Title: Demystifying Machine Learning: An Assessment of the AIcon2abs Approach

Abstract

Building upon earlier research, this paper introduces an evaluation of the AIcon2abs methodology—standing for "AI from Concrete to Abstract: Demystifying Artificial Intelligence to the general public." The primary objective of this study is to gauge the efficacy of this innovative framework in enhancing machine learning (ML) comprehension among a broad demographic, ranging from K-12 students to adults.

At the core of the AIcon2abs method is the WiSARD algorithm, a type of weightless neural network prized for its accessibility and structural simplicity. Because WiSARD operates independently of internet connectivity, it is particularly well-suited for non-technical audiences and settings with limited resources. This offline capability allows the system to learn efficiently from very small datasets, even single examples. Consequently, users can directly observe the machine’s accuracy improve incrementally as more data is introduced.

The approach facilitates an intuitive grasp of ML mechanics by engaging participants in hands-on activities that simulate the behavior of algorithms. Through these practical exercises, users can visualize and interact with the internal processes of training and classification. Furthermore, WiSARD creates mental images that represent what it has learned, thereby emphasizing the critical features of the data being classified.

To test the method, the researchers conducted a six-hour remote course involving 34 participants from Brazil, comprising five children, five adolescents, and 24 adults. The data analysis employed a dual perspective: a mixed-method pre-experiment that included hypothesis testing, alongside a qualitative phenomenological analysis. The findings indicated that nearly all participants held a positive view of AIcon2abs, with results reflecting a high level of satisfaction regarding the achievement of the study’s educational goals. The study received ethical approval from the CEP-HUCFF-UFRJ Research Ethics Committee.

Source: arXiv Generated at: 2026-06-04 00:00:00 UTC