Students Push for AI Literacy as Self-Taught Usage Outpaces Classroom Instruction
A Project Tomorrow survey of 95,000 K-12 stakeholders reveals students are learning AI on their own, prompting calls for better literacy and explainable AI in education.
A new national report released by education nonprofit Project Tomorrow reveals a growing gap between how students encounter artificial intelligence and how schools teach it. According to the 2025-26 Speak Up National Report, which surveyed over 95,000 K-12 students, parents, teachers, and administrators across 1,023 U.S. schools, students are increasingly exploring synthetic intelligence on their own. During an accompanying webinar, middle and high school students urged educators to provide structured instruction around AI mechanics and evaluation rather than relying on strict bans or passive technology adoption.
Key survey findings on student AI usage
The survey data demonstrates that self-directed discovery remains the primary entry point for AI adoption among adolescents. About 46 percent of high school respondents reported learning about artificial intelligence through personal experimentation, whereas only 30 percent learned about it during classroom instruction. Regular usage is rising rapidly: high schoolers using AI for studying and research on a daily or weekly basis expanded from 23 percent in the 2023-24 academic year to 40 percent in 2025-26. Similarly, routine use for drafting and brainstorming grew from 21 percent to 29 percent across the same timeframe.
Students participating in the webinar described utilizing these tools to assist learning without bypassing academic effort. Kennedy, an eighth-grade student in California, explained that she uses AI to rephrase difficult concepts and check her math calculations against synthetic outputs. Jordan, an Alabama high school junior, recalled using Google Earth combined with ChatGPT in ninth grade to redesign his school's courtyard to scale. Students also highlighted the importance of real-world context; 48 percent of surveyed high schoolers expressed a desire for lessons tied to practical situations, and an equal percentage sought training for future career skills.
Classroom implementation and teacher hesitancy
Despite student enthusiasm, pedagogical integration faces notable hurdles. Project Tomorrow CEO Julie Evans pointed out that technology should enable practical application rather than simply replicating traditional teaching methods on screens. However, educator confidence remains low: merely 11 percent of surveyed teachers reported feeling very comfortable letting students use AI tools, compared to 35 percent who felt very comfortable facilitating digital student collaboration. At the same time, 56 percent of teachers noted that digital tools aid in personalizing instruction, and 70 percent cited interactive learning as a clear benefit.
Students themselves acknowledged the potential risks of misinformation and classroom distraction. Tharun, a senior in Pennsylvania, emphasized that instruction must demystify how models construct output. "What happens behind the scenes? Where is AI getting this answer from?" he asked, adding that understanding these systems helps students grasp why outputs can be flawed. Virginia seventh grader Eloise noted that her research teacher provides AI-generated passages to help students locate inaccuracies and bias. Jordan summarized the dynamic by stating, "If students are taught how to use AI the right way, it is less likely that they'll use it the wrong way."
What it means for developers
For software engineers and product teams building educational software or AI-powered features, these findings highlight a shifting target audience. As younger generations become comfortable interacting with model interfaces outside formal instruction, developers must account for users who need explainable, transparent AI outputs rather than black-box answers. Building user interfaces that reveal underlying sources, highlight confidence levels, or explain step-by-step reasoning will align directly with the critical evaluation skills students are beginning to demand.
Furthermore, with teacher confidence in managing AI at just 11 percent, developers have a major opportunity to design administrative controls, guardrails, and guided learning modes that make AI tools approachable for non-technical educators. As AI integration accelerates across education and enterprise, developers looking to test or deploy multi-model pipelines can try top AI models cheaply through one API at https://apixoai.online. Ultimately, creating tools that support explainability, error-detection, and practical application will best prepare both students and educators for an evolving digital landscape.
Source: Students Call for AI Literacy, Purposeful Tech Use in Schools — GovTech AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
