Beyond AI (11): 3 takeaways for teachers: Letting computers compute and tutors tutor

In order to make the decision which GenAI tool to use how and when, teachers need to know what these tools are good at, considering that each tool relies on a number-crunching machine. In the context of language education, the answer is conversation practice and machine translation, and not language learning per se. This conclusion is rooted in the difference between human learning and machine learning, as understood in Activity Theory (Леонтьев, 1975):

  • Humans learn everything as an action. An action is intention-driven. Meeting or not meeting the intention is the main feedback for the learning human. Some learning can result in a new ability after a single action. When actions are repeated often, they get automatized and can become condition-triggered operations, which are less conscious and are not driven by an intention and can become a new skill.
  • Machines “learn” by computing probability weights and recording patterns in the artificial neural network of a large language model. Linear, deterministic feedback is often given through human reinforcement.
This is an excerpt from an early draft of a book chapter. The book will be about language teacher education and GenAI. I am posting these in smaller (mostly) self-contained posts. The posts are numbered consecutively. After they will have all come out, I will link them with each other.

In language use this results in the following basic facts:

  1. Humans create linguistic signs such as words, sentences, and texts that have form and meaning (Saussure, 1916 / 2011). Machines generate plausible language forms (tokens) that have no meaning on their own.
  2. Humans make communicative meaning (with texts) intentionally (Vygotsky, 1934/1978). Machines generate highly probable token sequences that mimic human language, when they are triggered by a prompt.
  3. Humans imbue generated texts with meaning and read it and react to it according to the meaning with which they themselves imbued the complex, plausible token sequences.
woman teaching girls sitting by the table
Photo by Gustavo Fring on Pexels.com

What does this mean for language education? On the one hand, sustained communication opportunities for learners are still lacking, although the general situation is better than it was for my English learning. Chatbots, when used in a planful and considered manner, can provide sustainable individual conversation practice opportunities. On the other hand, the situation is deceptive, because chatbots generate plausible synthetic text, text that is pure form and has no meaning for the machine and that depends on the quantity (Trosterud, 2023) and quality of the training data used for the LLM, and therefore are not truly communicating with the student (Searle, 1980). It is also important to remember that the texts of the training corpus and the processes of machine learning and of generating synthetic text are unknown to at least the large majority of GenAI users (Bender & Hanna, 2025). We are interacting with the computer, a machine.

The interaction of the learner with the computer was called tutorial CALL (Heift & Schulze, 2015; Hubbard & Bradin-Siskin, 2004; Schulze, 2024). When I started in CALL in the 1990s, the computer either functioned as a tool (an electronic dictionary, for example) or as a tutor (a grammar learning activity, for example). Since the mid-nineties and until the advent of GenAI chatbots in the mainstream, most CALL was computer-mediated communication (e.g., email tandems, chat, wikis and blogs, social media), the communication not with but via the computer with other humans: learners, a teacher, or first-language speakers. Today, when we interact with GenAI tools and chatbots, we interact with a machine, which is more akin to tutorial CALL than to computer-mediated communication. Yet, it “feels” more like the interaction with another human, if we do not reflect on the current interaction with the machine sufficiently.

A student model has information about each student over time and across learning areas, and it can infer future behavior with some probability.

A machine with a GenAI tool is not a tutor, because LLMs do not have a feedback algorithm, a student model (Schulze, 2008, 2025), or even the memory of past interactions. However, GenAI is a powerful tool for language practice, which teachers and students need to learn to use. On the other hand, tutoring – at the basic level: determining instructional sequences and providing contingent, contextualized, and individualized feedback – is one of the teacher’s responsibilities, until intelligent tutoring systems – another branch of AI research – have become mainstream in language education.

References

Bender, E. M., & Hanna, A. (2025). The AI con: How to fight big tech’s hype and create the future we want. Harper Collins Publishers.

Heift, T., & Schulze, M. (2015). Tutorial CALL. Language Teaching, 48(4), 471–490.

Hubbard, P., & Bradin-Siskin, C. (2004). Another Look at Tutorial CALL. ReCALL, 16(2), 448–461.

Saussure, F. d. (1916 / 2011). Course in general linguistics. Eds. Charles Bally & Albert Sechehaye. Trans. Wade Baskin, subsequently edited by Perry Meisel & Haun Saussy. Columbia University Press.

Schulze, M. (2008). Modeling SLA Processes Using NLP. In C. Chapelle, Y.-R. Chung, & J. Xu (Eds.), Towards Adaptive CALL: Natural Language Processing for Diagnostic Assessment. (pp. 149–166). Iowa State University.

Schulze, M. (2024). Tutorial CALL — Language practice with the computer. In R. Hampel & U. Stickler (Eds.), Bloomsbury Handbook of Language Learning and Technologies (pp. 35–47). Bloomsbury Publishing.

Schulze, M. (2025). ICALL and AI: Seven lessons from seventy years. In Y. Wang, A. Alm, & G. Dizon (Eds.), Insights into AI and language teaching and learning (pp. 11–31). Castedown Publishers.

Searle, J. (1980). Minds, brains, and programs. Behaviorial and Brain Sciences, 3(3), 417–457.

Trosterud, T. (2023). CALL for all languages? Why languages differ and what consequences that has for CALL. Keynote presentation Eurocall 2023, Reykjavik.

Vygotsky, L. S. (1934/1978). Mind and Society: The Development of Higher Mental Processes. Harvard University Press.

Леонтьев, А. Н. (1975). Деятельность. Сознание. Личность. Политиздат.