Teaching English for Specific Purposes in the Age of GenAI: Between Efficiency and Ethics 

The Efficiency Promise of GenAI in ESP 

The integration of Generative Artificial Intelligence (GenAI) into English for Specific Purposes (ESP) classrooms offers remarkable efficiency. Tasks that once required hours – drafting business correspondence, translating legal clauses, or composing technical reports – can now be completed in minutes. Students can generate multiple versions of a professional email or compare translation alternatives instantly. 

However, this efficiency raises a critical question: What is lost when linguistic production becomes automated? This article argues that while GenAI provides valuable support for ESP instruction, uncritical adoption risks undermining the very competencies ESP aims to develop: context-sensitive judgment, cultural pragmatics, and ethical responsibility. A balanced approach requires curriculum redesign and the cultivation of critical digital literacy. 

Hidden Costs of Automated Proficiency 

ESP differs from general English in its emphasis on professional genres and domain-specific terminology. In legal translation, the difference between shall and must carries distinct legal obligations. In business correspondence, direct versus indirect phrasing reflects relational hierarchy and cultural norms. GenAI models, trained on large but often decontextualized corpora, frequently flatten these nuances. 

Research shows that students tend to over-trust AI-generated output, assuming grammatical accuracy guarantees pragmatic appropriateness (Kohnke et al., 2023). In Indonesian ESP classrooms, many students lack the metalinguistic awareness to evaluate whether an AI-generated legal clause is enforceable or whether a business email appropriately reflects local politeness norms. Thus, efficiency without critical oversight produces an illusion of competence. 

Redesigning ESP Pedagogy for the AI Era 

Rather than banning AI – an unrealistic response – ESP educators must redesign learning tasks to position students as critical users of technology. Three strategies have proven effective in business correspondence and legal translation courses. 

First, process-oriented documentation. Students submit not only the final product but also a log of their AI prompts, revisions, and justifications. For example: “I asked ChatGPT to draft a reminder email. I changed ‘You haven’t paid’ to ‘We kindly remind you’ because the original was too direct for Indonesian business culture.” This makes decision-making visible and fosters metacognitive awareness (Hapsari, 2020). 

Second, comparative error analysis. Students produce two versions of a translation or business letter: one generated solely by AI, and one produced collaboratively with AI assistance. They then write a short reflection comparing accuracy, register, and potential professional risks. This sharpens their ability to identify AI hallucinations and cultural blind spots. 

Third, unplugged proficiency checks. Periodic in-class writing without AI access recalibrates students’ self-assessment. Many discover gaps in vocabulary or grammar that AI had been silently filling, prompting more focused study. 

Ethical Guidelines for ESP Programs 

Beyond classroom strategies, ESP programs need institutional guidelines. Three principles are recommended. 

The Transparency Principle: Students must declare any use of AI in submitted work, specifying the tool and purpose (e.g., brainstorming, grammar checking, translation). 

The Competence Principle: Any assessment measuring core professional competence — such as drafting a binding legal clause — should include an unassisted component to ensure independent performance. 

The Critical Literacy Principle: Every ESP curriculum should include explicit instruction on AI limitations: bias, hallucinations, cultural assumptions, and data privacy. This aligns with the essential goals of fostering critical thinking and student autonomy in ESP classrooms. 

Conclusion 

GenAI is not a threat to ESP teaching, but neither is it a neutral tool. Its integration into Indonesian ELT requires deliberate pedagogical design that prioritizes ethical judgment alongside efficiency. When students learn to question AI output, compare alternatives, and articulate their linguistic choices, they develop precisely the higher-order skills that ESP promises: professional competence, cultural mediation, and communicative responsibility. 

The future of ESP lies not in choosing between human instruction and machine assistance, but in teaching students to orchestrate both. As members of APSPBI, we have the opportunity to lead this transformation — by embedding technology within a framework of critical, ethical, and context-sensitive pedagogy. 

Reference 

Hapsari, A. D. (2020). Metacognitive strategy training in the teaching of reading comprehension: Is it effective in EFL classroom? Language-Edu, 9(1). 

Kohnke, L., Moorhouse, B. L., & Zou, D. (2023). ChatGPT for language teaching and learning. RELC Journal, 54(2), 537–550.