Bellwether is 15 this year. We’re using this moment to look back on some of the work we’re most proud of — work that we plan on scaling as new trends and challenges emerge.
Blended learning. Massive Online Open Courses. Smart Boards. Zoom. Since Bellwether’s founding in 2010, dozens of technology trends have come onto the education scene, but only some have fulfilled the hopes that practitioners in the field had for them.
Is artificial intelligence (AI) another technological advancement that seems primed to transform education in theory … and will do anything but in reality? Or is it a revolutionary technology that’s much bigger and much more potentially groundbreaking than many of the ed tech developments that came before it?
We wanted to see what the experts thought. So, we reached out to leading AI thinkers in the field — from Bellwether and beyond — to ask them the following:
Looking back over the past 15 years, what’s one ed tech trend that seemed transformative at the time? How can the lessons from that experience help the sector thoughtfully navigate the rise of AI in education and beyond?
The opinions on AI’s transformative potential in education varied significantly, but most of our respondents held a similar belief: No technological advancement can fully replace the human experiences that underlie teaching and learning.
Mary K. Wells
Co-Founder and Senior Partner, Bellwether
The hype around blended learning over a decade ago reminds me a lot of the hype around AI in schools today. Blended learning advocates believed it would lead to improvements in both efficiency and in meeting students’ individual needs, and many AI advocates believe that AI can do the same.
Blended learning, however, was a big disappointment for most schools. The products didn’t live up to the hype, implementation was difficult, and the envisioned advances in student learning proved to be mostly illusory. We should learn from this to avoid thinking about AI as a quick fix or silver bullet. Even though AI is a technology-based innovation, we can only move as fast as the people implementing solutions. We need to plan for this and help school practitioners implement AI solutions thoughtfully and strategically.
Gwen Baker
Venture Advisor, LearnerStudio; Former Chief Operating Officer and Partner, Bellwether
Over the past 15 years, technology has meaningfully transformed nearly every aspect of American life — except education. In K–12 schools, tech adoption has often been slow, fragmented, and met with skepticism, despite universal use outside of the school building.
When I was at Bellwether in 2019, we urged the sector to break this pattern. We called for a systemic, cross-sector approach to innovation — one that unified important but siloed efforts in education — in order to drive real progress.
That call is even more urgent today. The education sector is severely lagging in its understanding of AI’s impact and its potential to transform outcomes for young people. We now have a chance to build a new system that understands we are already living in the Age of AI. We must stop relying on outdated tools and fragmented solutions and instead build a flexible, open, and public digital infrastructure in which students are inspired and prepared to thrive in life, careers, and democracy. Technology is no longer optional; it’s an essential tool to build fair systems that center human potential.
Alex Spurrier
Associate Partner, Bellwether
Massive Online Open Courses (MOOCs) were all the rage in the early 2010s, offering the promise of access to instruction from elite universities to anyone from a high school student to a mid-career professional. I count myself as a MOOC success story — the Johns Hopkins University data science offerings on Coursera helped me build skills with the R programming language that reshaped the arc of my career — but my story is an outlier. MOOC retention and completion rates were a huge disappointment to those hoping for a dramatic shift in educational opportunities.
There’s a real risk that AI tools end up producing outcomes similar to what MOOCs produced: providing additional opportunities to already-motivated learners. On top of that, if students are using AI as a cognitive crutch to replace the necessary productive struggle of learning, it may actually constrain their intellectual development. If AI is to be a force for expanding educational opportunity, it’ll require what many MOOCs lack: human support to keep learners motivated, engaged, and on track.
Michelle Culver
Founder, The Rithm Project
Over the past 15 years, personalized learning has reshaped classrooms. At its best, it has honored individual needs and given students agency. At its worst, it has turned learning into a solo experience between a young person and a screen.
Now, that model collides with a generation facing unprecedented loneliness and pressure. Now, more than ever, relationships aren’t just nice to have: they’re essential for academic success, mental health, economic mobility, and a thriving democracy.
At the same time, generative AI is rapidly reshaping how trust and relationships form, blurring lines between what’s real and what’s artificial. With AI companions and apps like character.ai, young people are not just learning through technology, but forming relationships with technology itself.
The question isn’t just how to accelerate learning; it’s how we do so while helping young people reclaim and redefine human connection in the age of AI. We must use this as a chance to design intentionally for a future where AI strengthens rather than erodes human connection.
Amy Chen Kulesa
Senior Associate Partner, Bellwether
Research indicates that implementing high-quality instructional materials (HQIM) can secure student achievement gains; the recent National Assessment of Educational Progress score bright spots (amid otherwise dismal results) of Louisiana and Mississippi could arguably be proof points. And it’s applaudable how many practitioners have adopted HQIM — along with aligned professional development practices — especially in light of tremendous COVID-recovery needs in K-12 schools across the country. At the same time, effective HQIM adoption is time- and resource-intensive, even for experienced educators.
AI could theoretically make it faster, cheaper, and easier to create and differentiate HQIM. It could also improve teacher-sourced content currently cobbled together from Google searches and Pinterest. Content and tools alone, however, are insufficient to drive meaningful change by themselves, no matter how advanced they are. Changing the mindsets, habits, and behaviors of educators through aligned professional development is still the most effective avenue for transformation.
Kristen DiCerbo
Chief Learning Officer, Khan Academy
In the early 2010s, I was heavily involved in game-based learning as a solution to student engagement challenges. Yet here we are in 2025, and games in schools mostly involve students answering multiple choice questions, as opposed to immersive digital experiences. This is because there is limited time in schools and a defined amount of content to be covered; our schools are not structured to support a technology that requires practice and deep exploration. As we navigate the AI world, we need to remember that how students learn has not changed and that our solutions must be able to fit into the structure of classrooms.
Marisa Mission
Senior Analyst, Bellwether
As a Generation Z kid, I grew up “alongside” technology — from floppy disks in elementary school, a chunky laptop in high school, and 3D printers in college. The projectors on wheels turned into Smart Boards, and in-person lecture halls became virtual Zoom breakout rooms. Each tech development seemed more revolutionary than the last, but through it all, learning never changed. Curiosity, discipline, supportive teachers, and the freedom to explore were still the keys that unlocked my growth.
As generative AI becomes the norm, we should remember that all the fancy toys in the world won’t change something fundamental: students are still humans. The way the human brain absorbs, processes, interprets, and applies information hasn’t changed just because computers can now mimic us. Learning hasn’t changed, but what has changed is the world around us. Students will need different and new skills to navigate an AI-enabled world. Cultivating agency, empathy, citizenship, curiosity, a healthy dose of skepticism, and sound judgment is paramount to preparing the next generation for a future beyond our wildest dreams.
