Global Education Assessment Results Released: AI Is Not the Culprit, but It’s Creating New Divides
Hello everyone, I’m your financial journalist and economist. Today, we’re talking about a recently released global education assessment report—the OECD’s “PISA 2025 Results.”
PISA (Program for International Student Assessment) is widely recognized as the most influential global education evaluation, conducted every three years. This time, it covered 91 countries and regions, involving 760,000 15-year-old students. You can think of it as a benchmark of the capabilities of teenagers worldwide.
The most surprising conclusion of this report is somewhat counterintuitive: Both reading and math scores have declined globally, but we can’t solely blame AI for this. Although AI has become increasingly integrated into classrooms, it is neither the primary cause of the decline nor a simple tool for cheating. Instead, AI acts like a double-edged sword—it makes the strong stronger and the weak weaker, and it may even prevent some people from developing the ability to think deeply.
Let me break down this report into five key points to help you understand the realities behind these educational changes.
---
1. Don’t Blame AI Too Quickly: The Decline in Scores Is an Old Problem, with AI Being a New Factor
Many people, upon seeing headlines like “AI Causing Students’ Scores to Drop,” immediately think, “This is the end; ChatGPT has ruined the kids.”
But the data shows this is a misunderstanding.
Looking at the timeline:
- Reading scores began to decline after reaching a peak in 2012, with a significant drop in 2018, and they dropped to 461 points in 2025.
- Math scores were relatively stable from 2003 to 2018, then began to plummet after 2018.
- The tipping point: ChatGPT became popular at the end of 2022.
This means that even before AI became widely used, students’ scores had been declining for at least a decade.
So what’s the real reason?
It’s a combination of the aftermath of the pandemic, the fragmented attention caused by smartphones and social media, a shortage of teachers, deteriorating classroom discipline, and a general lack of interest in reading long articles.
A particularly striking statistic in the report is that the proportion of “rush readers”—students who skim through texts and guess answers—doubled from 6.6% in 2018 to 11.4% in 2025. These students aren’t unable to understand the content; they simply don’t want to read carefully. This decline in deep reading skills** is the result of a long-term trend.
Conclusion: AI is not the root cause of the education crisis; it’s more like a magnifier. In an already fragile and distracted educational system, AI has exacerbated certain issues, but the problems were already present ten years ago.
---
2. The “Inverted U-Curve” of AI Use: How You Use It Determines Its Impact
One of the most interesting findings in the report is the relationship between how AI is used and students’ scores. This isn’t a linear relationship but follows an “inverted U-curve”:
- Low or no use of AI: Moderate scores.
- Moderate use (several times a week or month): The best scores.
- High frequency of use (almost daily): Poorer scores.
Why is moderate use better, while frequent use worse?
It’s similar to exercising:
- Moderate use of AI = Hiring a personal trainer: You use AI to explain concepts or check your logic—your brain is still working, and you’re engaging in productive cognitive activities.
- Frequent use of AI = Letting the trainer do the work for you: You rely on AI to write summaries, essays, or provide answers—your brain becomes inactive.
PISA’s founder, Schleicher, made a great analogy: “Watching a sports game won’t make you stronger; you have to actually participate in the activity.”
If you let AI generate answers for you, you’re depriving yourself of the opportunity to develop critical thinking and deep understanding. Students who rely on AI to summarize texts perform worse in science because they skip the most crucial step of thinking.
Conclusion: AI itself isn’t good or bad; it depends on whether you use it as a tool to enhance your thinking or as a crutch that prevents you from thinking for yourself. The former enhances abilities, while the latter leads to decline.
---
3. The Invisible Divide: AI Creating New Class Segregation
This is the most concerning aspect of the report. In the AI era, the wealth gap may no longer be just about money but also about AI literacy.
Data shows that students from higher socioeconomic backgrounds use AI more frequently and have better resources to learn how to use it effectively.
- Advantaged students not only use AI to aid their learning but also take special classes to learn how to evaluate the reliability of AI-generated information. This creates a positive cycle: the better they use AI, the better their scores and the more opportunities they have in the future.
- Disadvantaged students either lack access to AI due to lack of devices or internet at home or use it blindly, copying and pasting content without understanding its limitations.
This leads to a new digital divide:**
Previously, we worried about having or not having a computer; now, we worry about whether we can use AI effectively. If schools don’t take proactive steps to provide equal AI literacy education, AI will become a tool that exacerbates inequality rather than promoting fairness.
Conclusion: In the AI era, knowing how to use AI is a privilege. If only the elite have this skill, education fairness will face unprecedented challenges.
---
4. The Crisis of “Cognitive Endurance”: Do We Still Want to Think When AI Can Provide Answers Instantly?
The report highlights that the decline in reading scores is due to a decline in higher-order skills such as evaluating information, integrating multiple sources, and critical thinking.
In the AI age, this is particularly dangerous because AI-generated content can be persuasive and authoritative but may contain biases or errors.
If students get used to asking AI for answers, they lose the ability to ask questions and evaluate evidence.
Schleicher noted, “In a world of information overload, it’s harder for students to distinguish between useful and irrelevant information.”
Imagine a 15-year-old facing a complex scientific article. Instead of reading it carefully, they could ask AI to summarize it for them in seconds. Over time, this reduces their cognitive endurance—their ability to focus and think deeply.
Conclusion: Education must protect time for deep reading and **critical thinking.* If students can’t read long articles thoroughly, no matter how advanced AI becomes, they won’t be able to become well-informed adults.
---
5. The Future of Education: AI as a Tool, Not the Core
Facing these challenges, the report doesn’t suggest banning AI but proposes a new direction for education that emphasizes human agency:
- AI as a Tool, Not a Crutch: Use AI to support learning, not replace human thought.
The report suggests three approaches:
1. **Protect Time for “Slow Thinking”: Countries like France are reemphasizing literature and reading comprehension; Singapore maintains a lot of printed books in high-tech classrooms; Sweden promotes deep reading through library programs. The more digitalized we become, the more we need to preserve non-digital experiences.
2. Make AI Literacy a Mandatory Skill: All students, regardless of background, should learn to use AI critically. Teachers should shift from being knowledge dispensers to guides, teaching students how to question and verify AI.
3. Redefine Core Competencies: Machines are great at routine tasks, but education must focus on human abilities that machines can’t replace, such as metacognition (reflecting on one’s own thinking), systemic thinking (understanding complex systems), ethical judgment (making choices based on human values), and creativity (creating new things).
Conclusion: If education doesn’t protect these human abilities, the next generation may become mere appendages or substitutes of machines. The goal of education in the AI era is not to teach students how to defeat AI but to teach them how to coexist with it while maintaining their autonomy and creativity.
---
Summary
The PISA 2025 report serves as a wake-up call, but it also points the way forward:
1. Don’t demonize AI: The decline in scores is a long-term trend, with AI being just one of the factors.
2. Use AI wisely: Moderate use can enhance learning; over-reliance can be harmful.
3. Be wary of new divides: AI literacy must be a foundation for educational equity.
4. Protect Deep Thinking: In a fragmented world, focus and attention are scarce resources.
5. Emphasize Human Value: Education should aim to cultivate people who can use AI, not just operate it.
For parents and educators, the question is no longer whether to use AI but how to use it. We need to help students develop critical thinking skills, understanding that while AI is a powerful tool, the owner of their thoughts remains the student themselves.