (about $25 per student). And at least one major publisher has a similar competing product: Cengage’s OpenNow courseware, based on open education resources, also starts at $25 per student.
Knewton’s pitch is that its mix of OER and an adaptive-learning system will set it apart. The basic idea is that students are shown different problems depending on how well they do on each question they face. Get a tough problem wrong, and the system shows an easier one, and then tries progressively harder questions.
Anna L. Cox, a math professor at Kellogg Community College, is among those testing Knewton’s new calculus title. (Her students have been given free access to entice her to try it). She says her best students love the new system because they are given only a few problems for homework if they do well. “They can be done in a half an hour—and for a calculus class that’s a pretty quick homework assignment,” she says. Meanwhile, students who miss early questions effectively have more homework because they are shown more questions before the system decides whether they have mastered the concepts. Cox is noticing that students are spending time searching the internet for extra help so they get early questions right to minimize their workload.
There are downsides to the adaptive approach, however. During one recent class, for instance, a student asked the professor to work out the “circle problem” that was in the homework. The professor had no idea which problem that was, and she had never seen it—because students get different questions based on how they do on each one. “I thought I was thoroughly prepared, but I wasn’t because I didn’t know all the problems it might give to them,” she says. She’s asked the company to add an option that prints out all possible questions within a topic to be ready in such cases, and she says they’ve added it to a list of things they’re considering changing.
And while Knewton officials are marketing their courseware as a textbook replacement (with content and automated homework), Cox sees Knewton’s product as mainly an “online homework system.” “The content is there if a student needs it,” she admits, “but almost no student volunteers to go out and read a math book.”
The company is essentially betting that the old idea of publishers touting the quality of their content—and seeking out Nobel-prize winning professors as authors—is fading away with so much information online for free. “The idea of the superstar author in higher education—and I have to be careful here—has had its day to a degree,” says Kibby, Knewton’s new CEO. “Strong content is—I wouldn’t say it’s a commodity now, but it’s not as hard to come by. It’s just out there.”
In an op-ed published on EdSurge in 2014, Knewton’s former CEO, Ferreira, went further. “OER will commoditize education content. Nothing can stop that,” he wrote. He remains on the board of Knewton.
These days it’s trendy for textbook publishers to say they are no longer publishers, but rather software companies. And Kibby says Knewton has an advantage in that environment because its roots are as a tech company (and Kibby says it employs 80 software engineers).
Knewton is also sticking to its old business model of selling its adaptive-learning platform to other organizations. One area it is especially eyeing is the corporate-training market, for companies interested in improving how they offer courses to their employees. And Knewton sees K-12 as a future market for online textbooks as well.
“Oftentimes you have to start in higher education because the decision-making dynamics are different,” Kibby said. “But we will be a K-12 company, a higher-education company, and a corporate-learning company.”
Many in higher education remain skeptical about the promise of adaptive learning, though—or at the least, fuzzy on what the concept looks like in reality—in part because of Knewton’s earlier missteps.
That reputation will be hard to overcome, argues Feldstein. “These incredibly overhyped products have created a backlash against anything that sounds cutting-edge because there’s just a deep skepticism and often a poor ability to evaluate what’s real and what’s hype,” he says. “Fake news came to edtech long before it came to politics.”