# Fix the Mind, Not the Move: Interpretable AI Assistance via Knowledge-Gap Localization

arXiv: https://arxiv.org/abs/2606.05602 · HTML: https://mlsanity.com/paper/2606.05602/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 18 of its 44 references that are in the map.
- First posted on arXiv 4 Jun 2026.

## Builds on (most cited first, 18 of 18)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Gaussian Error Linear Units (GELUs)](https://mlsanity.com/paper/1606.08415/) (2016)
- [On the Utility of Learning about Humans for Human-AI Coordination](https://mlsanity.com/paper/1910.05789/) (2019)
- [CodeT5+: Open Code Large Language Models for Code Understanding and Generation](https://mlsanity.com/paper/2305.07922/) (2023)
- [Deep Knowledge Tracing](https://mlsanity.com/paper/1506.05908/) (2015)
- [Shared Autonomy via Deep Reinforcement Learning](https://mlsanity.com/paper/1802.01744/) (2018)
- [Explicability? Legibility? Predictability? Transparency? Privacy? Security? The Emerging Landscape of Interpretable Agent Behavior](https://mlsanity.com/paper/1811.09722/) (2018)
- [Learning to Cooperate with Humans using Generative Agents](https://mlsanity.com/paper/2411.13934/) (2024)
- [InteRACT: Transformer Models for Human Intent Prediction Conditioned on Robot Actions](https://mlsanity.com/paper/2311.12943/) (2023)
- [Assistive Teaching of Motor Control Tasks to Humans](https://mlsanity.com/paper/2211.14003/) (2022)
- [Computational Teaching for Driving via Multi-Task Imitation Learning](https://mlsanity.com/paper/2410.01608/) (2024)
- [Hierarchical Expertise-Level Modeling for User Specific Robot-Behavior Explanations](https://mlsanity.com/paper/1802.06895/) (2018)
- [Robots that Take Advantage of Human Trust](https://mlsanity.com/paper/1909.05777/) (2019)
- [The Invisible Mentor: Inferring User Actions from Screen Recordings to Recommend Better Workflows](https://mlsanity.com/paper/2509.26557/) (2025)
- [Timing the Message: Language-Based Notifications for Time-Critical Assistive Settings](https://mlsanity.com/paper/2509.07438/) (2025)
- [Integrating Field of View in Human-Aware Collaborative Planning](https://mlsanity.com/paper/2505.14805/) (2025)

## Abstract (arXiv)

AI assistants in human-AI collaboration often correct suboptimal human actions through behavioral feedback (e.g., alerts or steering-wheel nudges in assistive driving). Such interventions can mitigate immediate errors, but long-term improvement requires addressing the underlying misconceptions that cause repeated mistakes. We introduce SENSEI, a framework that infers user misconceptions from interaction behavior and provides targeted, minimal yet sufficient suggestions to correct them. Our approach departs from action- or trajectory-level interventions by operating over a structured knowledge representation to localize and correct the sources of erroneous behavior. Across three long-horizon tasks with diverse misconceptions and corresponding behaviors, SENSEI demonstrates zero-shot compositional generalization, disentangling multiple overlapping misconceptions despite training only on single-misconception cases. A user study further shows that our method identifies real human misconceptions and provides effective guidance that improves long-horizon task performance, successfully correcting 90% of student misconceptions. Code and project page are available at https://misoshiruseijin.github.io/SENSEI/.