The Coach's hint budget was managed poorly: students split into early-spenders and savers, and within the Coach condition only those with a deliberate hint strategy scored higher¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r1 · q2
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q2 i? Interaction logs showed two distinct hint-budget patterns: spending most hints early on first problems, or saving hints and finishing with most of the budget unused. → Azimi 2026 q2 i? Within the Coach condition, students with a deliberate strategy for when to spend a hint scored higher on assignments than students who saved hints without a plan. → Azimi 2026 (2)

Evidence¶

Azimi 2026¶

Azimi, S. (2026). AI Literacy over Tool Design: A Mixed-Methods Study of Scaffolded versus Unrestricted Generative AI in Programming Education. Delft University of Technology. https://arxiv.org/abs/2609.16784 doi:10.48550/arxiv.2609.16784

q2 · i? · causal · r1

Log analysis of Coach interactions in the pilot, examining how students distributed the 25-hint budget over a session. The logs show students who "spent most of their budget early in a session" and others who saved hints; several complained about the limit.

"Some students spent most of their budget early in a session, on the first problems they encountered, and had few hints left for the more demanding tasks at the end of the session."

Azimi 2026 (2)¶

Azimi, S. (2026). AI Literacy over Tool Design: A Mixed-Methods Study of Scaffolded versus Unrestricted Generative AI in Programming Education. Delft University of Technology. https://arxiv.org/abs/2609.16784. doi:10.48550/arxiv.2609.16784

q2 · i? · causal · r1

Descriptive within-Coach-condition comparison of assignment rubric scores against logged hint-use patterns. The article concludes the budget "rewarded a capacity for self -governance that some students already possessed; it did not develop that capacity"; no effect size is printed.

"the students who had developed a deliberate strategy for when to spend a hint and when to hold back scored higher on the assignments than the students who saved their hints without a plan."

Discussion¶

Related Claims¶

• Effective AI help-seeking is itself a skill unevenly distributed among students, and access to an assistant is not equivalent to effective help-seeking — the general claim this finding bears on, with the studies behind it

• Persistence and performance costs concentrate among participants who used the AI to obtain direct solutions, while hint-users resemble controls — the general claim this finding bears on, with the studies behind it

• Letting learners pace instruction helped on complex material in one experiment, while learner control in general showed almost no effect across 18 studies — related
• Scaffolding improves learning — related
• A scaffolded AI Study Coach produced no assignment-performance or concept-inventory advantage over unrestricted AI use in a master's programming course pilot — related
• Students who formulated their own rules for when to use AI performed better on assignments in both conditions — related
• Prompt-literacy guidance was associated with higher transcript-coded rule-following compliance than unrestricted LLM access, mainly in process-over-answer and stepwise-hint behaviors — related
• Students with the best self-taught understanding of how language models work used AI tools most deliberately and achieved the highest assignment scores — related