Introduction
The emergence of ChatGPT, or generative AI (Artificial Intelligence), signals a seismic shift in the landscape of education. This new entity, arising from advancements in various fields of academia and digital technology, fascinates not only ESL teachers and learners but also those in other content areas with its unprecedented performance. Such remarkable performance has been well reflected in the phenomenal response from the public worldwide since its disclosure in November 2022, including its benefits to general education, medicine, business, arts, and various fields in ESL education.
Meanwhile, there is a growing tendency among scientists and educators not to view generative AI merely as a machine under the control of Homo sapiens, but to treat it as a potential candidate for a new species with intelligence that may pose a threat to human beings. The escalating concern is not solely the responsibility of AI scientists; it should also be addressed by ESL professionals. The performance of ChatGPT is not far from ESL education, as it produces the English language with almost perfect accuracy and fluency, largely overlapping what ESL teachers and learners do. Additionally, a significant distinction between ChatGPT and previous CALL (Computer-Assisted Language Learning) applications is the level of agency involved. In traditional CALL environments, ESL teachers and learners could exert control or participate actively as agents. However, with ChatGPT, such direct involvement might become less essential in the near future.
The Chronicles of AI: Turing, Perceptron, and Deep Learning
In the beginning of a long journey to discuss the genesis of the universe, Tonelli (2021) sanguinely imagines that Neanderthals living in Europe 40,000 years ago created various symbols (e.g., paints on cave walls) and used them directly or indirectly to tell their stories from generation to generation. He confidently states that the ancient tradition continues to go all the way down to the present. A historical overview of the development of AI since the early 1900s tells us a story: A story with imagination, hope, disgruntlement, rigor, tension, aspiration, collaboration and triumph.
In this story, there appears not the ancient symbols but a modern craft, thinking machine. Depending on story tellers in the field of philosophy, psychology and computer science, the beginning of the story varies from the ancient Greek philosophers (e.g., Aristotle), the philosophers or mathematicians in the 16th and 17th century (e.g., Hobbes and Descartes) to the modern scientists. Although many scholars are said to philosophically and conceptually contribute to the birth of the thinking machine in the history of science, Turing (1950) is commonly regarded as the starting point of its story from both epistemological and ontological point of view.
Innerspace
Swans on a lake appear serene, yet beneath the water, they are busy paddling hard to maintain their serenity. Similarly, although the outcomes from ChatGPT may seem to come easily, the underlying artificial neural networks are engaged in rapid and intricate processes. The same is true for our brains. When we see a cat and simply say “a cat,” our brains are actively processing this ‘cat’ information through a swift and convoluted mechanism.
As an ESL professional and merely a consumer of generative AI, it is acceptable to use it for simple, straightforward outcomes; however, as an ESL professional and a pedagogical user of generative AI, it is far more beneficial to understand the underlying principles behind the apparent outcomes. This not only helps in using generative AI as a teaching or research aid but also deepens our understanding of it. Similar to ESL educators in reading programs, who are not satisfied with simply knowing a learner’s level of reading proficiency, they thus aim to probe into the cognitive processes occurring within the learner’s mind during reading. These coveted insights are provided by L2 reading process models, knowledge of which significantly helps to enhance ESL teachers’ instructional strategies and enrich their understanding of how learners read and interpret text in the target language.
Homo sapiens and Machina sapiens
Upon opening to the public at the end of 2022, ChatGPT set numerous records in online history, including the number of users within a week, the growth rate of users, and the speed of growth. With its phenomenal popularity, debates have been ongoing regarding the epistemological and ontological issues of ChatGPT. Among the debates, this evokes a galactic collision between two leading figures, Chomsky and Hinton, from two different disciplines in terms of philosophical issues of ChatGPT: whether it has intelligence or not.
In the article of about 1,610 words, Chomsky and his colleagues (Chomsky, Roberts, & Watumull, 2023) degrade the achievement of ChatGPT, viewing it as “something like the banality of evil: plagiarism and apathy and obviation.” Seemingly opposed to the intelligence of ChatGPT by means of thought and language, their argument disguisedly disavows ChatGPT itself, as read in the concluding remark, “Given the amorality, faux science and linguistic incompetence of these systems, we can only laugh or cry at their popularity.” And Hinton (University of Oxford, 2024) directly refutes them.
Il buono, Il brutoo, Ilcattivo
ChatGPT’s exceptional performance has sparked public fascination and excitement, reminiscent of the response to the unveiling of the Perceptron approximately 70 years ago. However, the excitement has recently faltered. This shift is not attributed to deficiencies in ChatGPT’s performance; rather, it is because its capabilities are so preeminent that they are poised to soon surpass human capacities, both physical and mental. A symbolic example of this dithering over the advancement of generative AI is Hinton’s resignation from the advisory boards of major AI tech companies. His proactive effort to publicly highlight the potential dangers associated with AI’s rapid advancements cautions that its exceptional capabilities necessitate more measured approaches to development. The heightened awareness of the risks has spread widely among societal leaders, researchers, and the public.
In May 2024, global leaders convened at the United Nations’ (UN’s) AI for Good Global Summit to deliberate on the dual aspects of “the hope and fear of AI development.” The summit primarily assessed the alignment of recent AI advancements with the UN’s Sustainable Development Goals (SDGs). However, the explicit mention of “fear” highlights an escalating concern about both the rapid pace of AI advancements and their significant achievements. Heikkilä (2024b) provides an analysis of the post-conference sentiments, emphasizing a predominant inclination towards “fear” rather than “hope”.
AI and ESL Education: From CALL 3.0 toward CALL 4.0 ?
What measures should be taken when confronted with a replicant AI teacher whose proficiency either matches or surpasses that of a human ESL educator? Reflecting on Chapelle’s (2001, p. 1) comments, it will become necessary to initially address the question of whether a replicant AI teacher “should be used” before considering how a replicant AI teacher “can best be used.” If the relevance of the latter inquiry proliferates, one might then question the continued necessity of human teachers. Moreover, as Otto (2017, p. 20) has highlighted, what implications would arise if replicant AI teachers remain constant in presence but evolves in format and capability, eventually becoming “a normal part of everyday practice” in all aspects of future ESL education? The human-like AI teacher “may initially be developed by humans, but as it grows it follows its own path, going where no human has gone before – and where no human can follow” (Harari, 2017, p. 458).
The potential for unexpected and previously unencountered scenarios can be attributed to the introduction of an obscure and enigmatic dimension of AI within CALL, which extends more broadly to the field of ESL education. Such development may require ESL professionals to adopt novel conceptual and instructional strategies to effectively tackle these challenges. As Lanier (Bloomberg Originals, 2023, 3:35-3:39) articulates, “to be an optimist, you have to have the courage to be a fearsome critic” when integrating AI technology. As depicted in Figure 9, the future trajectory of CALL could diverge significantly depending on the efficacy with which these new AI-related challenges are addressed. Specifically, this could result in the evolution of CALL into either CALL 4.0, where AI technologies enhance the richness and efficacy of L2 research and instruction, making it more conceptually and pedagogically robust, or CALL X, where the AI technology could lead to CALL becoming lost and conceptually and pedagogically deprived due to AI’s capacity to autonomously manage all aspects of L2 learning.
