What Past Education Technology Failures Can Teach Us About the Future of AI in Schools
By Justin Reich, Massachusetts Institute of Technology (MIT)
Published: October 3, 2025
For over a century, American technologists have urged educators to quickly embrace their latest inventions, promising revolutionary improvements in learning. Yet history reveals that rapid adoption of new educational technologies rarely yields sustained benefits for students. As artificial intelligence (AI) increasingly permeates classrooms, looking back on past experiences offers vital lessons to help navigate this unprecedented educational transformation wisely.
The Cautionary Tale of Past Innovations
In 1922, Thomas Edison famously claimed that film strips would replace all school textbooks because film was “100% efficient” compared to text’s purported “2% efficiency.” This striking but misleading statistic illustrates a recurring phenomenon: brilliant technologists often overestimate the educational impact of their inventions while underestimating the complexities of teaching and learning.
Similarly, more recent introductions like mobile phones and internet-connected classrooms promised great progress. However, studies reveal that schools pioneering these technologies did not outperform more cautious counterparts in terms of student achievement, economic growth, or overall well-being. The technology’s potential can only be realized within supportive learning communities fostering appropriate practices, norms, and resources.
Reflecting on Digital Literacy Lessons
Justin Reich, a professor specializing in education technology at MIT, recalls his experiences teaching students web literacy beginning in 2003. At the time, educators used frameworks such as the CRAAP test (evaluating Currency, Reliability, Authority, Accuracy, and Purpose) to teach students how to judge online information credibility. Students were discouraged from trusting sites like Wikipedia and encouraged to favor domains ending in .edu or .org.
However, a 2019 peer-reviewed study found that these methods were ineffective. In reality, expert researchers employed "lateral reading"—leaving a webpage quickly to cross-check information on other sites—which led to superior ability to distinguish truth from falsehood online. This revelation indicated that educators had spent nearly two decades teaching millions ineffective evaluation strategies.
This hard lesson serves as a timely reminder about the risks of overconfidence in unproven educational approaches—even those backed by seemingly reasonable evidence.
The Urgency and Uncertainty of AI in Education
Today, AI is being heralded as another transformative technology for schools. But unlike past tools, AI is an “arrival technology” — entering classrooms not gradually but explosively, disrupting established routines and requiring rapid responses.
Over the past two years, Reich’s research team has spoken with nearly 100 U.S. educators, many expressing a sense of urgency to integrate AI responsibly, yet a widespread need for support and guidance. While numerous groups and consultants offer frameworks and training in AI literacy, these initiatives largely lack solid research evidence, reminiscent of early web literacy efforts.
A Prudent Path Forward: Humility, Experimentation, and Assessment
Until the education science community produces robust findings—likely taking another decade—teachers must become scientists themselves, testing and refining AI use carefully in their classrooms. Reich recommends three key strategies:
-
Embrace Humility
Educators and students should accept that current understandings of AI’s educational impact are provisional. As with past technologies, today’s best practices may later be found flawed, requiring continual revision. -
Conduct Thoughtful Experiments
Schools should identify where AI might be explored creatively and where greater caution is needed. For instance, Eric Timmons, a California filmmaking teacher, openly integrates AI to assist his students in scripting and technical problem-solving while preserving the authentic filmmaking experience they enjoy. In contrast, foundational courses like freshman English writing may warrant more conservative approaches. -
Local Assessment and Iteration
When experimenting with AI tools, educators can monitor student outcomes by comparing work produced before and after AI’s introduction. Gathering and reviewing such evidence helps determine whether AI supports learning goals, enabling data-informed adjustments.
Looking Ahead: Navigating AI’s Place in Education
By 2035, a richer understanding of AI’s impact on schools will emerge. AI could become, much like the internet, a crucial resource that enriches education despite some risks. Alternatively, it might resemble mobile phones, where negative effects on well-being and learning prompt tighter restrictions.
The pressure to resolve AI’s role in education feels intense. Yet, as history teaches, the challenge is not to rush adoption but to prioritize doing it right—grounded in evidence, thoughtful experimentation, and ongoing reflection. Only then can AI fulfill its promise as a powerful educational tool rather than a fleeting trend.
About the Author:
Justin Reich is a professor of Digital Media at the Massachusetts Institute of Technology (MIT), specializing in the study of education technology’s history and future. His research investigates how digital tools affect teaching and learning and emphasizes evidence-based approaches to integrating innovations like AI in schools.
For more information, access the full article at The Conversation.
