Y10W41PA - Should AI Companies Disclose Their Training Data Sources?

This week you wrote a persuasive submission about AI training data disclosure. Now you'll read another student's submission and judge how strong it is. Working through how assessors evaluate formal persuasive writing sharpens your ability to apply the same lens to your own work.

Part 1

The Assessor Scorecard for

Persuasive – Submission

A strong persuasive submission takes a clear position, supports it with specific reasoning and evidence, addresses the strongest counterargument, and closes with a practically specific recommendation. Assessors judge whether the argument truly convinces its professional audience.

Ideas & Content

Specific reasoning — not just asserting a position, but naming the mechanism behind the problem. Evidence that genuinely supports the claim. The precise way the strongest objection fails to undermine the case.

  • Specific reasoning: shows mechanism, evidence and objection handling instead of assertion alone.

Structure & Cohesion

Deliberate movement from position statement to positive case to counterargument to recommendation. Clear transitions between sections. A recommendation that is specific, not vague.

  • Submission pathway: moves from position to case, counterargument and recommendation with purpose.

Audience & Purpose

Framing calibrated for a specific professional audience. Framing that matches what that audience is equipped to evaluate. A recommendation that is actionable for them.

  • Framing in terms: the professional audience is equipped to evaluate is the primary mark of audience strength.

Language Choices

Precise analytical language throughout. Key claims expressed exactly, and the recommendation stated in specific, actionable terms. No vague or informal language that weakens formal credibility.

  • Actionable precision: states claims and recommendations in exact, formal terms.

Conventions

Accurate spelling, grammar and punctuation, as expected in formal submissions. Errors reduce professional credibility. Sentence variety and controlled complexity show command of formal written expression.

  • Formal control: uses accurate mechanics and controlled sentence complexity to sustain credibility.

Part 2

Today’s Marking Targets

Task in one sentence

Write a submission to the technology regulation inquiry arguing for or against requiring AI companies to publicly disclose the data sources used to train their models.

Let’s Focus

Three strands matter most this week: Ideas & Content, Structure & Cohesion and Language Choices. The quality of ideas decides whether the argument is genuinely analytical, with specific mechanisms and evidential grounding. The structure decides whether the argument moves clearly with explicit transitions. The precision of language decides how clearly the case is expressed.

Ideas & Content

Strong writing this week shows Ideas & Content applied consistently — not just in isolated moments. Assessors look for reasoning that serves this task: specific mechanisms and evidence behind disclosure, not a general appeal to transparency.

What markers scan for

  • Ideas & Content applied consistently throughout — not only in isolated moments.
  • The specific task and topic visibly shaping how the strand is demonstrated.

Score Bands

  • Basic

    Ideas & Content is present but applied inconsistently or only at a surface level.

  • Strong

    Ideas & Content is applied consistently, with genuine understanding of what this task requires.

  • Excellent

    Ideas & Content is applied with sustained precision throughout, shaped by the specific demands of this task.

Structure & Cohesion

Strong writing this week shows Structure & Cohesion applied consistently — not just in isolated moments. Assessors look for shaping that serves this task: an argument moving clearly through its three claims with explicit transitions.

What markers scan for

  • Structure & Cohesion applied consistently throughout — not only in isolated moments.
  • The specific task and topic visibly shaping how the strand is demonstrated.

Score Bands

  • Basic

    Structure & Cohesion is present but applied inconsistently or only at a surface level.

  • Strong

    Structure & Cohesion is applied consistently, with genuine understanding of what this task requires.

  • Excellent

    Structure & Cohesion is applied with sustained precision throughout, shaped by the specific demands of this task.

Language Choices

Strong writing this week shows Language Choices applied consistently — not just in isolated moments. Assessors look for precision that serves this task: exact phrasing at key argumentative moments, so each claim reads clearly.

What markers scan for

  • Language Choices applied consistently throughout — not only in isolated moments.
  • The specific task and topic visibly shaping how the strand is demonstrated.

Score Bands

  • Basic

    Language Choices is present but applied inconsistently or only at a surface level.

  • Strong

    Language Choices is applied consistently, with genuine understanding of what this task requires.

  • Excellent

    Language Choices is applied with sustained precision throughout, shaped by the specific demands of this task.

Now read · Student sample

Should AI Companies Disclose Their Training Data Sources?

Year 10 sample · \~300 words

Student sample for assessment

Written by a Year 10 student in Cairns, Queensland, Australia.

This submission argues in favour of requiring AI companies to publicly disclose the data sources used to train their models, on the grounds that transparency about training data is a precondition for meaningful accountability, informed consent and accurate public understanding of AI capabilities and limitations. The case for mandatory disclosure rests on three related arguments. First, AI systems trained on data that is undisclosed cannot be adequately audited for bias, errors or harmful content: if the training data is unknown, evaluating the reliability and fairness of the system’s outputs is impossible. Second, much of the data used to train AI systems has been taken from creators — writers, artists, musicians and coders — without their knowledge or consent. Disclosure is a necessary precondition for any framework of consent or compensation for the use of creative work. Third, public understanding of AI capabilities and limitations is significantly distorted when the provenance of training data is unknown: people cannot accurately assess what an AI system can and cannot do if they do not know what it has learned from. Disclosure does not solve these problems but it is a necessary precondition for addressing them. The most significant objection to mandatory disclosure is that training datasets contain commercially sensitive information — including proprietary data curation strategies and partnerships — that would be exposed to competitors through public disclosure. This objection has merit, and it does not require disclosure of the training methodology or proprietary filtering processes. What it requires is disclosure of the categories and sources of training data at a level of specificity sufficient to allow independent audit and public understanding. This is not a novel regulatory requirement: pharmaceutical companies are required to disclose the ingredients of their products without disclosing the manufacturing processes that combine them. The inquiry is invited to recommend mandatory training data disclosure at the category-and-source level, with a defined exemption process for genuinely commercially sensitive specifics, and with independent audit rights to verify the completeness of disclosure.