Analysis · Technology
Predicting Student Failure Is Only Half an AI Value Test
By AWEI · AI-compiled · Published · Analysis prepared · 1 source · doi.org
An AI system can flag students likely to struggle, yet that prediction alone does not show that anyone learns more. This distinction matters when reading a 2024 review by Muyideen Dele Adewale and colleagues on artificial intelligence in open and distance learning.
The authors report selecting 64 papers from an initial pool of 700, covering publications from 2017 to 2023. They searched Scopus, Google Scholar and Web of Science, screened titles and abstracts, assessed full texts and sought additional papers through snowballing. Their work synthesised existing research; it did not test a new teaching intervention.
A revealing finding concerns the evidence itself. The review includes substantial theoretical work alongside statistical studies and machine learning research. Its taxonomy table classifies 26 papers as nonempirical, including conceptual analyses and literature reviews. However, a later section gives a different percentage for that category, so the exact methodological breakdown requires caution.
The reviewed literature describes models using academic records and learning behaviour to forecast performance or identify students at risk. It also discusses personalised instruction and timely feedback. These are connected possibilities, but prediction accuracy and educational benefit are separate outcomes. Associations between AI use and achievement do not, by themselves, establish that adopting AI caused improvement.
Despite calling one section a meta-analysis, the authors explain that differing methods and outcome measures prevented quantitative pooling. The review therefore supplies no combined estimate of achievement gains. Its English-language and database restrictions may omit relevant evidence. The recommended framework for assessing AI across settings, genders and regions also remains untested.
For distance-learning institutions, the practical implication is to evaluate the whole sequence: a prediction, an educator’s response and a measured learning outcome. Technology access and institutional readiness belong in that assessment. This review helps identify questions for a local evaluation; it cannot establish how much a particular deployment will improve results.
Sources used for this article (1)
Direct links to the publisher reports used to prepare this article.
- Source 1
- Impact of artificial intelligence adoption on students' academic performance in open and distance learning: A systematic literature review M. D. Adewale, A. A. Azeta, Adebayo Abayomi‐Alli …
