Artificial intelligence and the mobilities of inclusion

TitleArtificial intelligence and the mobilities of inclusion
Publication TypeBook Chapter
Year of Publication2019
AuthorsGallagher M
EditorKnox J, Wang Y, Gallagher M
Book TitleInclusive Education, ICT, and Artificial Intelligence
Series TitlePerspectives on Rethinking and Reforming Education
Pagination179 - 194
PublisherSpringer
CityCham
Keywordsartificial inteligence, digital divide, ICT4Development, INCLUSION, Mobile Applications/utilization, mobile learning
Abstract

The burgeoning use of artificial intelligence in a learning increasingly mediated through mobile technology makes inclusion problematic. This is largely due to the sheer ubiquity of mobile technology globally, the complexity of the machine learning regimens needed to function within increasingly sophisticated 5G cellular networks, and the legions of professionals needed to initiate and maintain these AI and mobile ecosystems. The promise of artificial intelligence in inclusion is curtailed precisely due to the accumulated advantage (the Matthew effect) presented in such a technological sophistication: only those with the most sophisticated and agile of educational systems will stand to benefit, a scenario that poses significant impact on inclusion strategies increasingly mediated through ICT. Further, the accumulated advantage is not only a dichotomy between those with access to these sophisticated technologies and those that do not enjoy that same privilege. A further parallel is between the “curriculum” of machine learning of artificial intelligence in 5G networks as outlined in Li et al (2017) and traditional and human-centred educational curricula that is being increasingly redrawn as a reductionist enterprise aligned with national and international quantitative metrics. While AI has evolved to include multidisciplinary techniques such as machine learning, optimization theory, game theory, control theory, and meta-heuristics in both supervised and unsupervised formats, traditional educational curricula is increasingly influenced by third party commercial enterprises and reductionist moves towards computational thinking. This poses significant disadvantage to educational inclusion beyond technological advantage, the sophistication of machine learning curricula, or the general paucity of human curricula increasingly modeled on computational thinking; the biases at work in larger society are encoded into the datasets that machine learning operates. Inclusion operates, statistically, as an outlier in these data-driven environments; as an equitable model in education, largely designed to counter prevailing societal biases, rather than conform to them. The equity that this inclusion seeks to provide is not inherently a naturally-occuring entity and will not render naturally in the data that AI is learning from. As more and more education is engaged through mobile technology and more and more of that mobile education is driven by an artificial intelligence emerging from curricula of greater and greater sophistication, a situation emerges that poses great challenges for any sort of meaningful inclusion, particularly in the potential acceleration of entrenched advantage.

DOI10.1007/978-981-13-8161-4_11
Abstract

The burgeoning use of artificial intelligence in a learning increasingly mediated through mobile technology makes inclusion problematic. This is largely due to the sheer ubiquity of mobile technology globally, the complexity of the machine learning regimens needed to function within increasingly sophisticated 5G cellular networks, and the legions of professionals needed to initiate and maintain these AI and mobile ecosystems. The promise of artificial intelligence in inclusion is curtailed precisely due to the accumulated advantage (the Matthew effect) presented in such a technological sophistication: only those with the most sophisticated and agile of educational systems will stand to benefit, a scenario that poses significant impact on inclusion strategies increasingly mediated through ICT. Further, the accumulated advantage is not only a dichotomy between those with access to these sophisticated technologies and those that do not enjoy that same privilege. A further parallel is between the “curriculum” of machine learning of artificial intelligence in 5G networks as outlined in Li et al (2017) and traditional and human-centred educational curricula that is being increasingly redrawn as a reductionist enterprise aligned with national and international quantitative metrics. While AI has evolved to include multidisciplinary techniques such as machine learning, optimization theory, game theory, control theory, and meta-heuristics in both supervised and unsupervised formats, traditional educational curricula is increasingly influenced by third party commercial enterprises and reductionist moves towards computational thinking. This poses significant disadvantage to educational inclusion beyond technological advantage, the sophistication of machine learning curricula, or the general paucity of human curricula increasingly modeled on computational thinking; the biases at work in larger society are encoded into the datasets that machine learning operates. Inclusion operates, statistically, as an outlier in these data-driven environments; as an equitable model in education, largely designed to counter prevailing societal biases, rather than conform to them. The equity that this inclusion seeks to provide is not inherently a naturally-occuring entity and will not render naturally in the data that AI is learning from. As more and more education is engaged through mobile technology and more and more of that mobile education is driven by an artificial intelligence emerging from curricula of greater and greater sophistication, a situation emerges that poses great challenges for any sort of meaningful inclusion, particularly in the potential acceleration of entrenched advantage.