AI Fuels Dreams. But Data Has to Deliver.

[Tim_Flem]

There’s no doubt that AI is dominating the conversation right now, across industries and especially in education. In fact, about one in every four dollars in venture capital funding in 2024 went to AI-related companies. That energy was impossible to miss at the recent ASU+GSV Summit, where nearly every conversation, demo and pitch touched on artificial intelligence in some way.

New tools and startups are racing to embrace artificial intelligence, and many hold real potential. Yet within the education sector, the real differentiator shouldn’t be flashy technology with a shiny new interface but meaningful evidence of impact. Student success isn’t built on AI alone. It’s built on what actually works, backed by research and real-world outcomes.

While there’s no doubt that AI-enabled tools are becoming powerful assets in the educational toolkit, they are by no means a silver bullet. What actually moves the needle for students are technologies rooted in research, tested in real classrooms and refined based on real outcomes. The best tools, whether AI-powered or not, are those that improve retention, boost confidence, and offer insights to both students and instructors.

The Promise of AI is Real

Don’t get me wrong. In education, AI offers a world of possibilities from personalized learning pathways to efficiencies in assessment. But to fully realize that promise, our industry must stay rooted in outcomes. It’s not enough to marvel at what AI might do, we need to know what it actually does. And just how well it does it.

At ASU+GSV, I interacted with a wide range of companies bringing AI into education in thoughtful and innovative ways. Some were developing AI-assisted grading tools designed to give faculty time back. Others focused on scaling access to internships through project-based learning or using statistical modeling to enhance language acquisition. And yet others were reimagining how to deepen student practice in key subjects like math through highly adaptive platforms. 

What stood out were efforts to apply AI to real problems, like supporting student understanding and skill development, improving feedback, or increasing engagement, often in ways that could meaningfully support instructors and students alike. The common thread among the most compelling? A clear commitment to understanding and improving learning through data. And beyond that, measuring that data against things that matter, such as better engagement and improved course outcomes.

Ground AI Dreams in Real Data

At Macmillan Learning, we’ve taken a purposeful approach to AI, ensuring each tool we build rests on a foundation of data-supported outcomes. We ground our AI dreams in real data and proven efficacy.

It’s why we created our AI Tutor using a Socratic method of teaching, and are now in our fourth semester learning about its efficacy. While early findings taught us that students using it felt more confident in their problem-solving skills, this Fall we learned the benefits can extend beyond that. In fact, just 15 interactions with the AI Tutor helped students boost their course grades by three points. 

We continuously gather feedback to better understand how our tools are being used, and how they can improve. It’s that ongoing cycle of research, reflection, and refinement that keeps our focus where it belongs: helping instructors teach and students learn.

Our conversations with thought leaders and research-driven companies at ASU+GSV reaffirmed a belief we hold deeply, which is that the future belongs not just to the boldest innovators, but to those who can demonstrate genuine impact.

If the last year was about exploring AI’s potential, this next chapter should be about proving it. The question isn’t whether AI will shape education, because it already is. The question now is whether or not we can ensure it measurably improves learning.