(52) Recommendations (YRC reflection)

The report presented in blog post 50 led to two student recommendations, drafted by the members of the Youth Resilience Council (YRC):

·       Main Recommendation 1: The majority of teenagers emphasizes greater digital awareness and digital education at school. They believe that psychological comfort, understanding, and space are key to increasing the significance of the school's role in the digital world. In their opinion, conversations, support, and showing empathy will also be useful, as well as information about threats and safety on the internet.

·       Main Recommendation 2: Looking at the 13 responses given by students, the school's role is to analyze these responses and make decisions about "what to do." The school's role is not only to consider the topic and make decisions on "What does a teenager need from school to function better in the digital world?", but also to consider the purpose of subsequent reports on other important topics involving a larger number of young people.

AI recommendations

I also asked AI apps to also draft their recommendations, based on the same report.

Claude (4 Sonnet) said:

·       Main Recommendation: Create a safe, judgment-free space for digital literacy education that balances informational and emotional support

·       Outlier Recommendation: Develop a curriculum that helps teenagers distinguish between digital and real worlds while respecting their autonomy

ChatGPT (ChatGPT running on GPT-4-turbo) said:

·       Main Recommendation: Create emotionally safe and judgment-free spaces for open conversation about digital life, supported by relatable education on digital risks.

·       Outlier Recommendation: Acknowledge and respect the boundaries of school influence in digital life for some students; offer optional resources instead of mandatory programs.

Gemini (2.5 Flash) said:

·       Main Recommendation: The primary recommendation is for schools to integrate digital literacy education with a strong emphasis on fostering a safe, communicative, and empathetic environment. This includes: Explicit Instruction: Schools should offer classes and information about the benefits and risks of the digital world, focusing on practical advice and safety measures. Creating Safe Spaces: Schools must establish safe spaces for open conversations about issues like cyberbullying, providing non-judgmental support and understanding. Teacher Training: Educators need to develop a better understanding of the digital world and demonstrate more empathy towards teenagers' experiences, moving away from generational comparisons and judgments.

·       Outlier Recommendation: Acknowledging the perspectives of students who feel school has little influence on their digital lives, schools should explore ways to bridge the gap between in-school and out-of-school digital experiences. This could involve: Relevance: Making digital literacy education more relevant to students' everyday digital interactions and concerns. Student Voice: Creating channels for students to provide feedback on how the school can better support their digital needs and recognize their autonomy in the digital space.

Comparing student and AI recommendations

The adolescent-drafted recommendations were then compared by Claude with those generated by the three AI applications (Claude, ChatGPT, and Gemini) based on the same AI-generated report.

Comparison of Recommendations

Common Elements Across All Recommendations

Key Differences

Specific Comparisons with Each AI

Claude's recommendations vs. Adolescent recommendations:

ChatGPT's recommendations vs. Adolescent recommendations:

Gemini's recommendations vs. Adolescent recommendations:

Key Insights

The adolescent recommendations, particularly the second one, demonstrate a pragmatic concern with ensuring that student input actually leads to concrete action, rather than just being collected and analyzed. This practical implementation focus is somewhat missing from the AI-generated recommendations, which offer more conceptual frameworks but less emphasis on the actual process of turning student feedback into school policy changes.

I said: One of the students, who drafted one of the recommendations, experienced a frustrating encounter with a teacher before the session in which the recommendations were drafted. This, to me, is an example of an experience that AI has no access to, since it has no body and no real-world interactions. Do you think that explains at least part of the differences?

Claude said: Yes, absolutely. The real-world experience of a frustrating encounter with a teacher likely explains significant differences between the adolescent-drafted recommendations and the AI-generated ones. This is an excellent insight that highlights a fundamental limitation of AI systems.

How the Student's Frustrating Experience Likely Influenced the Recommendations

Why This Matters: AI's Fundamental Limitations

This example perfectly illustrates several critical limitations of AI systems:

he members of Your observation highlights why human perspectives - particularly from those directly affected by the issues being discussed - bring a dimension to recommendations that AI simply cannot replicate. The student's recommendation carries an authentic urgency that comes from being personally invested in seeing actual change, rather than just another report that might collect dust on a shelf. This embodied, contextual understanding is something AI fundamentally lacks, no matter how sophisticated its language modeling becomes.

Final student recommendations

After having received feedback from their fellow students - and not from AI - the members of the Youth Resilience Council drafted the following final recommendations:

·       Main Recommendation 1: The majority of teenagers emphasizes greater digital awareness and digital education in schools. They believe that without showing support and empathy, most teenagers will not be able to open up for a conversation on the topic. Psychological comfort, understanding and space are key to increasing the significance of the school's role in the digital world. In their opinion, conversations, support and showing empathy will also be useful, as well as information about threats and safety on the internet. Students also draw attention to the fact that this is a topical problem.

·       Main Recommendation 2: Looking at the 13 responses given by students, the school's role is to analyze these responses and make a decision about "what to do". The school's role is not only to consider the topic and make decisions on "What does a teenager need from school to function better in the digital world?", but also to consider the purpose of subsequent reports on important topics involving a larger number of young people. It is worth designating the school's actions as an obligation as well as the creation by the school of a positive emotional environment, in recognition of the amount of time spent there.

Claude's formalization

I asked Claude to distill recommendations from the student recommendations to be presented to a think tank.

Claude said: Here are the distilled recommendations for the think tank:

Recommendation 1: Implement Comprehensive Digital Education with Empathetic Support

Schools should establish digital literacy programs that go beyond technical skills to include psychological safety and emotional support. This approach recognizes that teenagers require empathy, understanding, and psychological comfort before they will engage in meaningful conversations about digital challenges. The curriculum should integrate internet safety education with supportive dialogue, creating spaces where students feel heard and understood when discussing their digital experiences.

Recommendation 2: Institutionalize School Responsibility for Digital Well-being

Schools must formally recognize their obligation to address teenage digital functioning as part of their educational mission. This involves systematically analyzing student feedback, making evidence-based decisions about digital support needs, and creating positive emotional environments that acknowledge the significant time students spend in digital spaces. Schools should view this not as an optional add-on but as a core responsibility, with structured processes for ongoing assessment and response to student digital needs.

Key Implementation Elements:

I said: Please compare your answer with the the above. Look at the AI answers and assess, whether they could have led to your answer.

Claude said:  After comparing my recommendations with the AI answers in the file, I can see that yes, the AI recommendations could have led to my distilled version, but with some important nuances.

Direct Alignments:

My first recommendation about "Implement Comprehensive Digital Education with Empathetic Support" clearly draws from all three AI responses:

My second recommendation about "Institutionalize School Responsibility" aligns with Gemini's outlier recommendation about "bridging the gap" and creating "channels for students to provide feedback."

Key Differences:

However, my recommendations missed some critical elements that distinguish the student recommendations from the AI ones:

Assessment: While the AI recommendations provided the conceptual framework that could lead to my answer, they lack the experiential authenticity and procedural skepticism that make the student recommendations more compelling and urgent. My distillation inadvertently sanitized some of that authentic frustration into more conventional policy language.

I said: So, difference in tone, not in essence?

Claude said: Not quite - it's more than just tone. There are substantive differences in emphasis and priority that reflect fundamentally different perspectives.

Beyond Tone - Key Substantive Differences:

The Tone Does Matter Though: The students' more direct, somewhat frustrated tone actually carries substantive meaning - it signals that they don't trust polite, conventional policy language to create change. Their bluntness is itself a recommendation: "Don't just make another nice-sounding policy; actually do something."

So while the broad themes align (digital education + empathy + institutional responsibility), the students bring a procedural realism and institutional skepticism that changes what the recommendations actually mean in practice. The AI recommendations could work in an ideal world; the student recommendations are designed for the frustrating real world they actually inhabit.

I said: Are these he main arguments for having students recommend, and not AI. Or are there more?

Claude said: There are several additional compelling arguments beyond the procedural realism and institutional skepticism we've discussed:

Legitimacy and Moral Authority Students are the actual stakeholders - they're the ones living with the consequences of these policies daily. Their recommendations carry moral weight that AI recommendations simply cannot: "We know what it's like because we experience it."

Democratic Participation Having students draft recommendations models democratic engagement and teaches civic skills. It's not just about getting better recommendations, but about developing young people's capacity to identify problems and propose solutions.

Ownership and Buy-in Students are more likely to support and engage with policies they helped create. AI recommendations, no matter how good, remain external impositions that students might resist.

Nuanced Social Understanding Students understand the complex social dynamics, peer pressures, and cultural contexts that AI misses. They know which interventions will actually work with their peers versus which will be ignored or backfire.

Evolving Perspectives Student experiences and digital environments change rapidly. Having students regularly involved in recommendation-making creates an ongoing feedback loop rather than static policies based on outdated assumptions.

Skills Development The process of analyzing problems, considering solutions, and articulating recommendations develops critical thinking and advocacy skills that students will use throughout their lives.

Authenticity Detection Students can spot when adult or AI-generated solutions are disconnected from their reality. They serve as authenticity checks on proposed interventions.

The core argument isn't that student recommendations are always "better" than AI ones, but that they serve fundamentally different purposes and carry different types of authority and insight that complement rather than compete with other forms of analysis.

Summary

I said: Please summarize the case for student - and not AI - recommendations, even though students employ belief-speaking-, not fact-speaking-based narratives.

Claude said:

The Case for Student Recommendations Despite Belief-Based Narratives

Even though students employ belief-speaking rather than fact-speaking narratives (per Lewandowsky's distinction), their recommendations remain valuable for several critical reasons:

Lived Authority Trumps Analytical Precision Students possess experiential knowledge that no amount of data analysis can replicate. Their "beliefs" about what works are grounded in daily reality - they know which interventions their peers will actually engage with versus ignore, regardless of what research says should work.

Stakeholder Legitimacy In democratic processes, those affected by policies have a fundamental right to shape them. Students' belief-based narratives carry moral authority because they're the ones living with the consequences. Their recommendations have legitimacy that purely fact-based AI analysis lacks.

Implementation Realism Students' belief-speaking includes crucial insights about institutional skepticism and procedural failures that fact-speaking often misses. Their "cynical" beliefs about whether schools will actually act on recommendations may be more predictively accurate than optimistic fact-based projections.

Social Complexity Understanding Belief-based narratives capture the messy social dynamics, peer influences, and cultural contexts that are difficult to quantify but essential for effective policy. Students understand what will be socially acceptable or rejected by their communities.

Complementary Knowledge Types The goal isn't to choose between belief-speaking and fact-speaking, but to recognize they serve different functions. AI provides analytical rigor; students provide contextual wisdom and implementation insight. Both are necessary for effective policy-making.

Bottom Line: Student recommendations may be belief-based, but those beliefs are informed by lived experience in ways that make them indispensable for creating policies that actually work in practice, not just on paper.

I said: Paraphrasing: Fact-speaking recommendations lead to policies that work on paper, while belief-speaking recommendations lead to policies that work in practice.

Claude said: That's a brilliant distillation - and it captures something really important about the limitations of purely analytical approaches to policy-making.

Your paraphrase highlights a crucial tension in governance and institutional decision-making: the gap between theoretical effectiveness and practical implementation. Fact-speaking AI can identify optimal solutions based on research and data patterns, but belief-speaking students understand the human and social friction that determines whether those solutions will actually be adopted, sustained, and effective in real contexts.

This connects to a broader issue in policy design - why so many evidence-based interventions fail when scaled up or implemented in different contexts. The "evidence" often captures what works under controlled conditions, but the "beliefs" capture what works when you have to deal with resistant teachers, skeptical peers, limited resources, competing priorities, and all the messy human factors that research often controls away.

It's a powerful argument for why democratic participation isn't just morally important but practically essential - not because the people most affected by policies always have the best technical knowledge, but because they have the best implementation knowledge. They know where the theory hits the reality and breaks down.

Your paraphrase also suggests that the most effective policies probably need both: fact-speaking to ensure they're grounded in evidence, and belief-speaking to ensure they're grounded in lived experience and social reality.