Many popular habits, like rereading, highlighting, and last-minute cramming, create a feeling of fluency that rarely turns into durable learning. Decades of research point to a different toolkit: retrieval practice, spaced practice, and, when appropriate, interleaving, paired with timely feedback. These approaches improve long-term retention and transfer across subjects, including the SAT and ACT. Early randomized trials of AI, also called intelligent tutors or LLM chatbots, show promising gains when these tools are designed around the same principles, although safeguards for privacy, accuracy, and equity are essential. The fix is not to study more, it is to study differently, and to measure learning with delayed, mixed, and self-tested work rather than immediate ease. (Dunlosky et al., 2013; Pan & Rickard, 2018; Kestin et al., 2025). SAGE JournalsNature

TL;DR

• Rereading and highlighting have low utility for durable learning (2013). SAGE Journals
• Retrieval practice reliably boosts retention and transfer (2018).
• Spacing study over days or weeks increases long-term memory (2006). augmentingcognition.com
• Interleaving often helps, but not for every kind of rule learning (2025). PMC
• AI tutors can double learning in some RCTs, design and safeguards matter (2025). Nature

Why do “comfortable” strategies fail?

Rereading textbooks, highlighting, and massed repetition make information feel familiar, which increases confidence during study. That fluency is often mistaken for learning, a well-documented metacognitive bias. Reviews stress that performance during practice can diverge from later learning, especially when study is easy and feedback is immediate. In other words, the brain confuses ease now with learning later. See Soderstrom & Bjork, 2015, and the “desirable difficulties” framework summarized by Bjork & Bjork, 2011. PubMedBjork Lab

A landmark review of ten common techniques found that rereading and highlighting usually offer “low utility” for long-term retention across materials and learners, while practice testing and distributed practice were “high utility.” This ranking has guided study-skills instruction for over a decade. See Dunlosky et al., 2013. SAGE Journals

Which “study habits” underperform, and why?

Rereading and highlighting

Passive review inflates familiarity without strengthening retrieval paths. Benefits are small and short-lived unless combined with active retrieval or spacing. The comprehensive evaluation by Dunlosky et al., 2013 rated both strategies as low utility for durable learning. SAGE Journals

Massed practice and cramming

Short bursts can lift performance on immediate tests, but retention drops steeply over days and weeks. A meta-analysis across a century of research shows spaced practice outperforms massed practice for long-term memory, with optimal spacing depending on how far in the future the test is. See Cepeda et al., 2006. augmentingcognition.com

Overconfidence during easy practice

When studying feels fluent, learners overestimate mastery and under-invest in effortful strategies such as self-quizzing. This learning–performance disconnect is reviewed in Soderstrom & Bjork, 2015. PubMed

What works better, and what are the limits?

Retrieval practice

Self-testing, low-stakes quizzes, and explaining answers from memory strengthen recall and concept use. A meta-analytic review concluded that test-enhanced learning improves not only retention but also transfer to new problems. See Pan & Rickard, 2018. Experimental work demonstrates retrieval practice outperforming elaborative methods like concept mapping under randomized conditions, for example Karpicke & Blunt, 2011. PubMed

How to apply:

1. After reading a passage, close the text and write what you remember.
2. Answer mixed practice questions without notes, then check answers and correct.
3. Schedule brief “cumulative quizzes” weekly to revisit earlier topics.

Spaced practice

Spacing short sessions across days or weeks produces larger and more durable gains than the same time massed together. Optimal intervals scale with the exam date, which is why SAT and ACT plans should lengthen spacing early and tighten spacing near test day. See Cepeda et al., 2006. augmentingcognition.com

How to apply:

1. Plan 20–40 minutes per topic, per day, over several days.
2. Use a calendar to revisit weak items just before they would be forgotten.
3. Combine spacing with retrieval, often called “successive relearning.” See teacher guidance in Dunlosky, 2015. American Psychological Association

Interleaving, with nuance

Mixing problem types trains discrimination, for example deciding when to use which algebra strategy. Classroom and lab studies commonly show interleaving improves test performance relative to blocking, including in K-12 math, although not universally. See Taylor & Rohrer, 2010 and the school-based syntheses summarized in Rohrer, 2015. A 2025 study reports that when learning explicit rules, blocked sequences can sometimes outperform interleaving, so match sequence to goals. See Little & Nepangue, 2025. Wiley Online LibraryERICPMC

How should SAT or ACT students use these methods week to week?

A simple 4-day loop per topic

1. Day 1, Learn and retrieve, read an example, then explain the steps from memory.
2. Day 2, Spaced retrieval, do 5–10 problems cold, check, and correct.
3. Day 3, Interleave, mix that topic with two other skills, choose the method from the problem itself.
4. Day 7, Cumulative quiz, sample prior topics, short and timed.

This loop implements retrieval, spacing, and interleaving for both math problem-solving and reading comprehension. Evidence for each pillar comes from the sources above, applied to exam-style practice. augmentingcognition.comWiley Online Library

Can an AI tutor help, or hurt?

What the latest trials show

A randomized controlled trial with college STEM students compared an AI tutor to in-class active learning with the same pedagogy. The AI group achieved higher post-test scores in less time, with median learning gains over double those of the class condition. This peer-reviewed study suggests that well-scaffolded AI tutors can accelerate learning when they enforce retrieval, stepwise problem solving, and targeted feedback. See Kestin et al., 2025. Nature

Guardrails that matter for teens

U.S. guidance encourages responsible AI use that supplements, not replaces, educators, with attention to accuracy, bias, and data protection under FERPA and related laws. Districts and vendors are urged to define permissible data flows, human oversight, and transparency. See the U.S. Department of Education’s 2025 Dear Colleague Letter and AI guidance resources (ED, 2025; ED, 2023 report; Student Privacy Policy Office). U.S. Department of Education+1Protecting Student Privacy

What should be done by people or by software?

How much should families budget?

As of August 2025, typical private SAT tutoring on large marketplaces costs about $35–$60 per hour, with higher ranges for specialized tutors. Branded packages can be $1,999 for 10 hours from major providers. Examples include Wyzant’s average ranges and Kaplan’s current packages. (Wyzant; Kaplan). WyzantKaplan Test Prep

AI-enabled options range from $4 per month for Khan Academy’s Khanmigo for Families to $20 per month for general AI access such as ChatGPT Plus, which some students use for structured self-quizzing and explanations. Pricing varies by features and eligibility. (Khan Academy, 2025; OpenAI). Khan Academy BlogOpenAI Help Center

FAQ

Are AI tutors effective for SAT or ACT prep?
Evidence is emerging. A 2025 randomized trial in university physics reported over double learning gains with an AI tutor designed around retrieval and stepwise feedback. Translating that to SAT or ACT requires alignment to exam domains and high-quality item sets, but the mechanism, retrieval with scaffolding, matches what works in test prep. (Kestin et al., 2025). Nature

How do AI tutors compare to human tutors on learning gains?
When AI enforces proven methods, gains can be large in short windows, but humans still add value for planning, motivation, and judgment. U.S. guidance frames AI as augmenting, not replacing, teachers. Families often get best results with a hybrid approach. (ED, 2025). U.S. Department of Education

What does a good hybrid study plan look like week to week?
Start with spaced retrieval five days per week, add interleaved mixed sets twice weekly, and meet once weekly with a teacher or parent to review errors and adjust difficulty. Use cumulative quizzes every 7 days. This mirrors the strongest effects in the literature. (Pan & Rickard, 2018; Cepeda et al., 2006). augmentingcognition.com

Are AI tutors safe for teens’ data and privacy?
Schools should apply FERPA and state rules, define what data are collected, where they are stored, and how long they are kept, and ensure a human is accountable for decisions. Parents can ask vendors for data maps and opt-out options. See the Student Privacy Policy Office guidance. (SPPO). Protecting Student Privacy

How can parents monitor quality without micromanaging?
Set weekly goals, require short reflections after sessions, and review error logs. Look for delayed gains on mixed quizzes, not just immediate correctness during guided steps. This aligns with the learning versus performance distinction. (Soderstrom & Bjork, 2015). PubMed

What are red flags that an AI tutor is giving bad advice?
Hallucinated formulas, steps out of sequence, or hints that never ask the student to retrieve. Prefer tools that chunk problems, check reasoning, and provide sources. The 2025 trial engineered guardrails for stepwise scaffolding for this reason. (Kestin et al., 2025). Nature

Do AI tutors help struggling students or widen gaps?
Design and support determine the answer. When AI is free or low-cost and paired with coaching, it can expand access to high-quality practice. Public guidance emphasizes equitable access and human oversight. (ED, 2025). U.S. Department of Education

Check out ExamJam and how we use these techniques to make your study more effective.

Sources

• Bjork, E. L., & Bjork, R. A. (2011). Making things hard on yourself, but in a good way, creating desirable difficulties to enhance learning. In Psychology and the Real World (pp. 56–64). Worth. https://bjorklab.psych.ucla.edu/wp-content/uploads/sites/13/2016/04/EBjork_RBjork_2011.pdf Bjork Lab
• Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks, A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380. https://augmentingcognition.com/assets/Cepeda2006.pdf augmentingcognition.com
• Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students’ learning with effective learning techniques. Psychological Science in the Public Interest, 14(1), 4–58. https://journals.sagepub.com/doi/10.1177/1529100612453266 SAGE Journals
• Karpicke, J. D., & Blunt, J. R. (2011). Retrieval practice produces more learning than elaborative studying with concept mapping. Science, 331(6018), 772–775. https://pubmed.ncbi.nlm.nih.gov/21252317/ PubMed
• Kestin, G., Miller, K., Klales, A., Milbourne, T., & Ponti, G. (2025). AI tutoring outperforms in-class active learning, An RCT introducing a novel research-based design in an authentic educational setting. Scientific Reports, 15, Article 97652-6. https://www.nature.com/articles/s41598-025-97652-6 Nature
• Little, J. L., & Nepangue, J. A. (2025). Whether interleaving or blocking is more effective for long-term learning depends on one’s learning strategy. Psychology. https://pmc.ncbi.nlm.nih.gov/articles/PMC12108632/ PMC
• Pan, S. C., & Rickard, T. C. (2018). Transfer of test-enhanced learning, A meta-analytic review. Psychological Bulletin, 144(7), 710–756. https://pubmed.ncbi.nlm.nih.gov/29595261/
• Rohrer, D., & Taylor, K. (2010). The effects of interleaved practice. Applied Cognitive Psychology, 24(6), 837–848. https://onlinelibrary.wiley.com/doi/10.1002/acp.1598 Wiley Online Library
• Rohrer, D. (2015). Interleaved practice improves mathematics learning. ERIC brief synthesizing classroom trials. https://files.eric.ed.gov/fulltext/ED557355.pdf ERIC
• Soderstrom, N. C., & Bjork, R. A. (2015). Learning versus performance, An integrative review. Perspectives on Psychological Science, 10(2), 176–199. https://pubmed.ncbi.nlm.nih.gov/25910388/ PubMed
• U.S. Department of Education. (2025, July 22). Dear Colleague Letter on AI. Press release and guidance page. https://www.ed.gov/about/news/press-release/us-department-of-education-issues-guidance-artificial-intelligence-use-schools-proposes-additional-supplemental-priority U.S. Department of Education
• U.S. Department of Education, Office of Educational Technology. (2023). Artificial Intelligence and the Future of Teaching and Learning. https://www.ed.gov/sites/ed/files/documents/ai-report/ai-report.pdf U.S. Department of Education
• U.S. Department of Education, Student Privacy Policy Office. Guidance index for FERPA and PPRA. https://studentprivacy.ed.gov/guidance Protecting Student Privacy