The “Colorado Roadmap for AI in K-12 Education” will lead us toward worse educational outcomes than we currently are experiencing. While teachers may view the use of AI as adding efficiency to their work, the cost of that efficiency is a lack of depth in teaching.

As AI is likely to be addressed in Colorado’s upcoming special legislative session, it is imperative that these issues be brought to the forefront.

The current state policies could lead to the emergence of a dystopian education system. It is not hard to imagine a scenario in which a teacher might ask students to write about a prompt that the teacher generated with AI. Students, whether for efficiency or lack of effort, would offer an AI-generated response — which then would be graded by AI. This interaction would appear to the outsider as learning, but the entire exchange would be entirely superficial. Such a scenario not only seems possible today but also inevitable.

If we choose to rely too heavily on AI in education, we risk losing the human core of teaching and learning. According to a new study from MIT, AI decreases the physical traits of learning by reducing widespread neural activation while writing. Allowing teachers to use AI to create lesson plans and provide feedback to students will lead to a homogenized education system with little depth. Policies implemented by the state must continue to emphasize the importance of teachers in learning.

While schools have already recognized the dangers AI poses for students, the use of AI by teachers remains relatively unregulated. Colorado’s plan for AI in K-12 education encourages the use of AI in nearly every part of teachers’ work, particularly in grading and lesson plan generation. The plan does not warn teachers of the risks of using it, only giving vague guidelines, saying that AI struggles to make complex decisions, and has difficulties understanding cultural nuances.

It is key that teachers feel connected to their work. The MIT study outlines AI’s risks. Participants who used AI to produce papers had significantly less neural activation and felt less connected to their writing than those who generated papers on their own. Research shows that teachers with higher cognitive complexity can better individuate their teaching to accommodate for different learning styles.

Teachers who use AI may be less likely to improve their technique or teaching methods, as they will not be generating the necessary neural activation to do so.

As a student and Daniel’s Fellow at Common Sense Institute, I’ve been exposed to AI in both school and work settings. Recently, I had a biology professor who openly admitted using another professor’s slide deck — and was less engaging and could not answer questions about the content. Their disconnection from their own work was obvious and led to a lack of student engagement and worse student performance. The use of AI to generate content only serves to make this problem more common for students and worsen already-low test scores.

Colorado’s plan describes a vision in which teachers offload tasks to AI and are left with more time for student interaction. Research shows that the more AI is used in the writing process, the more similar the resulting papers are. Teachers using AI will tend to generate similar lesson plans, leading to an efficient, but depersonalized system.

I believe that the state needs to create frameworks in which AI is only used to support teachers’ work, not replace it, supporting educators rather than encouraging them to offload their work.

Colorado’s state performance among high schoolers is already relatively low, ranking 37th out of 50 states in their average SAT score. According to the Common Sense Institute, this trend is not new, as Colorado’s test scores continue to decline. Expecting AI to save the state from that decline distracts from the current issues facing schools, including high rates of absenteeism and lack of engagement from students.

Despite appearing to produce results, AI still is not capable of helping students learn unless applied effectively. Colorado’s AI-centric approach will lead students to have a far less effective education. If we hope to produce a creative and educated populous, AI must be a supplemental tool, only being used in places in which humans cannot do the work. If misapplied, it will lower neural activation and long-term retention.

In my work as a student, AI is often helpful in creating personalized learning tools such as practice tests. When I’ve overused it though, it detracts from learning. If we choose to relegate teachers to a passive role, we lose the human ingenuity and creativity that has the potential to make education in Colorado great.

Judah Weir is a rising junior at Colorado State University and a Daniels Scholar Junior Fellow with the Common Sense Institute.