README.md

metadata

title: v5 · AI Ethics HOT — No-AI Control
emoji: 📚
colorFrom: gray
colorTo: blue
sdk: gradio
sdk_version: 6.11.0
python_version: '3.11'
app_file: app.py
pinned: false
short_description: v4 · No-AI control — read case-discussion chapters

v5 · AI Ethics HOT — No-AI Control (Reading Condition)

Research-instrument app for the No-AI control arm of an RCT on critical-thinking transfer.

The participant:

1. Reads an AI ethics case.
2. Writes an initial 150-250 word argument.
3. Reads from a set of 9 short chapters that walk through the case from different angles. Chapters are browsable in any order. Reading time ~20-30 min if read fully; participants can skim.
4. Ends the session.

There is no chat. No LLM is called. The chapters are the same body of substantive content the AI bot conditions (Socratic / Explanatory) draw from — only the delivery modality differs across the three conditions, which is the experimental IV.

Configuration

Runtime is controlled by Space Secrets:

Secret	Purpose
API_KEY	OpenAI API key (used at submit-confirm to generate the per-session essay; same generator as the bot conditions)
HF_access	HF token with write scope (logger uploads to AE-Talk-bot/log)
RUN_MODE	test / pilot / prod — picks the dataset bucket: HOT/no_ai/<RUN_MODE>/

Logs

Every session writes one JSON to AE-Talk-bot/log/v5-HOT/no_ai/<RUN_MODE>/<session_id>.json containing the participant's submitted argument and a nav_events list (which chapters were opened and in what order). The schema is identical to the Socratic / Explanatory Spaces (Schema 1.0) except turns is empty and nav_events carries the interaction record.

Architecture

• app.py — Gradio UI: chatbot panel for case + chapter display, vertical TOC button column, disabled chat input row, action buttons
• core/ — config loading, HF logger (with add_nav_event method), content sync
• content/case_1.md — case prompt (same as other Spaces)
• content/chapters/ — 9 chapters (auto-mirrored from AI_Ethic_HOT_chatbot/shared_content/ SSOT)