AI from concrete to abstract: demystifying artificial intelligence to the general public

A methodology called AI from concrete to abstract (AIcon2abs) aims to demystify artificial intelligence by having the public build and observe learning machines through hands-on activities, according to its creator [1]. The approach, detailed in a paper by Rubens Lacerda Queiroz, combines visual programming with WiSARD weightless artificial neural networks to make machine learning processes visible and understandable [1]. Unlike conventional AI teaching tools that treat a trained model as an external module plugged into an application, AIcon2abs integrates training and classification tasks as blocks within the main program itself [1]. This design, Queiroz writes, makes the distinction between a program that learns from data and a conventional program more evident [1]. The WiSARD algorithm is central to the method’s accessibility. It is a weightless neural network that does not require an internet connection and can learn from a minimal dataset, even a single example [2]. Users can observe how the machine incrementally improves its accuracy as it receives more data, and the system generates mental images representing what it has learned, highlighting essential features of the classified data [2]. A subsequent study evaluated AIcon2abs through a six-hour remote course with 34 Brazilian participants: 5 children, 5 adolescents, and 24 adults [2]. The research, approved by the CEP-HUCFF-UFRJ Research Ethics Committee, analyzed data through a mixed-method pre-experiment and a qualitative phenomenological analysis [2]. Nearly all participants rated the experience positively, and results demonstrated a high degree of satisfaction in achieving the intended outcomes [2]. The methodology’s stated goal is to equip people with skills that help them become insightful actors in debates and decisions involving the adoption of artificial intelligence mechanisms [1][3]. By enabling participants to interact with machine learning processes as if they were the algorithms themselves, the method aims to build understanding through direct observation rather than abstract explanation [2].