75447 2712-9934 18+ 7 3 24 2026 1-300

EDI RUS 1-7 0000-0003-0432-4603 Moscow State Institute of International Relations Baykov Andrey Ho Chi Minh city University of Foreign Languages Nguyen Ngoc Vu

Ho Chi Minh City Southeast Region Vietnam

Machine Learning for Learning Machines This special issue explores the profound intersection of machine learning, language, and pedagogy, questioning the fundamental role of artificial intelligence in educational and linguistic coordination. Rooted in the age-old dream of combinatorial devices – echoing historical figures from Ramon Llull to Jorge Luis Borges – the collection investigates whether language and thought can be automated without appeals to meaning and reference. As generative AI becomes ubiquitous, we confront a landscape where machines train humans, write academic prose, and simulate understanding. This issue brings together philosophers of technology, linguists, and educational theorists to analyze this paradigm shift. It spans historical genealogies of educational automation, the cognitive potential of human-AI interfaces, and empirical studies on AI literacy and coaching, alongside critical examinations of AI detection systems and the capacity of machines to truly “think”. Furthermore, a dedicated subsection reflects on Borges’s 1937 essay “Ramon Llull’s Thinking Machine,” using this historical lens to dissect contemporary generative models. Taken together, the collected papers reveal an increasingly recursive relationship between machine learning and human learning, bringing questions of epistemic agency, disciplinary expertise, semantic judgement, and evaluative competence to the center of debates about what it means to learn in an age of generative AI. This issue frames these diverse contributions, charting a course through the promises and perils of the automated word. 10.48417/technolang.2026.03.01 81'322:004.8 Generative AI Philosophy of Technology Language Pedagogy Combinatorics Epistemic Agency Machine Learning https://soctech.spbstu.ru/article/2026.24.1/

RAR RUS 10-31 University of Naples “Federico II” De Stefano Lorenzo

Naples, Italy

Learning Machines: From the Automatic Teacher to Large Language Models A reconstruction is offered of the genealogy of educational automation, personalisation, and epistemic agency. The contemporary debate on large language models in education often treats generative AI as an unprecedented rupture. Yet its promises of personalisation, immediate feedback, and relief from repetitive teaching belong to a longer history. This article reconstructs educational automation from Pressey and Skinner through cybernetics, computer-assisted instruction (CAI), and intelligent tutoring systems (ITS) to large language models. It advances three claims. Educational automation redistributes epistemic labour rather than expressing a linear increase in machine intelligence. Personalisation repeatedly depends on prior standardisation of objectives, knowledge, and evidence of learning. Generative AI introduces a discontinuity by producing the explanatory discourse through which understanding is ordinarily recognised. Bringing these claims together, the genealogy reveals how successful performance can become detached from the formation of judgement. Its critical force lies in examining this tension against educational commitments to developing learners’ capacities, while making the normative priority of epistemic agency explicit. The decisive question concerns how relations among learners, teachers, and machines enable subjects to direct inquiry, evaluate claims, and regulate their dependencies. This criterion remains relevant even if models exhibit limited forms of understanding: the competence of an assisting system does not establish the educational development of its user. 10.48417/technolang.2026.03.02 084:37 Teaching machines Programmed instruction Computer-assisted instruction Adaptive learning Large language models Epistemic agency Personalisation https://soctech.spbstu.ru/article/2026.24.2/ 10-34.pdf

RAR RUS 35-47 57217591576 0000-0002-8825-3760 Peter the Great St. Petersburg Polytechnic University Lisenkova Anastasia

St. Petersburg, Russia

B-2975-2017 56426509300 0000-0001-8953-7434 Department of Social Science, Peter the Great St.Petersburg Polytechnic University Shipunova Olga Cognitive Capabilities of the Interface in Machine Learning: A Philosophical Perspective The actual problems of digital technology in the machine learning field are related to the expansion of autonomous intelligent systems in respect to cognitive activity. The growth in the volume of information in modern networks requires new formats for working with knowledge, based on hybrid systems and neural models of artificial intelligence (AI). Neural network training involves the formation of experience of semantic orientations in multi-valued contexts using self-correction mechanisms. In the modeling of virtual machine learning environments, the role of the interface increases. The article examines the cognitive potential of the interface as a factor shaping the boundaries and modes of functional autonomy of artificial intelligence systems in machine learning. The interface is treated not as a neutral technical channel of information transfer but as a semantic matrix that programs a sequence of cognitive operations – selection, context-binding, inference, revision, and action. Particular attention is given to the temporal dimension of the interface's cognitive potential. The authors introduce the concept of chronotope (Mikhail Bakhtin) as a model for organizing the interplay of space and time within the virtual learning environment, and analyze temporal event models grounded in metaphors, image schemas, and grammatical constructions. The communicative design of the interface is discussed through the lens of cultural archetypes (Carl Gustav Jung), which shape “trusted agent” models in human–AI dialogue. Factors determining the interface's cognitive function – structuring the space of relevant information, temporal dynamics, inference formation, optimization regimes, and actor motivation – are systematized. The authors conclude that the interface's cognitive potential establishes the conditions for the subject's semantic orientation and the digital agent's goal-directed behavior, ensuring the coupling of human and artificial intelligence in the expanded production of knowledge. 10.48417/technolang.2026.03.03 1:81'22:004.8 Interface Machine learning Artificial intelligence Chronotope Semantic matrix Digital epistemic agent Temporal models https://soctech.spbstu.ru/article/2026.24.3/ 35-47.pdf

RAR RUS 48-67 0000-0002-2068-9250 Institute of Business Career Sakhnevich Sergey

Moscow, 109029, Russia

Towards a Hermeneutic AI: A Lemmatic Approach to Nietzsche and Robbins This study investigates the hermeneutic potential of artificial intelligence in lemmatic analysis–a method that reduces word forms to canonical lemmas and, in its extended version, identifies stable semantic patterns that form the conceptual framework of a text. The article tests the hypothesis that AI can perform not only an auxiliary but also a heuristic function, revealing implicit semantic structures inaccessible to traditional methods. Based on two previously manually analysed, rather disparate corpora – Friedrich Nietzsche’s metatexts (personal correspondence, drafts, notes) and Harold Robbins’s popular novels – the study presents a three-level model of AI-assisted lemmatic analysis. Level 1 automates lemmatization and frequency counting; Level 2 identifies stable lemmatic patterns and binary oppositions; Level 3 integrates these results into a hermeneutic interpretation. To validate the model, an experiment compares AI-generated lemmatic analysis (using GPT-4) with manual results. The experiment shows that AI successfully replicates core-periphery dichotomies and detects instrumental functions of thematic lemmas (e.g., money as a means of escape in Robbins), but may miss certain peripheral lemmas (e.g., “wound”) identified by human researchers. The article introduces the concept of “hermeneutic AI” – an algorithm capable of dialogical, traceable, and verifiable interpretation – and proposes five validation criteria: traceability, pattern stability, inter-coder verification, completeness, and heuristic productivity. The findings demonstrate that AI-assisted lemmatic analysis offers scalable, reproducible tools for digital humanities while remaining complementary to human expertise. The study contributes to the methodology of text interpretation and opens new perspectives for the integration of machine learning into literary and philosophical hermeneutics. 10.48417/technolang.2026.03.04 1:81'22:004.8 Hermeneutic AI Lemmatic analysis Friedrich Nietzsche Harold Robbins Digital humanities Text interpretation Experimental verification Implicit semantic structures https://soctech.spbstu.ru/article/2026.24.4/ 48-67.pdf

RAR RUS 68-85 Link David Covert Systems – On the Intelligence of Large Language Models In light of recent developments in Artificial Intelligence, particularly the development of Large Language Models (LLMs), the question if machines can be said to think is examined. The problem was already addressed by Alan Turing in the 1950s, laying the groundwork for our argumentation. Our text refutes four common objections. Firstly, it challenges Ada Lovelace’s claim that computers are incapable of creating truly original content, demonstrating that neural networks, such as Transformer models, independently generate novel structures. Secondly, the assertion that LLMs merely statistically predict the next word is discussed. They are shown to be capable of constructing compressed internal representations from their input data through nonlinear operations, distinguishing them from statistical models. A third argument concerns AI’s lack of sensory access to the world. However, as Turing has demonstrated, the development of mental capabilities requires only communication between a teacher and a student, and sensory experience can be technically replicated at any time. The Chinese Room thought experiment, which suggests that algorithms merely simulate intelligence through formal symbol manipulation, is already anticipated by the Turing Test, which posits that any machine successfully feigning mental functions should be deemed to possess them. The text explores how LLMs generate a perfect compression of their input data, resulting in an accurate model of the world, a convincing simulation of human intelligence, and mastery of language. It discusses the phenomenon of “emergence”, the appearance of higher-order cognitive capabilities with increasing amounts of training data. Although the successes in implementing intelligence are notable, the neural network methodology employed results in the enigmaticity of the underlying basis, failing to illuminate the true nature of the mind as originally hoped. This obscurity was already evident in the foundational works of Kurt Gödel and Alan Turing, highlighting the complexities inherent in understanding human and machine intelligence alike. 10.48417/technolang.2026.03.05 1:004.81 Large Language Models Mechanical Intelligence Emergence Perfect Compression Enigmaticity https://soctech.spbstu.ru/article/2026.24.5/ 68-85.pdf

RAR RUS 87-93 0000-0001-8000-7613 Darmstadt Technical University Hähnle Reiner

Darmstadt, Germany

The Color of LLMs In 1937, Jorge Luis Borges wrote a characteristically engimatic historical note and literary reflection on „Ramon Llull' s Thinking Machine.“ It was the year when the idea of „thinking machines“ picked up momentum with Harold Aiken’s mechanical Mark I early computer that became the electronic Mark IV. 85 years ahead of ChatGPT (first released in 2022), Borges is looking back some 650 years at Ramon Llull‘s great arts which includes an ars inveniendi that inspired rational speculation ever since. This is one five essays that look though the lens of Borges at contemporary learning machines. It explores how Jorge Luis Borges’ account illuminates the nature and limitations of contemporary large language models. The author recalls that Jonathan Swift’s Gulliver’s Travels already describes an engine that mechanically assembles books from broken sentences, and Borges associates it with the combinatorial tradition of thinking machines. This tradition, including Llull’s disks, Raymond Queneau’s cut-up book, and the Infinite Monkey Theorem, relies on generation by chance and recombination. It is contrasted with the algorithmic tradition, represented by the Antikythera mechanism, astronomical clocks, and the calculating machines by Schickard, Pascal, Leibniz, and Babbage, where problems are solved by rules. The author notes that astrology failed because it depended largely on correlation rather than computation, and that modern LLMs are also based on correlations learned from human artifacts. At the same time, LLMs contain a combinatorial element guided by prompting. A parallel is drawn with Oulipo: literary constraints, such as writing a novel without the letter “e,” foster creativity, whereas LLM prompting directs and constrains searches. The analysis of Borges’ color motifs – “red,” “zenithal,” the tiger, Blake – shows that Borges’ text is a subtle critique of simplistic thinking machines and an affirmation of human creativity and intertextuality, which cannot be reduced to mechanical recombination and correlation. 10.48417/technolang.2026.03.06 1:81'22:004.8 Ramon Llull Jorge Luis Borges LLMs Thinking machines Combinatorics Algorithms Prompting Oulipo Creativity Intertextuality Color https://soctech.spbstu.ru/article/2026.24.6/ 87-93.pdf

RAR RUS 94-107 J-9548-2017 57210142445 0000-0002-7956-4647 Department of Social Science, Peter the Great St. Petersburg Polytechnic University Bylieva Daria

St. Petersburg, Russia

Between Determinism and Chance: From Llull's Machine to Generative AI Models In 1937, Jorge Luis Borges looked back some 650 years at Ramon Llull‘s machinic ars inveniendi.  This is one of five essays that looks at contemporary learning machines through the lens of Borges‘s characteristically engimatic historical note and literary reflection. It examines the epistemological tension between determinism and randomness in projects of formalising thought – from Ramon Llull's Ars Magna (1305) to contemporary generative transformers. Borges's 1937 critique of Llull revealed a fundamental impasse: closed combinatorial systems, lacking contextual sensitivity, produce syntactic noise rather than truth. Yet Borges proposed an alternative, namely, to use such machines as generators of random combinations subject to human selection. This shift finds unexpected validation in AI transformer models. By replacing discrete symbols with continuous semantic embeddings and dynamic attention mechanisms, transformers overcome Llull's linear blindness. However, in deterministic mode (low temperature), they merely reproduce linguistic clichés, exhibiting neural text degeneration. Genuine novelty emerges only through stochastic sampling–temperature-based deviation into the probability distribution's long tail, actualising latent semantic projections (as in the Borgesian “red tiger”). Drawing on recent empirical studies, the paper demonstrates a fundamental trade-off between alignment due to reinforcement learning (RLHF) and stochastic creativity. As an alternative, calibrated uncertainty is proposed, allowing models to acknowledge the limits of knowledge. The transformer that legitimises its stochastic nature thereby approaches Borges's ideal of a poetic machine – an instrument that does not prove but suggests, that does not close off truth but opens a space for play. 10.48417/technolang.2026.03.07 7+117 Ramon Llull Jorge Luis Borges Generative AI Transformers Temperature sampling Randomness Creativity Combinatorics Calibrated uncertainty TRIZ https://soctech.spbstu.ru/article/2026.24.7/ 94-107.pdf

RAR RUS 108-117 University of Aarhus Hasse Cathrine

Aarhus, Denmark

The Thinking Machine and the Tiger In 1937, Jorge Luis Borges wrote a characteristically engimatic historical note and literary reflection on „Ramon Llull' s Thinking Machine.“ It was the year when the idea of „thinking machines“ picked up momentum with Harold Aiken’s mechanical Mark I early computer that became the electronic Mark IV. 85 years ahead of ChatGPT (first released in 2022), Borges is looking back some 650 years at Ramon Llull‘s great arts which includes an ars inveniendi and inspired rational speculation ever since. This is one of five essays that look through the lens of Borges at contemporary learning machines. On one level, it traces the lineage of combinatorial textual generation from Ramon Llull’s late-thirteenth-century Ars Magna through André Breton’s surrealist automatic writing and Ray Kurzweil’s early Cybernetic Poet to contemporary large language models. Relating Jorge Luis Borges’s literary critique of thinking machines to Lev Vygotsky’s cultural-historical psychology, the essay suggests how both human imagination and generative algorithms rely fundamentally on the recombination of culturally available materials. At the structural level of output, the text-generation mechanics of a human poet and a machine can appear functionally identical, yet contemporary generative AI achieves a highly selective combinatorial excess that obscures the underlying absence of an internal semantic architecture. Drawing on Vygotsky’s theory of word meaning as an evolving process forged through prior learning, historical embodiment, and social practice, this essay argues that modern machines dissociate the production of contextually appropriate, meaningful language from the socio-affective history of meaningful thinking. This finally leads to the conclusion that it is becoming impossible to infer authentic meaningful thinking solely from the presence of meaningful language, suggesting that meaning should be viewed not as an isolated asset within the human or the machine, but rather as an emergent property generated within the relational spaces where machine-generated cultural resources encounter human histories of learning. 10.48417/technolang.2026.03.08 1:81'22:004.8 Jorge Luis Borges Ramon Llull Thinking machines Generative AI Large language models Recombination Word meaning Lev Vygotsky Creativity Intertextuality Automatic writing Cybernetic Poet https://soctech.spbstu.ru/article/2026.24.8/ 108-117.pdf

RAR RUS 118-124 0000-0002-0702-0253 Moscow State Institute of International Relations Samorodova Ekaterina

Moscow, Russia

Ramon Llull's Thinking Machine: Between Uselessness and Poetry In 1937, Jorge Luis Borges wrote a characteristically engimatic historical note and literary reflection on „Ramon Llull' s Thinking Machine.“ It was a time in history when the idea of thinking machines to support human learning became ever more prominent. The latest generation of these machines was conceived in laboratories all over the world, their development still continuing today. At that time, Borges is looking back some 650 years at Ramon Llull‘s great arts which include an ars inveniendi and inspired rational speculation ever since. This is one of five essays that looks through the lens of Borges at contemporary learning machines. It focuses on Llull‘s idea of mechanical combinatorics as an instrument of cognition, and Borges’s paradoxical conclusion that an unsound method turns out to be a productive tool for creativity and especially a tool for forging poetic metaphors. The essay considers parallels to modern technologies of artificial intelligence, neural networks, and generative creativity, and also reflects on the role of humans and machines in the process of thinking and the nature of meaning. In the gap between design and realization and from technical failure Borges shows that creativity cannot arise out of emptiness; it is not creation ex nihilo, it is choice from a multitude of possibilities. 10.48417/technolang.2026.03.09 1:81'22:004.8 Ramon Llull Jorge Luis Borges Thinking machine Combinatorics Artificial intelligence Formalization of thinking Poetry Randomness Creativity https://soctech.spbstu.ru/article/2026.24.9/ 118-124.pdf

RAR RUS 125-139 Leuphana Universität Lüneburg Xylander Cheryce von

Universitätsallee 1, 21335 Lüneburg, Germany

Llullaby This is one of five essays that extend, to the current moment of digital infrastructure transformation, musings on synthetic intellect generation put forth by Jorge Luis Borges in one of his lesser-known essays from 1937. “Ramon Llull’s Thinking Machine” showcases an instrument for analytical enhancement, devised in the 13th century, to settle theological conflicts between the three monotheistic world religions (Judaism, Christianity, Islam). Borges observes that delegating intellectual agency to automated routines fails to fortify reason per se – but succeeds in other respects, namely in quickening concept proliferation and lateral associations. Borges treats Llull’s “thinking machine” as an object lesson that conveys a core historiographical insight concerning the prosthetics of mind in general: New idea-generating technologies entail unintended cogito-cultural ramifications. This essay sees itself as an addendum to Borges’ commentary. With reactionary politics once again on the rise, it recalls two thinking machines of a more recent vintage along with their respective cultural reverberations – the Chess Player (a late 18th century automaton), for one, and the Portable Bettmann Archive (a mid-20th century image search tool), for another. The commercial picture library Otto Bettmann founded in 1936, the year before the Borges essay appeared, articulated a critique of totalizing thought like the critique that Borges extracted from Llull. By the 1960s, Bettmann’s pictorial logic had entered mainstream US-advertising culture from whence its distinctive visual grammar continued to spread. To this day, the Bettmann/Borges vision of combinatorics informs the functional layout of the graphical user interface. This idiom of search and retrieval, first consolidated in the 1930s, has come to permeate online signposting and intuitive wayfinding. A few sample plates from the Bettmann Portable Archive conclude this essay-update of Borges’ remarks on Llull’s thinking machine. They are tendered as an invitation to follow a visual argument for the broader claims here advanced. 10.48417/technolang.2026.03.10 1:004.8 Thinking Machine Ramon Llull Jorge Luis Borges Kantian AI Bettmann Portable Archive Totalizing Knowledge https://soctech.spbstu.ru/article/2026.24.10/ 125-139.pdf

RAR RUS 140-155 0000-0003-4659-314X Russian State Agrarian University – Moscow Timiryazev Agricultural Academy Komanova Alla

Moscow, Russia

0000-0001-8449-544X Russian State Agrarian University – Moscow Timiryazev Agricultural Academy Vigna-Taglianti Jacopo

Moscow, Russia

Chromatic Representation of Ontological Attributes: Colour-Based Phraseology in the Light of Ramon Llull’s Ars Magna In 1937, Jorge Luis Borges looked back some 650 years at Ramon Llull‘s great arts which includes an ars inveniendi that inspired rational speculation ever since. Along with five essays that look though the lens of Borges at contemporary learning machines, this research paper adapts Llull‘s approach to offer a linguosemiotic analysis of English colour-based phraseological units. The aim of the work is to identify the mechanisms by which sacral chromatic codes are transformed into the pragmatic meanings of English idioms, and to establish typological patterns in the transition from the metaphysical attributes of the Absolute to the empirical linguistic picture of the world. The object of the study is the chromatic encoding of nine attributes of the Absolute, represented in the form of a mandala-like diagram with the central symbol ‘A’. The scientific novelty of the work lies in the fact that English colour phraseology is examined for the first time not in isolation, but through the lens of a specific combinatorial-logical model (Llull’s circles), which allows us to reveal the deep mechanisms underlying the transfer of transcendent categories into the pragmatics of everyday language. We thereby show that the English language, refracting the ideal Llullian spectrum through the prism of utilitarian pragmatism and anthropocentric experience, creates a semantically contradictory picture of the world in which transcendent attributes undergo reduction, inversion, or functionalisation. 10.48417/technolang.2026.03.11 81’373.7 + 81’22 Colour semantics Cognitive linguistics English idioms Ontological model Semiotics Ramon Llull’s logical machine Combinatorics https://soctech.spbstu.ru/article/2026.24.11/ 140-155.pdf

RAR RUS 157-183 0000-0002-3340-3880 Ufa University of Science and Technology Shaykhulova Aygul

Ufa, Russia

0000-0002-8616-0042 Ufa University of Science and Technology Medvedev Andrey

Ufa, Russia

The False Positive Paradox: AI Detection Systems and the Erosion of Human Academic Authorship The widespread adoption of AI text detectors in academic settings since late 2022 has created an unrecognized problem: rigorous academic writing is increasingly penalized as “AI-generated.” Universities and journals have rapidly deployed commercial detection tools such as Turnitin, GPTZero, and Originality.ai despite limited evidence of their reliability regarding scholarly texts. This study examines the empirical reliability of AI text detectors on academic writing and investigates the epistemological and ethical implications of delegating authorship verification to algorithms. We conducted a systematic review of peer-reviewed studies published between 2022 and 2025 that met strict inclusion criteria, namely quantitative accuracy metrics on academic corpora, peer-reviewed publication, and reproducible methodology. Data were synthesized from 21 verified sources including controlled experiments, meta-analyses, and detection limit studies, while philosophical analysis drew on creativity theory, philosophy of science, and the Chinese Room argument. On academic texts, false positive rates vary substantially across detectors (range: 2–41%, mean: 14–22%), representing a three- to sixfold increase compared to news corpora, whereas false negative rates reach 44–47% with minimal human editing and exceed 60% under combined obfuscation. Inter-detector agreement is weak (Cohen's κ = 0.35–0.44), which makes single-detector decisions unreliable, and no systematic cross-linguistic studies meeting inclusion criteria exist, representing a critical research gap. Theoretically, large language models can produce combinatorial and limited exploratory novelty but cannot generate transformational novelty, that is, the capacity for paradigmatic rupture that defines human authorship. Current AI detection systems are therefore epistemologically flawed and technically unreliable for high-stakes academic decisions. We propose a three-level human-centered framework (screening – expert review – appeal) that restores human agency, since academic integrity requires trust, transparent procedures, and human judgment rather than algorithmic suspicion. 004.8:37 10.48417/technolang.2026.03.12 AI text detection False positives Academic writing Epistemological boundaries of AI Human authorship Algorithmic stylistic discrimination https://soctech.spbstu.ru/article/2026.24.12/ 157-183.pdf

RAR RUS 184-206 0009-0009-8231-8904 Masaryk University Polednikova Tereza

Brno, Czechia

0009-0002-2379-8062 Masaryk University Ricicova Monika

Brno, Czechia

0000-0002-3432-5298 Masaryk University Pilch Pavel

Brno, Czechia

Artificial Intelligence Literacy in Slavonic Studies: Student Reflections Artificial intelligence literacy is usually assessed through self-report scales. Such instruments show where a group stands, but not how its members reason about the technology in their own terms. Research on this topic has also concentrated on science and technology disciplines and on high-resource languages. It remains unclear how students working in low-resource languages, such as the Slavonic languages, articulate their artificial intelligence literacy, and what knowledge underlies their evaluative judgments. This study analyses written reflections and a background questionnaire from students of Slavonic studies, translation and language teaching who completed a one-semester course in artificial intelligence literacy. The reflections were produced after students used a generative artificial intelligence tool as both the commissioner and the collaborator of an individual academic project. They were examined through qualitative content analysis combining deductive and inductive categories. The findings show that the reflections centre overwhelmingly on evaluating outputs, while declarative knowledge of how such systems function remains a knowledge of symptoms rather than of mechanisms. Evaluative judgments depend heavily on disciplinary expertise in areas such as phonetics, lexicology, translation and language pedagogy, rather than on general critical thinking, and target precisely the errors a domain-uninformed reader would not detect. The questionnaire further shows that students already held a critical disposition toward artificial intelligence before the course began, which constrains how the reflective practice documented here can be interpreted. Taken together, the results reposition disciplinary expertise, more than technical understanding of artificial intelligence, as the primary resource for critical evaluation, with direct implications for teaching artificial intelligence literacy and language technology in higher education. 10.48417/technolang.2026.03.13 378.147.016: 004.8 Artificial intelligence literacy Slavonic studies Low-resource languages Disciplinary expertise Reflective practice Language technology Higher education https://soctech.spbstu.ru/article/2026.24.13/ 184-206.pdf

RAR RUS 207-232 0009-0008-5604-165X Ho Chi Minh University of Banking Huynh Minh Vinh Hien

Saigon Ward, HCMC, Vietnam

Ho Chi Minh University of Banking Huynh Minh Vinh Hien

Saigon Ward, HCMC, Vietnam

AI Translation between Interpretation and Hallucination – Evidence for the Evaluative Competence of Vietnamese Students Generative AI complicates translation evaluation because fluent outputs may introduce unsupported meaning, while non-literal renderings may represent legitimate interpretation. This study examined Vietnamese EFL students’ evaluation of AI-generated translations and its associations with digital competence, translation competence, and critical-evaluation confidence. A cross-sectional study involved 213 final-year English majors who completed a 25-item performance task and self-report scales. Students were considerably more accurate in recognising problematic outputs than in diagnosing their type (68.0% vs. 38.4%). Critical-evaluation confidence showed the strongest association with performance (r=.85), although its conceptual overlap with the performance task warrants caution. In the full regression model, translation competence (β=.31) and digital competence (β=.16) contributed significantly alongside critical-evaluation confidence (β=.64), whereas AI-use frequency did not. When the overlapping critical-evaluation measure was removed, the model still explained 55% of performance variance, with translation competence showing the strongest association (β=.61). The findings indicate that evaluating AI-generated translation depends less on frequency of AI use than on students’ translation knowledge and evaluative judgement, particularly their ability to distinguish unsupported generation from defensible interpretation. Rather than simply as a translation tool Generative AI should thus  be used as an object of critical inquiry. Alternative AI renderings may expose students to interpretations they had not considered, including potentially productive expansions of source meaning. This approach preserves the interpretive character of translation while making students accountable for the decisions they ultimately accept. 10.48417/technolang.2026.03.14 004.8:378.016 AI-assisted translation; AI hallucination detection; Translation evaluation; Translation competence; Critical evaluation; EFL students https://soctech.spbstu.ru/article/2026.24.14/ 207-232.pdf

RAR RUS 234-262 0000-0002-5889-4292 American University of the Middle East AlAfnan Mohammad Awad

Egaila, Kuwait

Revealing Cultural and Linguistic Variation in Generative AI–Mediated Language Learning: A Mixed-Methods Study The rapid expansion of generative artificial intelligence (AI) in language education has created new possibilities for personalized and adaptive learning, while raising questions about how learners from different linguistic and educational backgrounds interact with AI systems. This exploratory mixed-methods study examines culturally interpretable patterns of AI-mediated language learning among 100 adult English language learners from Japan, China, Saudi Arabia, Egypt, North Macedonia, and Bulgaria across four CEFR levels. Data were collected over eight weeks through interaction logs, learner diaries, pre- and post-assessments, surveys, and semi-structured interviews. Quantitative indicators are used descriptively, while qualitative evidence provides the primary basis for interpretation. The findings suggest broadly positive language-development trajectories alongside substantial variation in how learners request, interpret, evaluate, and use AI-generated feedback. Four overlapping feedback-utilization patterns, calibration, negotiation, experimentation, and uptake, were identified, although their realization varied across learners and contexts. Narrative cases further indicate that interaction can develop from correction-focused use toward clarification, comparison, dialogue, and strategic engagement, while substantial within-group variation highlights the importance of individual agency. Because all participants interacted with the same AI system under uniform conditions, the observed differences reflect learner interpretation and adaptive use rather than system-level cultural customization. The study therefore treats generative AI as a culturally interpreted rather than inherently culturally responsive system and advances three design-oriented hypotheses concerning adaptive feedback calibration, interaction-style modulation, and learner-controlled customization. These hypotheses provide an exploratory foundation for future research examining whether culturally responsive learning agents can support more inclusive and meaningful AI-mediated language learning. 10.48417/technolang.2025.03.15 81'243:004.89 Generative artificial intelligence Language learning Cultural and linguistic diversity AI-mediated interaction Personalized learning systems https://soctech.spbstu.ru/article/2026.24.15/ 234-262.pdf

RAR RUS 263-284 The University of Da Nang Nguyen Quoc Khanh

Hai Chau Da Nang, South Central Coast Vietnam

Artificial Intelligence Coaching for Confident Speaking and Spontaneous Argumentation Artificial intelligence is increasingly used to support spoken language practice. However, different AI-assisted support systems do not support learners’ confidence equally. This explanatory sequential mixed-methods study examines associations among the perceived usefulness of AI feedback, AI-mediated argumentative scaffolding, self-reported cognitive engagement, and speaking self-efficacy in the context of spontaneous argumentation. The quantitative phase involved 130 university EFL learners who had used an AI-assisted argumentation coaching application for at least eight weeks. Regression analysis showed that perceived usefulness of AI feedback was positively associated with so-called speaking self-efficacy, that is, the confidence to perform spoken tasks successfully (β = .410, p < .001), as was AI-mediated argumentative scaffolding (β = .248, p = .002). In contrast, self-reported cognitive engagement was not significantly associated with speaking self-efficacy after the other predictors were taken into account (β = −.028, p = .725). The qualitative phase involved eight purposively selected participants representing variation in gender, year of study, IELTS target, and experiences relevant to the quantitative patterns. Thematic analysis of semi-structured interviews identified three explanatory themes: structural scaffolding, psychological safety, and cognitive limits. Participants frequently described organizational guidance and a nonjudgmental practice environment as useful, while some also reported repetition, fatigue, and dependence during extended AI use. Integrating the two phases suggests that perceived usefulness, scaffolding, and cognitive engagement should not be treated as interchangeable dimensions of AI-assisted speaking practice. The study proposes a provisional cognitive-threshold hypothesis for future testing, but does not claim a causal or nonlinear effect. The findings support strategically designed AI-assisted speaking practice that maintains opportunities for independent reasoning and human interaction. 10.48417/technolang.2026.03.16 81'243:004.89 Artificial intelligence Speaking self-efficacy Spontaneous argumentation Cognitive engagement Instructional scaffolding English as a foreign language IELTS https://soctech.spbstu.ru/article/2026.24.16/ 263-284.pdf

BRV RUS 286-294 0000-0001-9002-5604 Institute of Philosophy and Law of the Siberian Branch of the Russian Academy of Science Chistanov Marat

Novosibirsk, Russia

0000-0003-1347-7302 Novosibirsk State Technical University Chistanova Svetlana

Novosibirsk, Russia

Technology as Creativity: A Review of Anna Demina and Alexander Nesterov‘s Semiotic Theory of Creativity The philosophy of technology has long suffered from a consequentialist bias, focusing on the effects and risks of technology rather than on its essence. Technical creativity, in particular, remains underexplored compared to artistic creativity, often being reduced to applied science or accidental invention. This review examines the monograph Semiotic Theory of Creativity: The Problem of the New in Technology and Art by Anna Demina and Alexander Nesterov which attempts to construct a universal theoretical model of creativity applicable to both technical and artistic domains. The authors of the monograph synthesize the three-act theory of technical creativity by Pyotr Engelmeyer, the concept of form-giving forces by Friedrich Dessauer, and the ontological model of general semiotics (Frege, Peirce, Morris). They correlate the three dimensions of semiosis (pragmatics, syntactics, semantics) with the three levels of cognition (sensory perception, understanding, reason) and introduce a fourth dimension related to the material expression of signs. The reviewed book offers a systematic analysis of the epistemological problems of creativity. It defines creativity as a process of novelty generation through the transformation of pragmatic, semantic, or syntactic rules of semiosis, resulting in new objects at the level of sensory perception, new things at the level of understanding, and new ideas at the level of reason. The authors identify nine possible ways of novelty emergence in technical and artistic creativity, using science fiction as illustrative material. The monograph represents a significant contribution to the philosophy of technology, breaking with both existentialist alarmism and Marxist reductionism. While the choice of science fiction as empirical material may be debatable, and some parts of the text serve a ritual rather than analytical function, the work provides a much-needed, operationalizable framework for understanding technical creativity. It is a substantial, necessary, and important publication that opens new directions for research. 10.48417/technolang.2026.03.17 130.2:62 Technical creativity Semiotics of technology Problem of novelty Philosophy of technology Semiotics of art https://soctech.spbstu.ru/article/2026.24.17/ 286-294.pdf