Why Mark Zuckerberg Wants to Spend on Personalized Learning

December 2, 2015 1:59 PM EST

When Mark Zuckerberg and Priscilla Chan announced Tuesday night that they will give away almost all of their Facebook shares for charitable purposes, they wrote that their initial focus would be “personalized learning, curing disease, connecting people and building strong communities.”

Curing disease and building communities seem straightforward, but you may be wondering what “personalized learning” is.

Sometimes called “teach to one” or “adaptive technology,” personalized learning uses interactive software to tailor lessons and assignments to individual students, in order to reflect their strengths and weaknesses, and the pace at which they learn.

That’s not the same as educational computer games.

Instead of colorful characters tromping through mazes or fighting villains by solving math problems—’90s kids will remember games like “Math Blaster”—students interact with personalized learning software in much the same way they might interact with an old-school paper workbook. Only instead of ripping out pages and turning them in, they check boxes and type answers into software that corrects and tracks their progress, allowing teachers to quickly see that, say, little Johnny understood the concept of long division, but consistently got questions about remainders wrong.

The whole idea is to create what some designers call “a conversation” between students and teachers, making it easy for teachers to administer individualized class assignments, quizzes and homework that react to what a student “gets” and what he’s still struggling with.

“We’re starting to build this technology now, and the results are already promising,” Zuckerberg and Chan wrote in an open letter to their newborn daughter, Max, about the charitable work. “Not only do students perform better on tests, but they gain the skills and confidence to learn anything they want. And this journey is just beginning. The technology and teaching will rapidly improve every year you’re in school,” they wrote.

While the general concept of personalized learning isn’t new—teachers have long tried to design lessons to reach individual kids—the explosion of new technology, apps, and the “smart” software has only begun to penetrate classrooms in the last few years. As a result, we don’t know yet how well it works or, really, whether it works at all.

A study last March by McGraw-Hill Education Research found that 85% of students saw a “moderate or major improvement” in their grades after using adaptive technologies. Another study last month by the Gates Foundation found that students, mostly at charter schools, who used personalized learning tools made more progress over the course of two years than a control group.

But skeptics are pumping the brakes. One problem, they say, is that personalized learning software is only as good as the teacher using it—and teachers often need training and tech assistance to use it properly. That often feels like an additional burden on an already over-stretched and underpaid teaching force.

Another problem is that evidence of the efficacy of personalized learning is, at this point, mixed. Even that optimistic Gates study relied on a virtual control group that may or may not have been using tech-based teaching methods; subsequent studies could easily show that personalized teaching tools are no more effective than, say, smaller class sizes.

“There is no good evidence that online or blended learning works to improve teaching or leads to real personalized learning,” Leonie Haimson, who co-chairs the Parent Coalition for Student Privacy, told TIME in an email. “Instead this is really depersonalized learning through machines.”

Haimson argues that personalized learning tools are detrimental to the development of students’ critical thinking skills, since they reduce human interaction in the classroom.

In the Gates study, the most signifiant improvements were among elementary school kids, in both math and reading. Personalized learning tools appeared to be less helpful, and in some cases detrimental, to high school students in both subjects.

The findings suggest either that personalized learning apps are less well-developed for more complex subjects or that complex subjects simply don’t lend themselves as easily to such technology. It might be easy, for example, to track young students’ understanding of arithmetic and basic reading comprehension. But using technology to introduce new lessons in trigonometry or measuring whether a kid has grasped tonal themes in epic poetry? Might need to do that the old-school way.

Last, advocates concerned with student privacy worry that Facebook—an organization that has a less-than-stellar reputation for privacy—is entering an arena where its software will collect reams of information about students and their skills. The day Zuckerberg and Chan announced their initiative, an electronic toy company, VTech Toys, announced that 6.3 million kids’ personal data had been breached. Facebook has dismissed such concerns and promised to keep all student data safe.

At any rate, this isn’t Zuckerberg’s first rodeo. In 2010, the Facebook founder gave $100 million to refurbish the public school system in Newark, New Jersey. The effort was later criticized for being largely ineffective.

In September, Facebook announced its partnership with a Bay Area-based charter school network, Summit Public Schools, to develop personalized learning platforms. The Facebook-made software is now being used by both Summit schools and others. The software will be free for everyone.

Whether personalized learning revolutionizes learning or becomes another educational fad, the money and attention paid to it by Zuckerberg and Chan ensures that it will be a major topic in the coming years.

More Must-Reads from TIME