arXiv AI
Jul 24
AI Assistants Overassist
arXiv:2607. 21306v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used as tutors and thought partners, helping users reason through problems.
AI Assistance Reduces Persistence and Hurts Independent Performance
arXiv:2604. 04721v3 Announce Type: replace Abstract: People often optimize for long-term goals in collaboration: A mentor or companion doesn't just answer questions, but also scaffolds learning, tracks progress, and prioritizes the other person's growth over immediate results.
Related stories
arXiv AI
Aug 25
How AI Assistance Affects Human Skill Development: A Study of Learning with Logic Puzzles
arXiv:2608.23543v1 Announce Type: new Abstract: While AI assistance can improve human task performance in the short term, it may also undermine the development of skills in the longer term. We examin...
arXiv AI
Aug 20
What is Missing from AI Post-Training AI: An Empirical Analysis
The paper investigates the limitations of post-training AI agents that can autonomously train large language models. It distinguishes between execution-level capability—making adjustments within a chosen training strategy—and strategy-level capability—revising the overall approach based on new evidence. Analysis of many public post-training runs shows that agents lock into a strategy early and then only perform local tweaks, regardless of task. Experiments with experience scaffolds, human guidance, and extra compute improve execution but do not enable strategy reevaluation, indicating that agents lack a mechanism to spontaneously reassess their strategy during training.
arXiv Machine Learning
Sep 21
Rewarding Efficient Reasoning Improves Abstention on Underspecified Tasks in Reasoning Models
The paper introduces a new reward, GRPO, that encourages large reasoning models (LRMs) to efficiently determine whether a task is solvable before generating a full chain of thought. Fine‑tuning 4B LRMs with this reward improves their ability to abstain from answering unanswerable prompts by an average of 12.8% while producing 44% shorter chains of thought. The approach also preserves the models’ overall answering performance.
arXiv AI
Sep 4
Efficient Test-Time Adaptation through Human-AI Interaction
The paper introduces Test-Time Adaptation through Human‑Agent Interaction (TAHI), a method that uses iterative human feedback to adapt AI agents to individual users’ criteria. By integrating cross‑session interaction data into agent context and weights, and building an evolving rubric module, the authors demonstrate that agents can improve task success by 4.5–20.9% after only a few interactions. The evolving rubric also serves as a scalable annotation tool, detecting 16.0–22.3% more failures than language models or humans alone, and personalized agents can even generalize improvements up to 8.8% across users.
arXiv AI
Aug 19
On the Fragility of Self-Improving Agents: Variance, Task Order, and Underspecification
The paper re‑evaluates memory‑based self‑improving agents by adding multiple runs to measure variance and by randomizing task order. It finds that agent performance is noisy in complex, multi‑step environments and that improvement depends heavily on the sequence of tasks, revealing a hidden curriculum effect. The authors suggest that underspecification of tasks and environments contributes to this fragility and demonstrate that adding detailed rubrics and feedback can partially mitigate performance drops, though gaps remain.
arXiv AI
Jun 2
Beyond Access: Guided LLM Scaffolding for Independent Learning in Undergraduate Statistics
arXiv:2606. 01375v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly entering students' learning practices, but their educational value depends on whether they support reasoning or enable task completion without engagement.
arXiv AI
Aug 17
Demystifying Agent Skills: Why They Work-Until They Don't
arXiv:2608. 14036v1 Announce Type: new Abstract: Skills have emerged as a practical and effective approach for enhancing LLM agents at inference time through structured packages of knowledge.
arXiv AI
3d ago
Examining Variation in How Guided AI Tutors Resolve Student Impasses
The study analyzes 20,462 student turns from 1,260 sessions with a guided LLM chemistry tutor, identifying 6,630 impasse turns categorized as conceptual errors, expressed uncertainty, or help‑seeking. Three tutoring conditions—baseline, no‑direct‑answer, and guided—were simulated, revealing that the baseline tutor often gave direct answers, the no‑direct‑answer tutor always asked follow‑up questions, and the guided tutor varied its responses based on context. Impasse trajectories showed that each additional impasse turn reduced the likelihood of recovery, while addressing errors became increasingly beneficial compared to repeated scripted questioning.
arXiv AI
Aug 28
TutorTrace: A Dataset and Taxonomy for Classifying Learner Behavioral States during AI-Assisted Programming Education
TutorTrace is a new dataset and behavioral abstraction pipeline that captures learners’ low‑level IDE telemetry to make their behavioral context visible and computable in real time. The dataset, collected across 480 students in two introductory Python courses, includes 180 K telemetry events, 13 633 behavioral segments, and 27 continuously computed metrics, and it underpins a taxonomy of learner activity before, between, and after AI queries. Preliminary classroom tests show that behavior‑aware prompts reduce the time between queries, and the system can predict upcoming queries with AUROC scores of .726 and .717 on two held‑out tasks.
arXiv AI
Jun 4
Learning While Acting: A Skill-Enhanced Test-Time Co-Evolution Framework for Online Lifelong Learning Agents
arXiv:2606. 04815v1 Announce Type: cross Abstract: Lifelong learning is essential for Large Language Model (LLM) agents operating in dynamic, interactive environments.
arXiv AI
6d ago
Learning to Stop without Learning to Stop: Self-Supervised Confidence Training Improves Reasoning Efficiency
The paper demonstrates that fine‑tuning reasoning models to predict their own confidence at intermediate steps—using only 600 self‑supervised examples—substantially improves inference efficiency. Without adding any explicit stopping or length penalties, the models generate up to 25 % fewer tokens while maintaining accuracy on mathematical, scientific, and coding benchmarks across several architectures. The study finds that confidence supervision preserves the models’ high‑level reasoning structure rather than merely suppressing specific behaviors.