Updated at 10:30 a.m. ET

If you do a Google image search for "classroom," you'll mostly see one familiar scene: rows or groups of desks, with a spot at the front of the room for the teacher.

One teacher, many students: It's basically the definition of school as we know it, going back to the earliest days of the Republic. "We couldn't afford to have an individual teacher for every student, so we developed a way of teaching large groups," as John Pane, an education researcher at the RAND Corporation, puts it.

Pane is among a wave of education watchers getting excited by the idea that technology may finally offer a solution to the historic constraints of one-to-many teaching.

It's called personalized learning: What if each student had something like a private tutor, and more power over what and how they learned?

Pane is the lead author of one of the few empirical studies to date of this idea, published late last year. It found that schools using some form of personalized learning were, on average, performing better ( there were some wrinkles we'll talk about later on).

"In a personalized system," he says, "students are receiving instruction exactly at the point where they need it."

It's a concept grounded in the psychology of motivation, learning science and growing technologies like artificial intelligence (AI). And the hype around it is blowing up. Personalized learning is the No. 1 educational technology priority around the country, according to a recent survey by the Center for Digital Education, a news service that promotes ed-tech. More than nine out of 10 districts polled said they were directing devices, software and professional development resources toward personalized learning.

Personalized learning is also a major priority of the Bill & Melinda Gates Foundation (which is a supporter of NPR's education coverage) and the Chan Zuckerberg Initiative. The commitment by the Facebook founder's philanthropy is expected to run into the hundreds of millions of dollars per year.

But there's already a backlash to the idea: it's drawn teacher, parent and student protests--even walkouts--in several states.

So what is personalized learning, exactly? The term has buzz, for sure. But it's also a bit — or more than a bit — baggy.

In fact, in speaking about it with more than a dozen educators, technologists, innovation experts and researchers, I've developed a theory: "Personalized learning" has become a Janus-faced word, with at least two meanings in tension:

1. The use of software to allow each student to proceed through a pre-determined body of knowledge, most often math, at his or her own pace.

Paragraph 20 0
No paragraph-level conversations. Start one.

Paragraph 20, Sentence 1 0
No sentence-level conversations. Start one.
2. A whole new way of doing school, not necessarily focused on technology, where students set their own goals. They work both independently and together on projects that match their interests, while adults facilitate and invest in getting to know each student one-on-one, both their strengths and their challenges.

Paragraph 21 0
No paragraph-level conversations. Start one.

Paragraph 21, Sentence 1 0
No sentence-level conversations. Start one.

Paragraph 21, Sentence 2 0
No sentence-level conversations. Start one.

Which vision of personalization will prevail? Pace alone, or "Personalize it all"? And what proportion of the hype will be realized?

At your own pace

The first version of personalization is less radical and, by that token, already more common. It's the selling point of software programs, primarily in math, that are already found in millions of classrooms around the country. Two examples are McGraw Hill's ALEKS and Khan Academy.

In a traditional 3rd grade classroom, the teacher may give a test one Friday on adding and subtracting numbers up to a thousand.

Let's say you don't quite get it, and you bomb that test. On the following Monday, the teacher will introduce multiplication. What are the chances that you're going to grasp the new concept? And what about the student sitting next to you? She already learned her multiplication tables over the summer. She's doodling in her notebook and passing notes during the lesson.

Sal Khan, the founder of Khan Academy, defines personalization by pace. He tells me: "It's about every student getting to remediate if necessary, or accelerate if they can."

Khan Academy is a giant online library, viewed by tens of millions of people worldwide, of multiple-choice practice exercises and short instructional videos, with the strongest offerings in STEM disciplines.

In theory, it's possible to follow Khan's roadmap step-by-step, node by node, from simple counting all the way through AP calculus. Students, parents or teachers can keep track of progress using a dashboard.

When it comes to the transformation of education, "I strongly believe the biggest lever is moving from fixed-pace to mastery-based education," Khan says.

What he means by "mastery-based," is that students move on to the next topic only when they are ready. It's simple in concept, yet it's not the way school usually works.

In our example of a third grader using Khan or another software system, you'd get the chance to keep doing practice problems and watching videos on addition and subtraction. You wouldn't move on until you'd answered a certain number of problems correctly. Your teacher would be put on notice that you haven't quite grasped the concept before you bombed a test, so she could give you extra help. Meanwhile, your friend could move from multiplication on to division and beyond.

With Khan Academy, you can show "mastery" by getting a certain number of questions right in a row. Khan Academy has recently introduced more assessments, so that more of the exercises in their free library can be used in this way.

So there you have it. Personalized learning: a cost-effective, efficient way to improve direct instruction through pacing, while giving young people a little more autonomy. What's not to love?

Jade Davis has thoughts about that. She's an expert in emerging technologies in education, and the director of digital project management at Columbia University Libraries. When she thinks of personalized learning, "I think of kids with machines that have algorithms attached to them that move them through learning at the pace where the student is."

Does that excite her?

"No, it doesn't," she answers. "Because learning is a collaborative process. When you take away the ability for people to make things together, I think you lose something."

And, she adds, there's another issue. Many recent critics have pointed out how biases, such as racial biases, can be baked into all kinds of algorithms, from search engines to credit ratings. Davis argues that educational software is no exception. "It's going to sort students. It's going to stereotype, put up roadblocks and make assumptions about how students should be thinking." In other words, what's sold as "personalization" can actually become dehumanizing.

Teachers, I point out, can and do show biases as well. Point taken, she says. But, "teachers can attempt to remedy their bias ... teachers are learners in the space, too, but software is not."

Equating personalized learning simply with pacing is "a fairly large problem," according to Susan Patrick, the president and CEO of the International Association for K-12 Online Learning. She says part of the issue is that personalization has become a flimsy marketing term, with

"software vendors putting a sticker on a product because there's variation in pacing." That, she says, "does not equal a truly personalized approach."

I also talked to Ted Dintersmith. He's a technology venture capitalist who has visited schools in all 50 states. He presents himself as an expert, not in education, but in innovation, and is the author of What School Could Be, which features teachers talking about the promise of education.

For Dintersmith, the at-your-own-pace model falls well short of what personalization could be.

"If it's plopping down some obsolete or irrelevant curriculum on a laptop and letting every kid go at their own pace, It's hard to get excited about that," he says. "If it's giving students more voice, helping them find their own talents in distinct ways, that's better."

When it comes to software like Khan Academy, "I think it's a fair criticism to say most of what's on Khan has kids listening to lectures and practicing and taking multiple-choice tests to get good at some low-level procedure" — such as multiplication, say — "that the device they're working on does perfectly, instantly."

That's not good enough for the demands of the 21st century, Dintersmith adds. "Being pretty good — even very good — at the same thing that everyone else is pretty good to very good at doesn't get you anywhere. You really want bold, audacious, curious, creative problem-solving kids that embrace ambiguity."

He believes letting students choose more about what, and how, they learn is the way to awaken those qualities: letting them go off-roading, not merely letting them move at their own pace through a "closed course" of facts and skills that's already been set up for them.

Learn what you want

When you leave behind the narrow path of personalization simply as a matter of pacing, you enter a world that is broader. To some people that's more exciting, but it's also more difficult to sum up.

"At the beginning of a fad there's a naming problem,"Rich Halverson says. He's an education professor at the University of Wisconsin-Madison who has spent the last few years traveling around the country to see personalized learning in action at public schools.

He's found that, "what schools call personalized varies considerably," and also that "a lot of schools are doing personalized learning, but don't call it that."

Still, he's managed to identify some key common elements:

At the schools he's studied, students meet regularly, one on one, with teachers. They set individual learning goals, follow up and discuss progress. All of this may be recorded using some simple software, like a shared Google Doc. It's kind of like a schoolwide version of special education, with an IEP — an individualized education program — for every student.

This sounds simple, but face-to-face interaction is "expensive," says Halverson. Think 28 meetings of 15 minutes each — that's a full day of a teacher's time, somewhere between once a week and once a month. In fact, the entire school day, week, year may need to be reconfigured to allow for it.

Some schools Halverson has studied, especially charter schools with more freedom, have remade the curriculum to emphasize group projects and presentations, where students can prove the necessary knowledge and skills while pursuing topics that interest them. Students are grouped by ability and interest, not age, and may change groups from subject to subject or day to day. Scheduling and staffing is necessarily fluid; even the building may need to be reconfigured for maximum flexibility.

"I love school!"

James Murray is the principal of Waukesha STEM Academy, a K-8 charter school in Wisconsin that is one of Halverson's exemplars. It has elements of at-your-own-pace, software-enabled learning: In middle school, students have the ability to take whatever math they need, from 4th grade through calculus.

There's also flexible scheduling, with Tuesday and Thursday "flex time" blocks for whatever students want to do, Murray said. On any give day, a student can say, " 'If I need to work on a science lab, I go do that. When I'm done, I go to another class.'"

Murray says a lot of parents will ask, " 'Well what if my kid just takes gym class every day?' " The answer is, with guidance and feedback, "They really start to advocate for themselves and they start to understand what they need to do and why."

By middle school, his students propose their own long-term "capstone" projects, which range from raising money for a women's shelter to sharing their love of go-kart racing.

Sounds like fun. And indeed, a common element to personalized learning schools, Halverson has found, is that "when it's done well, there's a lot of parent and teacher enthusiasm."

Amy Bigelow is one of those enthusiastic parents. Her daughter started this fall at Murray's school, Waukesha STEM Academy. She's says she's seeing her daughter "thrive" and grow in self-confidence.

"She can think outside the box, and be creative and work with her hands," Bigelow says. "She has classes with seventh-graders, eighth-graders. It allows her to be with people on the same level, not based off age or grade, and that's been a refreshing outlook, too."

Last year, when her daughter was in fifth grade, Bigelow said, "she would come home from school just in a funk at the end of the day." But now? "She came home the first week and she said, 'Mom — I'm learning, but it doesn't feel like I'm learning.' "

John Pane, the researcher at Rand, says this enthusiasm comes from two places. The first is that students care more about their learning when they have an element of choice and agency.

Amy Bigelow agrees: "There are so many opportunities ... for her to be able to be empowered and take her schooling into her own hands."

The second point, Pane says, is that students care more about learning when they feel that teachers know them personally. And that happens through those regular one-on-one meetings, and through kids having the chance to share their passions.

It's what Halverson calls, "an effort to build the instruction on a personal relationship: 'What do you need to know and how can I guide you to get there?' "

"It's hard to implement."

So there you have it. Personalized learning: a transformative, labor-intensive approach giving students ownership over their learning. What's not to love?

Well, Sal Khan, for one, is a bit dismissive of what he calls this 'flavor' of interest-driven personalization. "We're all learning about factoring polynomials," he says, "but you're doing it in a context of something that interests you, say soccer, and I'm doing it in the context of something that interests me, say architecture. Or maybe there's instruction in different modalities. That's not the type that we focus on. There's not evidence it's effective, and it's hard to implement."

The research by Pane and his colleagues bears this view out, to a point. Their study of charter networks that were early adopters of personalized learning found large average effects on student achievement.

But a second study by Pane, with a more diverse set of schools, found a smaller average positive effect, which included negative impacts on learning at "a substantial number" of schools.

"So that, to me, is a warning sign that personalized learning appears not to be working every place that people are trying it," says Pane. "While conceptually they are good ideas, when you come down to analyzing it there are potential pitfalls."

One emerging issue is that, as the "fad" spreads, teachers may not always be getting the supports they need.

For a report published in 2018 by the Center on Reinventing Public Education, researchers interviewed and surveyed hundreds of teachers at schools that had received funding from the Gates Foundation to design and implement personalized learning. They found that, while many teachers were wildly enthusiastic, they were often left on their own.

They had little guidance to set meaningful learning outcomes for students outside the state frameworks of standardized tests. And, they had little support at the school- or district-level to change key elements of school, like age-based grouping or all-at-once scheduling. So personalization efforts often didn't spread beyond pilot classrooms.

The case of Summit Learning is another example of personalized learning's growing pains. It's a personalized learning platform that originated at a California-based charter school network called Summit Public Schools. After investments from the Chan Zuckerberg Initiative and some work from Facebook engineers, the platform and curriculum, plus training, was offered up for free, and has been adopted by almost 400 schools around the country.

Summit Learning is different from single-subject systems like ALEKS. It's been advertised more like a whole-school personalized learning transformation in a box: from mentoring sessions with teachers to "playlists" of lessons in every subject. The company says that participating schools are reporting academic gains for students who start out behind, as well as "greater student engagement, increased attendance, better behavior."

But not everyone loves the program. It's drawn teacher, parent and student protests in Cheshire, CT; Boone County, KY; Fairview Park City in Ohio; Indiana Area School District in Indiana, PA; Clearwater County, ID, and recently in New York City.

Some have privacy concerns about students' personal data reportedly being shared with Microsoft, Amazon and other companies. Some object to the quality of the curriculum and supplementary materials. Some say students are getting distracted by working on the laptop or merely Googling for answers to quizzes. Some just don't want to learn on their own at their own pace.

"It's annoying to just sit there staring at one screen for so long," Mitchel Storman, a ninth grader at the Secondary School for Journalism in Brooklyn, told the New York Post at a student walkout earlier this month. "You have to teach yourself."

Summit shared with NPR a letter from Andrew Goldin, the Chief Program Officer of Summit Learning, to the principal of the Secondary School for Journalism, Livingston Hilaire. Goldin stated that the school lacked enough laptops, Internet bandwidth, and teacher training to successfully implement the program, and recommended that they suspend it immediately for 11th and 12th graders.

Backlash to the backlash

Is personalized learning, aided by computers, destined to be just another ed reform flash-in-the-pan? Will it have a narrow impact in just a few subjects? Or will it be transformative, and is that a good thing?

As the Gates Foundation experience suggests, the future of personalized learning may hinge on what kinds of supports are offered teachers. The experience of the state of Maine is instructive here too.

In 2012, Maine became the first state to adopt what's called a "proficiency-based diploma." The idea behind it was that instead of needing to pass a certain set of classes to graduate, students in Maine now had to show they were "proficient" in certain skills and subjects.

To comply with the new law, many districts adopted "proficiency-based learning." The new system shared elements of personalized learning, like students being allowed to re-do assignments and work at their own pace. Yet schools received little funding or guidance on how to implement these changes, leaving some teachers lost and overwhelmed.

Heather Finn, a veteran math teacher at a high school in central Maine, told NPRit was "impossible ... so, so frustrating."

"It works really well, like, the first month," Finn says. Then, students started to progress at different speeds.

"So I have the kids who are on pace, and I have the kids who are perpetually, always behind. And it got to the point where I had 20 kids in 20 spots."

This past April, Maine lawmakers heard complaints from parents and teachers, as well as the statewide teachers union. Three months later, Gov. Paul LePage signed a bill to make "proficiency-based diplomas" optional. Some districts have already declared that they're leaving the new system behind and will return to a more traditional education style.

Some districts, though, like Kennebec Intra-District Schools in Maine, aren't going back. Kaylee Bodge, a fourth-grader at Marcia Buker Elementary School, says the appeal is simple. "We get to make choices instead of the teacher choosing. If you like something and you want to do that first, you get to do that first."