Among the many findings in this week's PISA 2025 release, one line from the report itself stands out for its specificity: the most troubling decline, the report states, is in the skills that matter most in an AI-saturated world, evaluating information, making connections across multiple sources, and thinking critically about what's being read. This isn't vague hand-wringing about "critical thinking" as an abstract goal. It's a precise description of the core skill set that research-based and evidence-based writing instruction has always tried to build, now flagged by a major international assessment as the area of sharpest concern.

This finding lands at a genuinely meaningful moment. As AI tools make it easier than ever to generate plausible-sounding text without necessarily engaging critically with underlying sources, the specific human skill of evaluating whether a source is credible, whether a claim is well-supported, and how multiple sources relate to and complicate each other becomes, if anything, more valuable, not less. The PISA report's framing, explicitly connecting this skill decline to an "AI age" context, suggests this isn't a coincidental finding but a genuinely pointed observation about what's at stake.

For writing teachers, particularly those teaching research writing, literature reviews, or any assignment requiring genuine source synthesis, this week's finding offers direct, timely, credible support for treating source evaluation as a first-order skill worth explicit, dedicated instruction, not an assumed byproduct of simply assigning research papers.

What source evaluation instruction actually requires

Genuine source evaluation skill, the kind this week's report flags as declining, requires more than teaching students to check whether a website looks professional or has a recognizable domain. It requires teaching students to ask specific, substantive questions: what's this source's underlying evidence or methodology, how does it compare to or conflict with other sources on the same topic, and what are its limitations or potential biases. This is genuinely demanding intellectual work, and it benefits from explicit, scaffolded instruction rather than an assumption that students will absorb it simply by being assigned enough research papers over time.

• Teach source evaluation as explicit, scaffolded instruction, not an assumed byproduct of assigning research writing
• Require students to compare and contrast multiple sources directly, not just cite several sources independently
• Ask students to articulate a source's specific limitations or potential bias, not just summarize what it claims
• Weight genuine source synthesis heavily in research writing rubrics, distinct from simple citation count or format compliance
• Use this week's PISA finding as current, credible context when explaining to students why this specific skill matters

Why this matters more, not less, in an AI-saturated environment

As AI tools make generating plausible-sounding written content increasingly effortless, the specific human skill of evaluating whether that content, or any content, genuinely holds up under scrutiny becomes a more, not less, valuable thing to teach explicitly. A student who can critically evaluate sources and synthesize across them brings genuine value to any research or writing task that a tool generating fluent but unverified text cannot replace. This week's PISA finding reinforces that this specific skill deserves protected, deliberate instructional time, not less time as other pressures compete for curriculum space.

This is a useful framing for departments making curriculum and time allocation decisions this year: source evaluation and critical synthesis aren't old-fashioned research skills made less relevant by AI. Based on this week's data, they're exactly the skills current evidence suggests need more, not less, dedicated attention.

A clear, current case for protecting this instructional time

Departments looking to justify protected time for genuine source evaluation and research synthesis instruction now have a specific, current, credible data point to point to: a major international assessment released this week explicitly naming this skill set as the area of sharpest, most concerning decline, and explicitly connecting its importance to the demands of an AI-saturated information environment.