The Hidden Risks of AI Teaching for Elementary and Middle School Students

(A Research-Based Article for Educators, Administrators, and Parents)

Introduction: Faster Learning – But for Whom, and at What Cost?

AI-based teaching systems offer great promise. A study published in 2025 in Nature - Scientific Reports showed impressive results: students who learned with an AI tutor covered more material in less time and achieved an average score of 4.2 out of 5 on tests, compared to only 3.5 in the traditional active learning group.

The Hidden Crisis: Brain Drain and Professional Burnout

Beyond the pedagogical aspect, there is a severe structural threat to the education system. In the US, recent reports point to a quiet wave of quality, values-driven, and experienced teachers leaving the profession. These teachers, who chose the profession out of a sense of mission and desire to educate generations, find themselves being pushed into the role of "AI tool coordinators" or "learning technicians."

Turning teachers into technology system operators drains the profession of its human essence. The result is paradoxical: the highest quality and most dedicated workforce – those who see education as human connection rather than data transfer – are the first to leave the system. This exodus deepens the existing staffing crisis and leaves students with fewer meaningful adults in their lives, precisely in an era when they need them more than ever.

Screen Time at the Expense of Social-Emotional Development

Adding hours of AI-based learning simply means adding screen time. Research has proven that replacing social, emotional, and educational processes with general screen time is misguided and harmful to students. A comprehensive meta-analysis of 117 longitudinal studies (involving approximately 292,000 children) found a clear connection: school-age students who spend many hours in front of screens develop more symptoms of anxiety, depression, aggression, and hyperactivity.

There is a "vicious cycle" – children experiencing social difficulties or anxiety tend to retreat to screens as a coping mechanism, which deepens their isolation. An AI tutor perceived as a patient "friend" who doesn't judge may inadvertently encourage this withdrawal at the expense of engaging with peers in the classroom.

A study examining preschool children (ages 4-5) illustrated this well: children participated in activities led by a human teacher versus a social robot. The results showed that children demonstrated significantly higher emotional and behavioral engagement with the human teacher — they laughed, pointed, and talked more. When the machine replaces the teacher, children lose critical opportunities to practice language and read social situations.

The Right Model: Human Instruction, Artificial Practice

Due to young students' intense need for a social framework, education experts draw a clear boundary regarding AI's role in elementary and middle school.

A survey conducted among grades 6-12 students in the US reinforced this need: half of the students felt less connected to their teachers when AI was integrated extensively in the classroom, and half of the teachers reported a decrease in interactions among students themselves.

Which Skills Erode When AI "Teaches"?

Transferring the teaching and knowledge acquisition stages to AI-based tools doesn't just harm knowledge – it systematically erodes deep skills critical to building a self-learner. When a child receives instruction through a screen instead of from a teacher and classroom, the following skills are directly impacted:

• Frustration Tolerance and Coping with Difficulty (Productive Struggle): A human teacher knows how to sit with the student in difficulty and encourage them to think independently before providing the solution. Artificial intelligence tends to provide immediate answers, which habituates the student to instant gratification and prevents healthy engagement with challenges.
• Reading Social Cues and Body Language: Listening to a teacher's explanation in class requires reading body language, intonation, and facial expressions. All of these atrophy when the "teacher" is a text box or a frozen figure on the screen.
• Joint Attention and Active Listening: The ability to listen to another person patiently, wait for one's turn to speak, and respond in the right context. These basic conversation skills are not practiced when facing a bot that responds in milliseconds.
• The Classroom as a Social Laboratory and Peer Learning: The classroom learning space is not just a place for absorbing information, but a laboratory where students learn through others' ways of thinking – their classmates and the teacher. Understanding why another student made a mistake or thought correctly, and exposure to different ways to reach the same solution, are critical for developing cognitive flexibility. Isolated learning with an AI system prevents this cross-pollination and the opportunity to enrich the thinking process through the peer group.
• Independent Creative Thinking (Preventing Fixation): Studies have shown that when students rely on AI for knowledge acquisition and ideas, they develop "cognitive fixation" and tend to adopt the machine's suggestions verbatim, instead of formulating their own original ideas. In a UK study, 62% of young people felt that AI made them less creative.

Recommendations and Safety Guidelines for the Education System

To enjoy the benefits of AI's personalized practice without harming student development, here are the key recommendations from the literature:

1. Cumulative Screen Time Limits (The 90-Minute Rule): Screen exposure time during school hours should be moderated. The recommendation is to limit digital learning to a maximum of 90 minutes per day cumulatively (for all digital tools combined). Beyond that, learning must be frontal, social, and hands-on.
2. The Teacher Instructs, AI Practices: AI should not be used as a tool for teaching new material. AI should come into play only after the teacher has completed instruction, as an aid for repetitive practice and personal training, not as a substitute for human explanation.
3. Preventing Emotional Dependency (APA Warning): The American Psychological Association warns against chatbots that pose as friends. Students should be made clear that AI is a work tool and not a comforting "friend," and they should be encouraged to turn to peers and teachers for emotional support.
4. Integrating Group Learning with Technology: Instead of each child sitting alone in front of a screen, it's recommended to incorporate pair or group work around the AI system (for example, students formulating a prompt together or discussing the received answer). This preserves the social context of learning.
5. Cultivating Critical AI Literacy: Students should be trained not to accept AI output as established fact. Teachers should ask students to explain in their own words the answer the AI gave, or to find potential errors in the machine's response.

Bottom Line

The consensus around integrating AI in elementary and middle schools is becoming clear: it's an amazing tool for accelerating practice and receiving immediate feedback, but it cannot and should not replace frontal teaching. When schools limit screen time to 90 minutes, leave knowledge acquisition to teachers, and use AI only for training purposes — they ensure that technology serves students rather than harming their social and cognitive development.

Key Sources

1. Muppalla, D., et al. (2023). Effects of Excessive Screen Time on Child Development: An Updated Review and Strategies for Management. Cureus / PMC.
2. Jose, B., et al. (2025). The cognitive paradox of AI in education: between enhancement and erosion. Heliyon (ScienceDirect).
3. Neumann, M. M., et al. (2023). Preschool children's engagement with a social robot compared to a human instructor. Early Childhood Research Quarterly.
4. Vilcarino, J., & Langreo, L. (2025). Rising use of AI in schools comes with big downsides for students. Education Week.
5. UNICEF Innocenti (2025). Childhood in a Digital World: Global trends in children's digital experiences. UNICEF Global Insight.
6. Burns, M. (2026). What the research shows about generative AI in tutoring. Brookings Institution.
7. Kestin, G., et al. (2025). AI tutoring outperforms in-class active learning: an RCT introducing a novel research-based design in an authentic educational setting. Scientific Reports (Nature).