A technical background on artificial intelligence and intelligent language models

Abstract

Stunning advancements in artificial intelligence (AI) over the last several years have undoubtedly opened new possibilities and challenges for the field of second language learning. Of course, AI is not new, and for decades it has attracted the interests and imaginations of philosophers, computer scientists, and the public. Though the performance of AI systems had typically been underwhelming until recently, the results of current systems have made AI relevant for nearly everyone. Much of the focus in language education has been on exploring how these new systems may be used by teachers, researchers, and especially learners. This paper provides a brief technical primer on the discoveries that have helped make the AI revolution possible, including algorithmic discoveries in deep learning such as backpropagation and the attention mechanism of transformers, and hardware improvements that have made it feasible to train such massive models. Understanding the background and underlying principles behind AI may benefit language teachers and researchers who are interested in demystifying the technology and understanding its long-term potential.