AI Scores on TrialAUC/ILACE 2026 workshop

Workshop home / Rubric

Direct participant access

Analytic writing rubric

Score only your assigned dimension. Use exact textual evidence and record one uncertainty.

Open essay

Content & argument

8 points

High · 7–8

Clear position; relevant reasons are developed.

Developing · 4–6

Mostly clear and relevant; support is uneven.

Limited / 0 · 1–3 / 0

Limited or off-task support. 0: No assessable response.

Organisation & coherence

6 points

High · 5–6

Clear progression; ideas connect coherently.

Developing · 3–4

Recognisable structure; connections are uneven.

Limited / 0 · 1–2 / 0

Weak sequencing or fragmentation. 0: No evidence.

Language & expression

6 points

High · 5–6

Clear control and sufficient range.

Developing · 3–4

Meaning generally clear; frequent errors.

Limited / 0 · 1–2 / 0

Meaning often obscured. 0: No assessable language.

Independent evidence record

Record your judgement

This record stays only in your browser. Upload no response. Lower down you can put your name to your rating.

1. Assigned dimension
2. Score

The available range changes to match your selected dimension.

3. Confidence

1 = very uncertain; 5 = very confident.

Your scoring record will appear here.
Continue to trial activity →

Optional controlled comparison

Compare your human rating with ChatGPT and Claude

Complete your independent human score first. Then give both AI systems the same anonymized essay and the same rubric prompt.

Human judgement comes first

The examiner’s score remains hidden. Do not open either AI tool until you have recorded your own dimension, score, evidence and uncertainty above.

  1. Copy one standard promptIt contains the anonymized transcription, the three rubric dimensions and a fixed response format.
  2. Ask both AI systemsPaste the identical prompt into a new ChatGPT chat and a new Claude chat.
  3. Record the same dimensionEnter each AI’s score for your assigned dimension and compare it with your human score.

Standardized scoring prompt

No name, candidate number, examiner mark or scanned document is included.

2 · Open ChatGPT 2 · Open Claude
Preview the exact prompt sent to both AI systems
3. Before the reveal: compare your assigned dimension

Select your human dimension and score above first.

4. After the facilitator’s reveal: compare overall totals

Leave the human examiner field blank until the facilitator reveals the benchmark. All totals use the same 0–20 scale.

Record your human score above, then enter the two AI ratings.
Discuss the disagreement →

Third-party notice: ChatGPT and Claude open on their own websites and may require an account. The workshop site does not transmit the essay automatically. Paste only the provided anonymized prompt and follow your institution’s approved-AI policy.

Send your rating to the facilitator

Put your name on your scores

Your name, your three scores and your comments go to the facilitator’s sheet, where the room’s ratings are compared. You can send again at any time — your latest entry replaces the earlier one.

Your confidence in this rating · 1 very uncertain, 5 very confident

Please note: When you send a rating you are asked for a name. That name, your scores and your comments appear on the facilitator’s screen, may be shown to the room, and are kept for analysis after the session. Use a label such as “Examiner 4” if you would rather not use your own name.