What Decades of Tech Revolutions Teach Us About Getting AI Right in Learning

Rajen Sheth and Philip Moyer have seen a few technology revolutions up close. Rajen, CEO and co-founder of Kyron Learning, spent 17 years at Google, where he led AI products for Google Cloud. Philip, now CEO of McGraw Hill, was his partner on that leadership team, with earlier stops at AWS, Microsoft, and Vimeo.

In a recent webinar, the two sat down to talk about what it will take to get AI right for students and educators, drawing on what they've learned building AI across industries. They were joined by Raina Rutti, department chair and clinical associate professor at Wichita State University, who has been bringing the McGraw Hill Connect Learning Coach, built jointly with Kyron, into her online courses.

What emerged was a clear through-line: the lessons from past tech shifts show up, almost one-for-one, in what's working in classrooms today.

LESSONS FROM DECADES OF TECH REVOLUTIONS

Lesson 1: Every hype cycle settles, and AI will become ambient

Philip pointed out that AI has been around for decades. What changed in 2022 was that it suddenly became approachable for everyday users. Since then, the hype cycle has moved faster than any before it. But, as with past technologies, he expects it to settle into something steady and everyday.

"Just as today we don't say, hey, I'm an internet company or I use the internet in education, eventually we're not going to say I use AI in education. It's going to become ambient." — Philip Moyer, CEO, McGraw Hill

Rajen, who moderated the conversation, noted how quickly the debate has shifted. AI has moved from a future consideration to something front and center, with responses ranging from outright bans to schools built entirely around it.

“Probably the true answer is somewhere in the middle and there’s a lot of nuance in terms of trying to figure out how do you actually make AI right for students.” — Rajen Sheth, CEO and Co-Founder, Kyron Learning

Lesson 2: Nobody is behind, and getting it right is hard

Every industry Philip works with — healthcare, financial services, construction, education — feels like it's lagging. His view: the playing field is more level than it looks, because building AI that works is genuinely difficult, even for people who've spent their careers doing it.

"Everybody feels like they're behind. And the reality is nobody's behind and nobody's ahead." — Philip Moyer

Lesson 3: AI is only as good as the people and data around it

Philip described two telecommunications companies he and Rajen worked with on customer support AI. One arrived with its support issues clearly categorized and its team brought along for the change. The other handed over its data and said, in effect, "have at it." Only the first succeeded.

"It's not just enough to paint a picture of the future. You have to actually paint everybody in your organization into that picture." — Philip Moyer

Rajen shared a similar story from an insurance project: technology that could assess car damage from a single photo, and worked extremely well, but never got used because it was never woven into how the organization actually operated.

"It's not about the technology. It's about the people and the key customer problem and how well you understand that problem." — Rajen Sheth

Lesson 4: Measure the outcome, not the activity

The last lesson, Philip said, is essential in education: hold AI to a measurable goal.

"It is absolutely essential that we don't just do AI for AI's sake, but that we're actually producing an outcome and that we're measuring that outcome." — Philip Moyer

Rajen added that measuring learning with AI brings its own challenge. Traditional studies can take years, while AI changes within months. His answer is to build on what learning science has already proven, like the outsized benefits of active learning over passive learning.

“Even though AI is changing, you can actually still ground it on things that are very much rock solid.” — Rajen Sheth

WHAT THOSE LESSONS MEAN FOR LEARNING

Applied to education, these lessons point toward a specific kind of AI: one that respects how people actually learn.

Widen the aperture over time. Philip compared AI in learning to a camera lens. For younger learners still building self-regulation, the aperture stays narrow and focused. As learners mature, it can open up, while still keeping them on a clear learning path.

Respect pedagogy. Learning is a sequence of building blocks. AI should stay within the context of what a student is studying right now, inside a focused learning environment that the educator curates.

Redefine safety for education. Safety usually means keeping AI from saying harmful things. In learning, it means something more.

"Safety inside of education is oftentimes not giving the answer." — Philip Moyer [~19:02]

Keep the educator at the center. Both leaders returned to this point again and again. The most successful AI deployments in any industry keep an expert human in the loop, and in education that expert is the educator.

"Having AI be an authentic extension of the educator is such a critical thing." — Rajen Sheth [~10:12]

Trust follows the same pattern. Philip shared that in McGraw Hill's recent global survey of teachers, 81% said they would trust AI embedded in the learning tools they already use in the classroom.

FROM THE CLASSROOM — RAINA'S EXPERIENCE

Raina has taught Principles of Management at Wichita State for eight years, mostly online, and has long used McGraw Hill Connect in her courses. When she was invited to try Learning Coach, her class had already started. She reworked her assignments anyway and made the Learning Coach a graded part of the course.

Learning Coach pairs a short video lesson on a concept students often struggle with, followed by a guided conversation. Rather than simply answering questions, the AI responds with specific instruction rooted in learning science, guiding each student toward understanding while surfacing misconceptions and insights for the educator.

Engagement without the wait. Student engagement in online courses had been a long-standing challenge for Raina.

"With Learning Coach, I felt like the engagement happened naturally." — Raina Rutti

Students watched the lesson, then immediately engaged in a conversation about it, which Raina believes helped both engagement and retention.

Fewer questions, more support. Raina didn’t receive questions from students about the Learning Coach assignments. Students found it easy to use and got instant feedback within each conversation, so the support they needed was right there in the moment. They also took to it readily, which Raina attributes partly to the experience being embedded and structured within the assignment. Her evaluations were positive, and students said they were getting through the material in less time. Easy to integrate. She found it simple to add as an assignment, and even reduced some required reading for topics the Learning Coach covered.

Her advice to other faculty is straightforward: use it. For colleagues worried that students will use AI to shortcut their learning, she sees structured AI-powered learning as a strong starting point.

WHERE THE LESSONS MEET THE CLASSROOM

Put side by side, Raina's experience reads like a case study in the lessons Rajen and Philip learned the hard way:

• Build it into the course. Raina didn't add AI off to the side. She made it part of the course, assigned and graded, and students engaged with it as a natural part of learning.
• Keep a focused environment. Students trusted Learning Coach because it stayed within the boundaries of their assignment — the same idea as Philip's camera aperture.
• Respect productive struggle. When asked whether students should be allowed to make mistakes, Raina's answer was immediate: "I personally think the more mistakes they make, the sooner the better."

"Students are actually much more willing to be candid and talk to our system than they are to raise their hand in class." — Rajen Sheth

Capturing those honest moments, the hesitations and the mistakes, gives educators visibility into exactly where to help.

-------------------------------

Asked what education will look like a decade from now, Philip described a future where knowledge is more modular, learners carry a longitudinal learning record with them, and teachers curate the journey. But the core, he said, won't change.

"It's going to continue to be, I would say, wonderfully human." — Philip Moyer

That may be the biggest lesson of all. Technology revolutions succeed when they're built around people. In learning, that means AI that keeps students thinking and keeps educators at the center.