Abstract
Purpose
To explore the interplay between human translators and AI tools, focusing on tertiary students' perceptions in the context of Portuguese-Chinese translations in Macao.
Design/methodology/approach
This research employed a mixed-methods approach. Quantitative surveys were complemented by qualitative responses. Qualitative class observations (participant and non-participant) and autoethnography further enriched the insights. Participants included undergraduate and postgraduate students in translation studies from the Macao Polytechnic University.
Findings
The data revealed a dual perspective: appreciation for AI’s efficiency contrasted with concerns about its potential to overshadow human touch in translations, especially in cultural nuances. Views on integrating AI into curricula were diverse, but a balanced, synergistic approach between human expertise and AI efficiency emerged as a common theme.
Originality/value
This study offers a fresh perspective by integrating various methodologies, capturing both statistical and experiential insights on the evolving relationship between AI and human translation efforts in academia.
Keywords
Citation
Amaro, V. and João Pires, M. (2024), "Found in translation, lost in education: artificial intelligence’s impacts on translation tertiary education in Macao", Asian Education and Development Studies, Vol. 13 No. 4, pp. 269-281. https://doi.org/10.1108/AEDS-01-2024-0012
Publisher
:Emerald Publishing Limited
Copyright © 2024, Emerald Publishing Limited
Introduction
The rapid evolution of artificial intelligence (AI) in various sectors, including communication, is undeniable. Particularly, the field of translation has witnessed transformative changes with the rise of sophisticated AI models, such as Large Language Models, exemplified by systems like ChatGPT. Though not originally designed for translation, these models have begun to compete with, or even surpass, established translation platforms like Google Translate and DeepL. Such shifts not only redefine the technological landscape but also hold significant implications for the broader academic and professional realms, potentially altering the global dynamic of translation and communication.
This surge in AI’s influence on translation presents both an opportunity and a challenge. As AI tools become more integrated into academic settings, there’s a pressing need to understand their impact on students' learning experiences. How are these tools affecting students' depth of understanding, analytical abilities, and originality? Do they enhance genuine learning, or do they inadvertently promote a superficial approach where expediency trumps comprehension? Addressing these questions is crucial, especially as educators strive to strike a balance between leveraging cutting-edge technology and maintaining the integrity of the educational experience.
This paper contributes a novel empirical study to a broader research endeavor assessing the impact of AI tools on the learning outcomes of translation students at Macao Polytechnic University. This institution stands as a leading authority in Chinese-Portuguese translation studies in Greater China. Utilizing a mixed-methods approach, this research aims to evaluate the repercussions of AI on students' academic journeys within the translation programs. The combination of quantitative data and qualitative insights is strategically chosen to yield a multifaceted understanding - quantitative findings establish the magnitude, while qualitative data bring forth the nuance and intricacy.
Drawing on their considerable experience as language educators, the authors have identified a significant trend: students are increasingly turning to AI tools for assistance in language learning and translation. This reliance reflects students' ability to adapt and their confidence in contemporary technological solutions. However, it also presents potential challenges, particularly if this dependence undermines the development of critical thinking and the capacity for independent, original analysis.
In exploring the dynamic interaction between emerging technology and educational strategies, this study endeavors to provide more than an analysis of the status quo; it aims to offer constructive recommendations that may inform curriculum development, teaching methodologies, and the future design of AI applications in translation education. The findings of this study hold the potential to extend their relevance beyond Macao’s classrooms, presenting valuable insights for educators, technology developers, and policy makers worldwide.
Review of literature
As instructors of Portuguese, teaching students for whom Chinese is the first language, we have employed diverse platforms for interaction, ranging from informal classroom dialogues to structured surveys and focus groups. These engagements have enriched our understanding of student perspectives on Machine Translation (MT) within the broader context of learning Portuguese as a second language (L2), and its application in Chinese-Portuguese and Portuguese-Chinese translation and interpretation. Our observations suggest that students rely on MT for a gamut of purposes, including vocabulary retrieval, grammatical validation, and pronunciation practice. A marked increase in the use of MT for drafting complete paragraphs has been noted, revealing a disparity between their conversational skills and the polished prose of their written tasks, suggesting a substantial reliance on MT.
Students readily engage with MT for translating sentences, valuing the tool for enhancing their understanding of Portuguese texts. The emergence of advanced tools like ChatGPT has been linked to an improvement in the linguistic precision of their assignments, now replete with complex structures not yet introduced in coursework. However, such technological advancements have often obscured the lack of depth in content, presenting a conundrum: while MT and AI tools can polish linguistic form, they do not necessarily translate to a deeper language comprehension.
Klekovkina and Denié-Higney (2022) articulate that educators have traditionally associated grammatical precision with language proficiency. Correct sentence structure and verb conjugation were seen as indicators of solid language grasp, and from the student’s vantage point, such accuracy often correlates with academic achievement. The advent of MT, however, calls for a reassessment of this correlation. Linguistic perfection achieved via MT does not definitively signify a student’s true understanding or capability in L2, necessitating a more nuanced evaluation of language mastery beyond grammatical accuracy.
Machine Translation, since its inception in the 1950s, has undergone a significant evolution, transitioning from rule-based systems to statistical methods and, most recently, neural machine translation (NMT) enabled by deep learning. This shift has greatly enhanced MT’s quality and necessitated a reevaluation of the translator’s role and the integration of AI in language education (Melby, 2020). Despite the documented apprehension toward MT among professional translators, such resistance tends to diminish when they directly participate in the development and implementation of MT systems (Rossi and Chevrot, 2019).
The incorporation of AI into language and translation education represents a complex narrative, shaped by divergent opinions within the academic community. Tools like Google Translate and AI platforms such as ChatGPT have incited debate, with some educators concerned about their impact on academic integrity and the potential obsolescence of traditional language instruction (Crossley, 2018; Hellmich and Vinall, 2021; Vinall and Hellmich, 2022; Klekovkina and Denié-Higney, 2022). However, research also illuminates the widespread adoption of AI tools among L2 learners, driven by their convenience and the immediate, cost-free assistance they provide. Alharbi (2023) underscores how the integration of AI in the writing process can significantly enhance the quality of student compositions, especially when paired with guided instruction (Zhang and Torres-Hostench, 2022).
Despite these positive developments, there is a growing concern that an over-reliance on AI might foster a simplistic view of language learning, sidelining the complexity and nuance intrinsic to human languages (Hellmich and Vinall, 2021). Yet, when AI is employed to foster engagement with a student’s native language, it can offer substantial pedagogical benefits. Such a strategy aligns with the multilingual turn in SLA research, encouraging a comprehensive exploration of lexicogrammatical structures and guiding learners towards a more usage-based language model (Crossley, 2018; Ellis, 2017; Klekovkina and Denié-Higney, 2022; Ortega, 2017; Tomasello, 2003).
As AI continues to permeate translation methods, it redefines the educational landscape, despite gaps in research on tertiary translation students' interaction with AI. Scholars like Deng and Yu (2022) emphasize MT’s pedagogical merits but also its shortcomings in capturing idiomatic expressions and contextual nuances. This calls for enhanced tools and holistic teaching methods (Shen et al., 2022). Cheng (2022) sees potential in merging traditional techniques with AI and virtual reality to improve learning outcomes. In contrast, dated pedagogical practices are critiqued for their inability to match the pace of AI innovation, pointing to misalignments in current academic frameworks and industry needs (He, 2021; Sun, 2023).
Considering AI’s broad educational impact, literature documents its diverse integrative roles, enhancing both administrative and pedagogical processes. However, challenges such as the risk of misinformation remain, highlighting the necessity for critical appraisal and the preservation of human interaction in education (Baidoo-Anu and Ansah, 2023; Chassignol et al., 2018; Chen et al., 2020).
In professional translation, a collaborative future between translators and AI suggests an evolution toward roles such as “translation + language engineers.” Such a partnership underscores the irreplaceable value of human expertise, positioning AI as an adjunct rather than a substitute (Ai, 2022; Lee, 2023; Yang, 2023). As education evolves, the balance between AI’s capabilities and the indispensable human intuition and expertise is critical, affirming AI’s role as an aid in the nuanced field of translation.
Methodology
This study adopts a quantitative-qualitative (Peterson, 2000; Saldaña, 2013) approach to examine the experiences of students enrolled in translation and interpretation programs at the Macao Polytechnic University. We targeted students from the bachelor’s degree of translation and interpretation in Chinese-Portuguese and Portuguese-Chinese, the master’s degree in Translation and Interpretation Chinese-Portuguese, and the doctoral programme of Portuguese Studies. The total population of these programmes is approximately 280 students.
The development of the questionnaire was guided by student interactions and qualitative class explorations, pinpointing essential areas of concern and interest. This method ensured its relevance and comprehensiveness. The questionnaire was validated through an initial distribution to a select sample of eight students, whose feedback informed subsequent refinements, thereby improving its reliability. Comprising 17 questions in Chinese, Portuguese, and English, the questionnaire featured closed-ended, semi-closed, and filter questions. It permitted open-ended responses in any of the three languages, enhancing flexibility. To capture the spectrum of student perspectives, a 10-point scale was employed for certain questions, enabling detailed expression of agreement or disagreement. Though this scale shares attributes with both Likert and semantic differential scales, it was chosen for its broader range and finer nuance in capturing attitudes. It is referred to as a modified Likert scale in our study for its simplicity and respondent accessibility.
The design of the questionnaire emphasized straightforwardness and user-friendliness. Additionally, it was crafted to be completed between three to five minutes, aiming to encourage and optimize student participation. Data collection was executed using Google Forms, leveraging its advantages in terms of speed, accessibility, and efficiency over conventional methods. The questionnaire was sent to the entire population of 280 students enrolled in the translation programs, encompassing both undergraduate and postgraduate levels. To ensure a broad and representative sample, we enlisted the help of teachers, who facilitated the collection of responses during their lecture times. Additionally, we sought the support of class representatives, who played a crucial role in motivating their peers to participate in the survey. This concerted effort ensured the survey reached students in various settings, both in-class and outside, effectively enhancing participation rates. The survey was administered in October 2023, targeting students across all academic years of their respective degrees. With the combined efforts of class representatives and academic staff in disseminating the survey, we engaged a significant portion of the student body in the process. Out of the total student populace of 280, 150 valid responses were acquired. Given this response rate, our findings can be interpreted with a confidence level of over 90%, considering a margin of error of 5%.
In addition to the questionnaire, this study also incorporated participant and non-participant observations conducted during Portuguese language classes. This added a qualitative dimension to our research, providing deeper insights into the classroom dynamics and student interactions. Furthermore, this research is enriched by an autoethnographic perspective (Ellis and Bochner, 2000), drawing upon the authors' experiences within academia and Chinese society over the years. This introspective lens offers a unique understanding of the cultural nuances influencing the translation and interpretation process, adding depth to our quantitative findings.
Ethical considerations were paramount. We ensured that data collection adhered to the ethical standards set by the Macao Polytechnic University and obtained explicit permissions from participants, ensuring their anonymity throughout.
Quantitative data underwent statistical analysis employing software tools for descriptive and inferential statistics. This phase aimed to unearth patterns, trends, and correlations within the dataset, focusing on students' engagement and reliance on AI translation tools. Simultaneously, qualitative data extracted from open-ended questionnaire responses, classroom observations, and the autoethnographic perspectives of the authors were coded and thematically analyzed. This qualitative exploration delved into the subtleties of student experiences, perceptions of AI’s utility, and the cultural and academic implications of AI integration in translation studies. Through this dual-pronged analytical strategy, the study achieved a layered understanding of the prevailing dynamics at the intersection of AI technology and translation education.
Findings
Respondents’ profile
As shown in Table 1, among the 150 survey participants, age ranged from 18 to 43 (average = 30.5, SD = 12.5), and the age groups that stood out were 21 years old at 28.7%, 20 years old at 15.3%, and 22 years old at 8%. Regarding gender distribution, females represented 72.0%, males 24%, with a small segment choosing not to specify their gender. A vast majority of 86% of the students were from Macao, while 14% originated from Mainland China.
The data also highlighted a dominant presence of students in their bachelor’s degree, making up a noteworthy 83.3% of the respondents. The fourth year of the bachelor’s program alone accounted for 41.3% of this figure, emphasizing the significant number of undergraduates in their final year. This was followed by third-year students at 22%, first-year students at 11.3%, and second-year students at 8.7%. Students in their Master’s degree made up a smaller 6%, split between first-year students at 4.0% and second-year students at 2.0%. Meanwhile, PhD students comprised 10.8% of the total, with a 7.4% majority in their third year, while the first and second years registered at 1.4 and 2.0%, respectively.
Utilization and dependence on AI translation tools
This questionnaire section contains six questions designed to gauge the frequency and extent of students' reliance on AI translation tools. Of those surveyed, 44.7% indicated they “often” employ AI tools for academic assignments. Another 29.3% reported “occasional” use, while 18.7% claimed they “always” use these tools. Remarkably, only one respondent (0.7%) mentioned they “never” utilize such tools. The primary platforms preferred by students were Google Translate at 84.5% and DeepL with 81.1%. Baidu Translate was also noted, though at a much lower usage rate of 14.9%. ChatGPT, not originally listed as an option in the survey, was mentioned by 8.2% of participants in the “other” category (see Table 2).
The majority of students (53.3%) either “fully rely” (8.0%) or “partially rely” (45.3%) on AI translation tools to comprehend complex texts. Conversely, 45.3% indicated they use these tools “only as a reference.” A mere five respondents (3.4%) expressed they “rarely rely” on them. When questioned about confidence in the translation’s accuracy, 43% felt “somewhat confident,” whereas a significant 40.5% remained “neutral.”
Most students “always” verify translated results (43.2%), with an additional 21.6% saying they “occasionally” review translations. Regarding assignments intended for grading, 31.8% “often” rely on AI translation tools. Meanwhile, 41.2% turn to these tools “occasionally” for assessed tasks, and 8.1% “always” make use of them (see Table 3).
Perspectives and experiences
This section of the questionnaire includes four close-ended questions and one open-ended question. It delves into the perceived benefits of AI tools, participants' opinions on errors and misinterpretations in translations, and the impact of these tools on student learning and skill enhancement.
When considering the main benefits of using AI translation tools, the majority highlighted “Assistance with unfamiliar terms” as a significant advantage, with 83.1% (123 responses) selecting this option. Close behind, “Speed” was also recognized as a primary benefit, chosen by 75.7% (112 responses). Additionally, 60.8% (90 responses) valued the ability of these tools to allow for “Comparison with my translations.” Conversely, “Consistency” was deemed less pivotal, with only 29.7% (44 responses) identifying it as a key benefit (see Table 4). This data suggests that users appreciate the quick and informative nature of AI translations, but consistency may be of secondary importance to them.
In response to the query about encountering significant errors or misconceptions due to AI translations, the majority of participants indicated they had not faced such issues. Specifically, 75.7% of the total 148 responses signaled “No” to experiencing significant misinterpretations. Conversely, a minority of 24.3% affirmed they had come across notable inaccuracies as a result of AI-powered translations. This indicates that while AI translation tools are generally reliable for most users, there remains a segment who have encountered issues, underscoring the importance of continuous improvement and refinement in the field of AI translation.
Participants who affirmed encountering discrepancies in AI-mediated translations were subsequently prompted to furnish illustrative instances. An analysis of their input yielded several recurrent themes. One predominant concern pertained to the intricate nature of sentence constructions, accentuated in translations between Portuguese and Chinese. Such complications were conspicuously present when discerning the subjects within these sentences. Translations of news articles posed a distinct challenge due to inaccuracies in rendering specific terminologies and appellations, consequently resulting in the dissemination of flawed information. Respondents underscored the system’s deficiency in faithfully translating Portuguese idiomatic expressions, with many translations adopting an overtly literal approach, thereby omitting the intrinsic connotations. There were reports of misinterpretations involving numerical values, pronouns, and personal nomenclatures. Frequent references were made to syntactic and morphological inconsistencies. Additionally, the rendition of specialized lexicon, literary passages, and colloquial terms was identified as problematic.
Regarding the impact of AI translation tools on individuals' learning and skill development in translation studies, a majority of the respondents (48%) felt that these tools had a positive influence. A notable proportion, representing 44.6%, maintained a neutral stance on the topic, neither attributing significant benefits nor drawbacks to AI’s role in their translation studies.
Lastly, in a Likert scale question about the importance of using AI tools for translation, the data underscores a prevailing belief in the value of these tools in the translation field. A significant majority of respondents, 77.1%, ranked its importance between 6 and 10, suggesting a general lean towards the higher importance of AI translation tools. Specifically, 11.5% viewed it as “extremely important,” providing the highest rating of 10. The middle ground, represented by a score of 5, was chosen by 14.2% of participants. On the other hand, 8.8% of respondents felt it had relatively lower importance, as reflected by their ratings between 1 and 4. Interestingly, not a single respondent felt it was not important at all, with a 0% response for scale 1.
Future prospects
In this final section of the questionnaire, students were asked to contemplate the future landscape of the translation profession and the integration of AI tools into the curriculum. Addressing the potential effects of AI on the demand for human translators, a pronounced 56.1% of respondents believe that the need for human expertise will wane with the rise of AI in translation. Interestingly, 15.5% of the students remained ambivalent, signifying their uncertainty about AI’s true impact on the translation realm.
In response to an open-ended question about the implications of AI translation tools in the professional translation sector, students freely expressed their thoughts and concerns. as shown in Table 5 The primary apprehensions centered on the potential unemployment among human translators, with many fearing they might be replaced by AI’s efficiency and evolving capabilities. Additionally, there were serious worries regarding the accuracy and precision of AI translations, especially for official or business documents. Many respondents felt that AI tools might miss out on capturing cultural nuances, emotions, and the inherent human touch in texts. An over-reliance on such tools, they believed, could lead to a decline in human translators' creativity and language skills. There were also concerns about the ethical implications, such as the undue influence of AI on academic writing. However, a small subset of respondents (only 7 out 148) indicated they held no particular reservations about the increasing role of AI in the translation sector.
The last question of the questionnaire asked for feedback and suggestions on how AI translation tools should be incorporated or treated in translation courses. Responses varied, with some respondents having no comments, while others offered diverse views. Some emphasized the importance of integrating AI tools as a reference, but not to rely on them entirely. Others suggested that AI can assist in specific domains or enhance translation quality and efficiency. There were concerns about over-reliance on AI leading to a loss in the essence of translation study. Recommendations also included teaching students the pros and cons of AI tools, ensuring they are used wisely. There were suggestions for interdisciplinary collaboration between translation and computer science, emphasizing the cultivation of scientific thinking in translators. Some felt that AI tools should be introduced only to senior students or those with higher language proficiency.
Discussion
The age distribution of the survey participants sheds light on a predominance of younger individuals, with ages 20 to 24 being notably overrepresented. This suggests that the views represented primarily come from a demographic that has grown up in the digital age, possibly making them more amenable and familiar with technological tools like AI translations. Considering the majority are from Macao and Mainland China, their cultural and educational backgrounds could significantly shape their views on AI’s role in translation.
Our research highlights a marked adoption of AI translation tools by students for their academic pursuits. Advancements in computational technology have enabled the seamless incorporation of AI into educational domains, heralding unprecedented prospects and conundrums. Our present research underscores a pronounced utilization of AI translation instruments by students in their academic endeavors. This trend not only exemplifies, but also resonates with Ouyang and Jiao’s (2021) discourse, suggesting that contemporary paradigmatic transitions within the realm of Artificial Intelligence in Education (AIEd) accentuate a trajectory towards personalized, data-centric educational paradigms, empowering learners to assume an increasingly autonomous role in their educational journey.
The prevalent use of platforms such as Google Translate and DeepL emphasizes their widespread approval and potential effectiveness. Notably, mentions of ChatGPT by certain respondents provide insight into the array of AI tools they utilize. It is important to note that ChatGPT is not directly accessible in Macao, necessitating the use of a Virtual Private Network (VPN) or platforms like Poe, which provide access to several AI-powered bots. Drawing from our classroom observations and autoethnographic records, the prevalence of ChatGPT for academic activities could potentially surpass the indications from the quantitative data. The fact that most students predominantly refer to use these tools as references, consistently cross-checking the translations, might initially suggest a tentative trust. However, this behavior could also reflect the students' heightened cognizance of the ethical discourse surrounding the deployment of AI in academia.
The perceived advantages of AI translation tools primarily centered on their speed and their assistance with unfamiliar terms. Despite these benefits, the consistency of AI translations was not deemed a priority by most participants. This may suggest that for many users, the immediacy and ease of access are paramount, potentially overshadowing the quest for perfect consistency. In the context of translation accuracy, a majority of participants believed that AI tools have not led them astray. However, it is paramount to consider the 24.3% who did encounter translation inaccuracies. Their feedback revolved around the challenges of translating complex sentence structures, particularly between Portuguese and Chinese, and the pitfalls of overly literal translations of idioms and expressions.
The abstract and metaphorical nature of the Chinese language presents unique challenges for AI translation when it comes to Portuguese. Portuguese has its own linear structure, enriched with distinct nuances and idiosyncrasies. While Portuguese often relies on a subject-predicate-object order and boasts a rich array of verb conjugations to express tense, mood, and aspect, Chinese offers a more spiral and diverse form. Chinese tends to emphasize subjects using an active voice and sometimes does not even require a subject or predicate to complete a sentence. The differing logical order between Portuguese and Chinese plays a pivotal role in translation. Portuguese employs specific words and verb forms to indicate relationships, whereas Chinese often conveys these relationships through mere sentence arrangement. Although AI has made strides in translation, there are evident gaps that are evidenced by our respondents’ experience when discussing misconceptions and misunderstandings in the translated materials by AI. This is in line with Wang’s (2023) results in regard to an experiment that assessed machine and human English-Chinese translations of several text typologies, and concluded that, when equipped with aids such as dictionaries, human translators have consistently shown better accuracy in translating between English and Chinese than AI. Yet, AI’s efficiency and ability to capture the overarching content of a text are commendable. Especially in business contexts, AI translations between English and Chinese are increasingly readable, accurate, and maintain a formal tone, thereby reshaping the role of human translators (Wang, 2023).
In terms of the broader academic implications, there appears to be a positive sentiment towards the impact of AI tools on students' learning in translation studies. Nevertheless, there remains a non-negligible portion of respondents who maintain a neutral position, perhaps reflecting a cautious optimism or reservations about the potential drawbacks.
A significant insight gleaned from the data is the anticipatory decline in demand for human translators due to the rise of AI. The open-ended responses underscored concerns regarding the potential erosion of the human touch in translations, especially in capturing cultural nuances and emotions. Despite these reservations, there seems to be a unanimous acknowledgment of the growing importance of AI tools in the translation field, with a majority emphasizing their value.
Regarding the curricular integration of AI tools, varied opinions emerged. While some emphasized the supplemental role of AI, others underscored the potential pitfalls of over-reliance. The suggestions for a strategic interdisciplinary approach between translation and computer science hint at the evolving landscape of translation studies in the age of AI.
This study offers a nuanced understanding of students' perceptions and utilization of AI translation tools. While the efficiency and accessibility of these tools are undoubtedly appreciated, there remains a consistent call for careful and critical engagement, ensuring that the essence of translation – with its intricacies and cultural nuances – is not lost in the quest for speed and convenience. The results underscore the need for a balanced approach, integrating AI advancements while preserving the core values of translation studies.
Conclusion
With the rapid advancement of AI and its expanding capabilities, the discourse around its role in translation has never been more pertinent. The findings from this study depict a generation of students who are not just welcoming the benefits of AI but are also acutely conscious of its shortcomings. Such discernment will be invaluable for future professionals who will be interfacing with AI tools regularly.
An important facet to address is the ethical ramifications of depending heavily on AI. This study briefly highlighted the potential influence of AI on academic writing. However, in professional contexts where the stakes for accuracy, accountability, and authenticity are high, errors in translation can lead to significant repercussions, especially in formal, legal, or sensitive documents. Here, the indispensable nature of the “human touch” and expert discretion becomes apparent.
As AI-driven translation tools gain traction, we recommend tertiary educational institutions re-evaluate their curricula. By doing so, they can both introduce students to this burgeoning technology and arm them with the analytical prowess required to decide when to employ AI versus when to rely on their own judgment.
Drawing from the initial findings of this ongoing research, we posit that the potential for collaboration between AI and human translators is enormous. However, it is crucial to emphasize that human Portuguese language experts and AI translators should be viewed as complements, not competitors. Human translators, serving as the conduit between the original author and the reader, have a knack for understanding the underlying essence and structure of texts, countering the limitations of AI translations. Conversely, AI excels at quickly pinpointing rudimentary grammatical oversights and facilitates a smoother translation process. For those fluent in both Portuguese and Chinese, their in-depth comprehension of both languages and cultures ensures translations remain true to the source text while resonating with the intended audience consequently, juxtaposing human cognition with artificial intelligence presents a paradigm of collaboration rather than rivalry. The synergy between human and artificial intelligences can serve as a confluence, collaboratively propelling advancements and augmenting both educational and overall life quality. By amalgamating these dual modalities of intelligence, we stand poised to sculpt a more advanced and enlightened global landscape, thus positioning ourselves to navigate and capitalize on the multifaceted prospects the impending future presents.
In light of the “Recommendations on Ethics of Artificial Intelligence (2021)” and UNESCO’s “Guide on ChatGPT and Artificial Intelligence in Higher Education (2023)”, it is imperative for Higher Education Institutions to champion an inclusive, multidisciplinary discourse on the aforementioned subjects. The unprecedented prowess and ubiquity of AI tools have not only underscored their indelibility in modern life but have also necessitated a contemplative reconsideration of their integration. It becomes crucial, therefore, to meticulously calibrate and assimilate the employment of AI within educational contexts, ensuring equilibrium and enlightenment. Institutions ought to rigorously reassess policies pertinent to academic integrity vis-à-vis AI utilities, offering stakeholders – students, academic staff, and faculty – a platform to deliberate upon the ramifications of AI within higher education and jointly formulate strategies for its judicious integration. Such endeavors could elucidate the potential of AI in bolstering pedagogical experiences and establish clear benchmarks. Contemporary circumstances mandate a transition to emergent paradigms, demanding intricate scrutiny and incorporation. However, this evolution may be fraught with challenges and discursive divergence, potentially unfolding at varied paces across geographical landscapes, given that certain nations yet retain prohibitions on tools such as ChatGPT.
Subsequent research can delve deeper into the exploration of specific frameworks or hybrid translation models wherein AI undertakes the preliminary bulk of translations, leaving humans to fine-tune, contextualize, and enhance the final output. This collaborative approach promises optimized efficiency without sacrificing the richness, quality, and cultural appropriateness of translations.
As a closing note, we argue that, while AI heralds an exciting horizon for the translation sector, the future seemingly leans towards a synergistic relationship between human expertise and machine efficiency. By amalgamating the unique strengths of both, we envision a translation landscape marked by unmatched efficiency, accuracy, and cultural fidelity.
Profile of survey respondents
| Characteristic | Category | Percentage (%) | Notes |
|---|---|---|---|
| Total respondents | – | 100 | n = 150 |
| Age | 18 years old | 6.7 | |
| 19 years old | 8.0 | ||
| 20 years old | 15.3 | ||
| 21 years old | 28.7 | ||
| 20 years old | 13.3 | ||
| 22 years old | 8.0 | ||
| 23 years old | 4.7 | ||
| 24 years old | 4.1 | ||
| above 25 years old | 11.2 | ||
| Gender | Female | 72.0 | |
| Male | 24.0 | ||
| Other/Prefer not to say | 4.0 | ||
| Region of origin | Macao | 86.0 | – |
| Mainland China | 14.0 | – | |
| Degree program | Bachelor’s | 83.3 | Primarily in final year (41.3%) |
| Master’s | 6.0 | Split between first (4.0%) and second year (2.0%) | |
| PhD | 10.8 | Majority in their third year (7.4%) |
Source(s): Created by authors
Utilization and dependence on AI translation tools (n = 150)
| Frequency of use | Percentage of respondents | Primary platforms used | Percentage of responses |
|---|---|---|---|
| Always | 18.7 | Google Translate | 84.5 |
| Often | 44.7 | DeepL | 81.1 |
| Occasionally | 29.3 | Baidu Translate | 14.9 |
| Rarely | 6.6 | ChatGPT | 8.2 |
| Never | 0.7 | Microsoft Translate | 2.7 |
Source(s): Created by authors
Perspectives on AI translation accuracy and confidence (n = 150)
| Verification of translations | Percentage of students | Confidence in accuracy | Percentage of students |
|---|---|---|---|
| Always | 43.2 | Somewhat confident | 43.3 |
| Often | 29.3 | Neutral | 40.7 |
| Occasionally | 21.6 | Not very confident | 12.7 |
| Rarely | 3.3 | Very confident | 3.3 |
| Never | 2.6 |
Source(s): Created by authors
Perceived benefits of AI translation tools
| Benefit | Percentage of students agreeing (n = 150) |
|---|---|
| Assistance with unfamiliar terms | 83.1 |
| Speed | 75.7 |
| Comparison with my own translations | 60.8 |
| Consistency | 29.7 |
Source(s): Created by authors
Consolidated concerns regarding the impact of AI translation tools on the professional translation field
| Primary concerns regarding AI in translation | Frequency mentioned |
|---|---|
| Job replacement/unemployment/AI replacing human translators | 52 |
| Inaccuracy in translations, complex sentences, and idioms | 29 |
| Accuracy and the quality of translations dropping | 22 |
| Overreliance on AI leading to loss of creativity and critical thinking | 15 |
| Reduced demand for human translators and interpreters | 14 |
| AI’s inability to understand cultural contexts and nuances | 12 |
| Loss of enthusiasm for learning languages | 12 |
| Decrease in opportunities for professional growth and learning | 9 |
| Impact on learning and skill development in translation studies | 7 |
| None/no concerns | 6 |
| Ethical concerns, including academic integrity | 5 |
| Concerns over AI handling unexpected situations or specific terminologies | 4 |
Source(s): Created by authors
References
Ai, Q. (2022), “Reflection on the machine translation and human translation in the age of artificial intelligence”, 2022 Fourth International Conference on Emerging Research in Electronics, Computer Science and Technology (ICERECT), Mandya, 2022, pp. 1-6, doi: 10.1109/ICERECT56837.2022.10060519.
Alharbi, W. (2023), “AI in the foreign language classroom: a pedagogical overview of automated writing assistance tools”, Education Research International, Vol. 2023, 4253331, doi: 10.1155/2023/4253331.
Baidoo-Anu, D. and Ansah, L.O. (2023), “Education in the era of generative artificial intelligence (AI): understanding the potential benefits of ChatGPT in promoting teaching and learning”, Journal of AI, Vol. 7 No. 1, pp. 52-62, doi: 10.2139/ssrn.4337484.
Chassignol, M., Khoroshavin, A., Klimova, A. and Bilyatdinova, A. (2018), “Artificial intelligence trends in education: a narrative overview”, Procedia Computer Science, Vol. 136, pp. 16-24, doi: 10.1016/j.procs.2018.08.233.
Chen, L., Chen, P. and Lin, Z. (2020), “Artificial intelligence in education: a review”, IEEE Access, Vol. 8, pp. 75264-75278, doi: 10.1109/ACCESS.2020.2988510.
Cheng, M. (2022), “Practical exploration of English translation activity courses in colleges and universities under the background of artificial intelligence”, Wireless Communications and Mobile Computing, Vol. 2022, 4547342, doi: 10.1155/2022/4547342.
Crossley, S.A. (2018), “Technological disruption in foreign language teaching: the rise of simultaneous machine translation”, Language Teaching, Vol. 51 No. 4, pp. 541-552, doi: 10.1017/s0261444818000253.
Deng, X. and Yu, Z. (2022), “A systematic review of machine-translation-assisted language learning for sustainable education”, Sustainability, Vol. 14 No. 13, p. 7598, doi: 10.3390/su14137598.
Ellis, N.C. (2017), “Cognition, corpora, and computing: triangulating research in usage-based language learning”, Language Learning, Vol. 67 No. S1, pp. 40-65, doi: 10.1111/lang.12215.
Ellis, C. and Bochner, A.P. (2000), “Autoethnography, personal narrative, reflexivity: researcher as subject”, in Denzin, N.K. and Lincoln, Y.S. (Eds), Handbook of Qualitative Research, Sage Publications, pp. 733-768.
He, Y. (2021), “Challenges and countermeasures of translation teaching in the era of artificial intelligence”, Journal of Physics: Conference Series, Vol. 1881 No. 2, 022086, doi: 10.1088/1742-6596/1881/2/022086.
Hellmich, E. and Vinall, K. (2021), “FL instructor beliefs about machine translation: ecological insights to guide research and practice”, International Journal of Computer-Assisted Language Learning and Teaching, Vol. 11 No. 4, pp. 1-18, doi: 10.4018/IJCALLT.2021100101.
Klekovkina, V. and Denié-Higney, L. (2022), “Machine translation: friend or foe in the language classroom?”, L2 Journal, Vol. 14 No. 1, pp. 105-135, doi: 10.5070/L214151723.
Lee, T.K. (2023), “Artificial intelligence and posthumanist translation: ChatGPT versus the translator”, Applied Linguistics Review, Vol. 0 No. 0, doi: 10.1515/applirev-2023-0122.
Melby, A.K. (2020), “Future of machine translation: musings on weaver's memo”, in O'Hagan, M. (Ed.), The Routledge Handbook of Translation and Technology, Routledge, London.
Ortega, L. (2017), “New CALL-SLA research interfaces for the 21st century: towards equitable multilingualism”, CALICO Journal, Vol. 34 No. 3, pp. 285-316, doi: 10.1558/cj.33855.
Ouyang, F. and Jiao, P. (2021), “Artificial intelligence in education: the three paradigms”, Computers and Education: Artificial Intelligence, Vol. 2, 100020, doi: 10.1016/j.caeai.2021.100020.
Peterson, R.A. (2000), Constructing Effective Questionnaires, Sage, Thousand Oaks.
Rossi, C. and Chevrot, J.P. (2019), “Uses and perceptions of machine translation at the European Commission”, Journal of Specialised Translation, Vol. 31, pp. 177-200.
Saldaña, J. (2013), The Coding Manual for Qualitative Researchers, 2nd ed., Sage, London.
Shen, G., Feng, T. and Cheng, Y. (2022), “Optimization of English learning mode under the influence of artificial intelligence translation”, Discrete Dynamics in Nature and Society, Vol. 2022, pp. 1-7, doi: 10.1155/2022/7755297.
Sun, Q. (2023), “On translation technology education at Chinese higher educational institutions in the age of artificial intelligence”, 2023 11th International Conference on Information and Education Technology (ICIET), Fujisawa, Japan, pp. 314-318, doi: 10.1109/ICIET56899.2023.10111488.
Tomasello, M. (2003), Constructing a Language: A Usage-Based Theory of Language Acquisition, Harvard University Press, Cambridge, MA.
UNESCO (2021), Recommendations on Ethics of Artificial Intelligence, UNESCO, Paris.
UNESCO (2023), ChatGPT and Artificial Intelligence in Higher Education: Quick Start Guide, UNESCO, Paris.
Vinall, K. and Hellmich, E. (2022), “Do you speak translate? Reflections on the nature and role of translation”, L2 Journal, Vol. 14 No. 1, doi: 10.5070/L214156150.
Wang, L. (2023), “The impacts and challenges of artificial intelligence translation tool on translation professionals”, SHS Web of Conferences, Vol. 163, pp. 1-6, doi: 10.1051/shsconf/202316302021.
Yang, C. (2023), “The application of artificial intelligence in translation teaching”, ICIST '22: Proceedings of the 4th International Conference on Intelligent Science and Technology, Harbin, 10-12 August 2022, pp. 56-60, doi: 10.1145/3568923.3568933.
Zhang, H. and Torres-Hostench, O. (2022), “Training in machine translation post-editing for foreign language students”, Language Learning and Technology, Vol. 26 No. 1, pp. 1-17, available at: http://hdl.handle.net/10125/73466
Further reading
Alam, A. (2021), “Possibilities and apprehensions in the landscape of artificial intelligence in education”, 2021 International Conference on Computational Intelligence and Computing Applications (ICCICA), Nagpur, 2021, pp. 1-8, doi: 10.1109/ICCICA52458.2021.9697272.
Huang, J., Saleh, S. and Liu, Y. (2021), “A review on artificial intelligence in education”, Academic Journal of Interdisciplinary Studies, Vol. 10 No. 3, p. 206, doi: 10.36941/ajis-2021-0077.
Muñoz-Basols, J., Neville, C., Lafford, B.A. and Godev, C. (2023), “Potentialities of applied translation for language learning in the era of artificial intelligence”, Hispania, Vol. 106 No. 2, pp. 171-194, doi: 10.1353/hpn.2023.a899427.
Acknowledgements
This project received funding from the Macao Polytechnic University (Research Project Code: RP/FLT-03/2023). Appreciation is extended to student Mok Man Sam for his work translating Chinese materials, providing language interpretation, and performing cultural verifications. Acknowledgement is also given to the participants who contributed data and engaged in surveys and interviews.