PISA Results, Growth Mindset, and the Role of Deep Learning

The 2025 PISA results bring news worthy of appreciation for Indonesian education. Reading scores increased from 359 in 2022 to 365 in 2025, while science scores rose from 383 to 389. This six-point increase in both fields signals a positive trend, suggesting that efforts to improve learning quality are beginning to yield results.

In mathematics, scores dipped slightly from 366 to 364. This area clearly requires more intensive national effort and sharper strategies. Overall, Indonesia’s achievements remain below the OECD average. However, viewing PISA solely as a list of deficiencies would mean missing important lessons from the progress that has been made (OECD, 2026).

This positive direction aligns with national data. Education reports show that the proportion of students reaching minimum literacy competence increased from 59.49% in 2022 to 70.03% in 2024. In numeracy, the increase was even greater, rising from 45.24% to 67.94% (Ministry of Primary and Secondary Education, 2025). This data is vital as it demonstrates that improvements in learning are not stagnating. The challenge now is to maintain consistency, expand impact, and ensure this progress reaches all student groups.

The story behind the scores is equally compelling. The report provides more than just scores; it reveals the narrative of how students learn, how they face difficulties, and how teachers support the learning process. Approximately 80% of Indonesian students expressed curiosity about many things, higher than the OECD average of 73%. Furthermore, 74% stated they increase their effort when tasks become difficult, compared to the OECD average of 60%. Only 9% consider school a waste of time, far below the OECD average of 24% (OECD, 2026).

This learning capital is further strengthened by self-regulation abilities. Around 90% of students seek help from teachers when needed, and 92% ask peers for assistance when they do not understand something. Teacher support is also encouraging: 82% of students stated that science teachers show attention to each student’s learning, 85% said teachers continue teaching until students understand, and 80% receive additional help when required (OECD, 2026). In other words, there is a sufficiently strong socio-pedagogical foundation to accelerate the improvement of learning outcomes.

However, this positive narrative also raises policy questions: why has this relatively strong learning capital not fully converted into high academic achievement? One answer lies in the ‘growth mindset’. PISA 2025 noted that only 30% of Indonesian students possess a growth mindset, compared to the OECD average of 69%. In Indonesia, students with a growth mindset achieved science scores approximately 17 points higher than those with a fixed mindset, after accounting for socio-economic backgrounds of students and schools (OECD, 2026).

This figure should not be read as a new stigma against Indonesian students. Instead, it indicates a very concrete area for intervention. A growth mindset is not an innate talent, but the belief that abilities can develop through effort, proper strategies, feedback, practice, and a willingness to continue learning. When students believe that abilities can grow, mistakes are no longer perceived as a final verdict, but as information to improve learning methods.

This is where Deep Learning (DL) gains strategic relevance. DL can serve as a bridge to transform curiosity into investigative skills, perseverance into problem-solving abilities, teacher support into learning independence, and reflection into an awareness of continuous growth. The principles of mindfulness, meaning, and enjoyment must be present in the experience of understanding, applying, and reflecting. Thus, DL is not an administrative addition, but a way to deepen the learning process that already possesses significant positive capital.

PISA 2025 also shows that 53% of Indonesian students use AI chatbots to assist their learning at least once a week, which is higher than the OECD average of 46% (OECD, 2026). This presents a massive opportunity but also demands clear pedagogical direction. Without Deep Learning, AI could easily become a shortcut for obtaining instant answers. With Deep Learning, AI can be used to ask questions, compare arguments, seek evidence, test ideas, and deepen reasoning.

All these findings are relevant to the mission of the Minister of Primary and Secondary Education, Prof. Abdul Mu’ti: ‘Quality Education for All’, which emphasises equal access alongside quality assurance (Ministry of Primary and Secondary Education, 2025). The word ‘quality’ demands an increase in achievement, while ‘for all’ demands that progress is not only enjoyed by elite schools or specific student groups. Therefore, future success should not be measured solely by the rise in national average achievement, but also by the ability of the education system to uplift the most disadvantaged students.

PISA 2025 gives us reason to be optimistic yet to work harder. Reading and science scores have increased, national literacy and numeracy are improving, and student curiosity and perseverance are strong, with teacher support showing a promising foundation. Mathematics still requires serious attention, while the growth mindset needs to be systematically strengthened among both students and teachers.

The policy task now is to convert all this positive capital into higher, deeper, and more equitable learning outcomes. If Deep Learning can play that role, PISA will no longer be merely a triennial report on Indonesia’s position, but a compass to bring us closer to the ideal of Quality Education for All.