---
source_url: https://arxiv.org/abs/2605.05598
ingested: 2026-05-09
sha256: 3983348294aecefd3b3394c3d1d6836d19b57bda144ee9006e259cc50085f15f
---

# Prober.ai: Gated Inquiry-Based Feedback via LLM-Constrained Personas for Argumentative Writing

**Authors:** Ran Bi, Shiyao Wei, Yuanyiyi Zhou
**Published:** 2026-05-07
**Venue:** NY EdTech Hackathon (2nd place)
**URL:** https://arxiv.org/abs/2605.05598

## Abstract
Inverts the conventional AI-tutoring paradigm: instead of generating or rewriting student text, Prober.ai constrains an LLM to produce only targeted, inquiry-based questions about argumentative weaknesses. A two-phase Challenge-Unlock architecture gates revision suggestions behind mandatory student reflection, creating pedagogical friction. Built in 36 hours, grounded in Toulmin's argumentation theory.