What the study found
The review found a dual impact of Generative AI, or GenAI, in computer science education: it can both hurt and help learning. The authors describe negative effects from hallucinated, meaning misleading or incorrect, outputs, as well as positive effects when GenAI is used in structured and pedagogically grounded settings.
Why the authors say this matters
The study suggests that GenAI should be designed and used in ways that support equitable, cognitively balanced, and instructionally effective learning environments. The authors conclude that their integrative framework may help explain how GenAI changes learning performance, hallucination dynamics, and problem-solving in computing education.
What the researchers tested
The researchers carried out a systematic review of 64 empirical studies on GenAI in computer science education. They focused on programming, debugging, algorithmic reasoning, and computational problem-solving, drawing on constructivist, sociocultural, cognitive load, adaptive learning, and metacognitive learning theories.
What worked and what didn't
According to the review, misleading GenAI outputs can raise extraneous cognitive load, which is the mental effort spent on irrelevant or confusing information, and can encourage over-reliance on system-generated content. These effects may disrupt error detection, self-monitoring, and problem-solving, and may widen educational disparities in low-resource settings or for culturally and linguistically diverse learners. In contrast, when GenAI is used in structured and equitable environments, it can support reflective programming practice, self-monitoring, verification, strategic adjustment, problem-solving skills, engagement, and personalized learning outcomes.
What to keep in mind
This is a review of published empirical studies rather than a single experiment. The abstract does not give detailed limitations beyond noting that effects depend on how GenAI is embedded in learning environments.
Key points
- The review synthesized 64 empirical studies on GenAI in computer science education.
- GenAI showed a dual impact: it could hinder learning when outputs were misleading and support learning when used in structured settings.
- Hallucinated outputs were linked to higher extraneous cognitive load and over-reliance on AI-generated content.
- Positive effects were reported for reflective programming, self-monitoring, verification, and strategic adjustment.
- The review notes potential inequities for low-resource settings and for culturally and linguistically diverse learners.
Disclosure
- Research title:
- GenAI shows mixed effects in computer science learning
- Authors:
- Adedeji Adefisoye Adejumo, Solomon Sunday Oyelere, Ismaila Temitayo Sanusi, Jarkko Suhonen
- Institutions:
- Luleå University of Technology, Modibbo Adama University of Technology, University of Eastern Finland, University of Eastern Finland, University of Eastern Finland, University of Exeter
- Publication date:
- 2026-03-14
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.