Idea #1
AI Human Brain Alignment
EEG-informed gating for responsible LLM assistance in academic work
Brief:
Gating system which only unlocks the LLM once the user has shown pre-commitment to thinking via:
- Behavioural Evidence
- Have they drafted enough original content?
- Have they created an outline/plan?
- Time-on-task + revision activity
- Neurophysiological Evidence
- EEG-derived proxies for engagement / cognitive effort during drafting
- Personalised thresholds from a short calibration phase
Aim:
Encourage authentic ideation before the use of LLM by unlocking AI support only after measurable engagement signals
Deliverables:
A desktop/browser middleware that controls LLM access + a small study comparing gated vs ungated use on originality and retention
Novelty:
Real-time neuroadaptive "cognitive warm start" for AI assistance
Research Questions:
Main:
- Does EEG + Behavioural gating increase original contribution and learning/retention compared to ungated LLM access?
Mechanism:
- Can we reliably classify "Idea Generation / Engaged Drafting" vs "Passive Copying / Low Effort" from consumer EEG + keystroke signals in real-time?
UX:
- Does gating harm user experience, or does it feel like a helpful "coach"?
Scope / Assumptions:
- EEG is treated as a noisy but useful signal
- System supports behaviour-only gating as a fallback
- Gating targets academic pipeline workflows
References:
- Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task
- https://arxiv.org/abs/2506.08872
- Not peer-reviewed
- Published on 10th June 2025 by MIT