T
The CTTL
Guest
- Thread Author
- #1
Integrity, equity, acceptable use, ethics, environmental cost. All real questions about educational technology and AI, and all downstream of one we rarely ask before answering them: how does learning actually happen?
That is the primary problem, and it usually goes unnamed.
A new position paper from the Center for Transformative Teaching and Learning argues that AI must be evaluated through a single lens:
The same question can be applied to any educational technology. Most guidance for schools comes from one of two places: technologists who have not studied learning, or researchers who do not teach. Almost none comes from someone doing both.
Dr. Ian Kelleher has taught high school science for more than thirty years and still teaches every day. He is also a Science of Learning researcher whose work is turning cognitive science into classroom practice. That translation is the hard part. Research rarely arrives in a form a teacher can use on Monday, and classroom practice rarely gets tested against what the evidence actually supports.
Every teacher already knows a sixth grader’s use of AI cannot be the same as a graduate student’s. What no one has supplied is the reason—a mechanistic account of how learning works and how that changes with age and expertise. Without it, guidance stays generic, and generic guidance fits nobody.
The first is a self-reinforcing cognitive bias loop: automation bias, the illusion of competence, the Dunning-Kruger effect, and performance illusory bias interact to erode the domain knowledge that would otherwise break the cycle. The more the cycle runs, the less equipped anyone inside it—students, teachers, leaders, parents—is to notice it running.
The second is heuristics deprivation—the failure to build the efficient, tacit patterns of expert reasoning that come only from sustained effortful practice. It quietly robs students of the very thing that makes experts expert. It does not run alongside the bias loop. It makes it permanent.
And a paradox: high achievers may be the most vulnerable. AI substitution is nearly undetectable inside a success narrative of strong grades and polished work.
This is not a case against technology. AI changes where, when, and how cognitive effort happens—and it still has real uses in schools. But sometimes it hollows out the exact work a lesson was designed to produce, often without anyone realizing. Telling the difference is a design question, and it belongs to teachers—teachers who are prepared with the knowledge of how learning actually happens.
The paper closes with a practical decision framework and addresses developmental, equity, and assessment implications. Its conclusions match the Brookings Institution’s 2026 global study. Different methods, same destination.
Read it. No email, no form. If your school is working through how to evaluate AI through the Science of Learning, the CTTL can help. Contact us to start a conversation.
The post We’re Answering AI and EdTech Questions in the Wrong Order appeared first on The Center for Transformative Teaching and Learning.
That is the primary problem, and it usually goes unnamed.
A new position paper from the Center for Transformative Teaching and Learning argues that AI must be evaluated through a single lens:
The same question can be applied to any educational technology. Most guidance for schools comes from one of two places: technologists who have not studied learning, or researchers who do not teach. Almost none comes from someone doing both.
Dr. Ian Kelleher has taught high school science for more than thirty years and still teaches every day. He is also a Science of Learning researcher whose work is turning cognitive science into classroom practice. That translation is the hard part. Research rarely arrives in a form a teacher can use on Monday, and classroom practice rarely gets tested against what the evidence actually supports.
Every teacher already knows a sixth grader’s use of AI cannot be the same as a graduate student’s. What no one has supplied is the reason—a mechanistic account of how learning works and how that changes with age and expertise. Without it, guidance stays generic, and generic guidance fits nobody.
The paper supplies the mechanism. Two of them.
The first is a self-reinforcing cognitive bias loop: automation bias, the illusion of competence, the Dunning-Kruger effect, and performance illusory bias interact to erode the domain knowledge that would otherwise break the cycle. The more the cycle runs, the less equipped anyone inside it—students, teachers, leaders, parents—is to notice it running.
The second is heuristics deprivation—the failure to build the efficient, tacit patterns of expert reasoning that come only from sustained effortful practice. It quietly robs students of the very thing that makes experts expert. It does not run alongside the bias loop. It makes it permanent.
And a paradox: high achievers may be the most vulnerable. AI substitution is nearly undetectable inside a success narrative of strong grades and polished work.
This is not a case against technology. AI changes where, when, and how cognitive effort happens—and it still has real uses in schools. But sometimes it hollows out the exact work a lesson was designed to produce, often without anyone realizing. Telling the difference is a design question, and it belongs to teachers—teachers who are prepared with the knowledge of how learning actually happens.
The paper closes with a practical decision framework and addresses developmental, equity, and assessment implications. Its conclusions match the Brookings Institution’s 2026 global study. Different methods, same destination.
Schools will change because of AI. The biology of learning will not.
Read it. No email, no form. If your school is working through how to evaluate AI through the Science of Learning, the CTTL can help. Contact us to start a conversation.
The post We’re Answering AI and EdTech Questions in the Wrong Order appeared first on The Center for Transformative Teaching and Learning.