Observed Signal · Sep 6, 2026 · Research Study · Source: t3n · Impact: 2/5 · Sentiment: Negative
Long-term study: AI tutoring requires human guidance
A two-year study by Brown University examined the use of the AI tutor Khanmigo among middle school students in Tennessee. Despite high initial usage (96% tried it), actual sustained use was low. Many students sent off-topic messages or tried to extract answers. Math performance improved slightly (1.3 percentile points per semester, 0.06-0.08 SD annually), similar to practice without AI. The researchers concluded that access alone is insufficient; human supervision is crucial to realize benefits.
Reveals limitations of AI tutoring without human oversight, relevant to EdTech and AI adoption.
Track Real-Time Education Technology Signals & Market Shifts
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Brown University conducted a two-year study at 18 middle schools in Tennessee.
- 96% of students tried Khanmigo at least once, but sustained use was rare.
- Math scores improved by 1.3 percentile ranks per semester with the AI tutor.
- Researchers found that human guidance is necessary for effective AI tutoring.
- Some students attempted to trick the AI into giving direct answers.
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
PISA Study: AI Chatbots Offer No Learning Boost
The PISA 2025 study reveals that students who frequently use AI chatbots for schoolwork tend to score lower than those who do not. Students who never use AI to summarize texts average over 505 points, while daily users average around 475 points. Similar gaps appear for AI-assisted research and text generation. However, moderate use of AI as a learning aid (1-2 times weekly) yields similar scores to non-use. The OECD emphasizes that the results do not prove causality but highlight the need for guided instruction on critical AI usage. They warn that socioeconomically disadvantaged students are less likely to be taught to scrutinize AI-generated content, risking a new digital divide. The report recommends integrating digital and media literacy, including AI, into curricula and investing in teacher training.
Students Left to Learn AI Themselves
A Studyflix press release and accompanying survey reported by t3n found that many young learners teach themselves how to use AI tools because schools seldom address AI use in class. In a sample of 1,012 users of the Studyflix platform (pupils, students, apprentices, teachers and parents), 74% said they had taught themselves how to use AI and only 4% said a teacher had taught them. Respondents reported using AI mainly to get explanations and ask follow-up questions rather than to outsource entire exams, though concerns remain about skill loss and unfair advantages from AI-assisted cheating. Studyflix CEO Mathias Klement urges schools to teach proper AI use and set clearer rules.
Peer Pressure Raises Parents' Willingness to Pay for AI Tutors
An international research team surveyed about 2,000 parents in the US, Canada and the UK to measure how peer adoption of AI homework tools affects parental willingness to pay. Parents reported the maximum they would pay for a three-month premium subscription under scenarios where 20% to 80% of peers used such tools. Results showed that each 10 percentage-point increase in assumed peer usage raised willingness to pay by $1.83; the gap between 20% and 80% adoption equated to roughly an $11 (≈60%) higher willingness to pay. A related field study found students who practiced math with GPT-4 later scored about 20% lower on tests without AI. Information about potential long-term harms increased parental support for school-level rules from 44% to 57%. The story notes policy responses in education, such as Denmark requiring oral defenses and the University of Chicago Law School implementing a laptop ban for first-year courses.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.