Abstract
Human intelligence (HI) has used artificial intelligence (AI) in professional translations for many years. What has been so far a helpful tool for translators, turns out to be a formidable competitor. The article tackles the topic of the danger represented by the dramatic reconfiguration of a job, which risks losing much of its consistency, getting closer and closer to post-editing. HI and AI performances in the translator profession are approached from an economic perspective, setting as criteria for analysis the elements that define the price and survival on the market: source language, target language, type of document, content subject, delivery date, the volume of text to be translated, the competence of the translator, availability of the translator, capability to learn, costs, accuracy and risk of errors. The methodological analysis of a representative sample of different texts from the economic field translated into five foreign languages, reveals that the results provided by AI are fully acceptable and competitive with the versions generated by HI. In this context, the article warns about the need to rethink the training of translators and the sustainability of their activity in the economic market.
You have full access to this open access chapter, Download conference paper PDF
Keywords
1 Introduction
Few innovations have possessed the power to completely and irreversibly disrupt a predictable order of things and the comfortable stability of the global population. 2022 brought the market launch and widespread availability of an innovation with the potential to profoundly alter the traditional constructs of humanity: Chat Generative Pre-training Transformer (ChatGPT), a product of Artificial Intelligence. Some have comprehended its benefits as well as imminent dangers; authorities hesitated between prohibition and acceptance-integration. Many who tested it have praised its potential, while most currently choose to overlook it, thus sidestepping the risk of jeopardizing their future job security. In this article, we have delved into the prospects of the translator profession, positioned at the border between tradition and innovation, and emblematic of careers susceptible to technological progress and AI capabilities.
2 Literature Review
Early evidence has shown that technology and artificial intelligence will become the new frontier to conquer in the digital era, and the relationship between translations and machines will become a favored study subject for specialists [1], extensively reflected in the press [2]. Initially, the field of translation studies focused on aspects related to human intelligence, such as translation techniques, lexical issues, grammar, style, and interculturality. However, with technological advancements, machines were introduced into the translation process, initially complementing and eventually replacing traditional writing tools: pen and paper, typewriter, or word processing software. The concern for the ethics of introducing artificial intelligence technologies in the field of translation, which is seen as the boundary between blessing and threat for linguistic professionals, is one of the preferred research directions. Studies provide valuable insights into the historical development [3, 4] and the ongoing advancements in machine translation, addressing topics like the future and the responsibilities of users and translators. The utility of machine translation (MT) for academic communication is highlighted, along with the ethical issues arising from using these robots concerning data and information confidentiality and biases related to personal health, finances, and other sensitive domains [5, 6]. In the context of traditional translation versus modern translation, the theme of human–machine interaction leads to the observation that Machine Translation (MT) can, in some aspects, partially replace Human Translation (HT), but not entirely [7]. Numerous studies pursue another direction in relation to the artificial intelligence-translation field, specifically the preparation of future translators who need to keep up with the rapid evolution of increasingly advanced software and applications. This is achievable through rethinking new training methods and evaluating future translators, who must be trained to handle tasks beyond the usual translation and acquire new skills to facilitate their work [4, 8]. This aspect raises new challenges for educators. Thus, the need to identify new skills experienced and novice translators must acquire to survive as translators in the digital world is emphasized [5, 6], with ethical preparation ensured during their training [9]. Consequently, preparing future translators and assessing their level of competency are becoming increasingly challenging [10]. However, the introduction of ChatGPT to the market represents a turning point. Research on the impact of this model and other similar models is still in its infancy [11], with the focus directed toward the medical field [12, 13] or education [14]. Nonetheless, we will undoubtedly witness an influx of studies. Many of these will certainly highlight a change in perspective regarding the perception and impact of AI on various domains of activity [15]. Thus, the aim of this paper is to address the challenges, focusing on the profession of translator in the wake of technological advances, namely AI.
3 Background Research
Internationalization and the need for new markets have increased the translation demand, which is why the translation industry is continuously evolving. Due to globalization, translation has become an integral aspect of corporate strategies across all sectors. Various types of translation exist, each with its particularities and specifications, with the most popular being literary, technical, administrative, financial, and legal translations [16]. There is an international ISO standard that regulates all aspects related to the translation process and human translation services [17], as well as an ISO standard that regulates the post-editing of machine translation output [18]. Both standards identify and define two distinct categories of automated translation. Machine translation is the translation carried out by software based on provided data. Examples include Google Translate, DeepL or Reverso, which perform automatic translations mechanically and quickly without considering the translation context. These free software applications started with weaker versions but have improved over time, becoming increasingly efficient. On the other hand, computer-aided translation (CAT) is when “software applications are used to support the task of human translation.” In this case, translators utilize platforms integrating machine translation technologies and translation management tools like Trados or WordFast to facilitate and optimize their translations. However, due to technological innovations, post-editing services are becoming increasingly prevalent in the market. Depending on demand, translators are more inclined to offer services such as full post-editing or light post-editing rather than traditional translation [19].
The emergence of new generative AI models like ChatGPT or similar ones marks a radical change, a fact that translation market players are well aware of. As Loney [20] noted, “Generative AI has taken the industry by storm recently, with many prognostications about its impact on our daily lives. It remains to be seen if any of those predictions will come true, but we want to make sure that any future is a place where translators thrive and gain advantages from this new technology” [20]. Our research focuses on the comparative analysis of the performances and limitations of translations made by human intelligence and artificial intelligence and aims to highlight the danger that the translator profession is facing due to the sophistication and competence of ChatGPT.
4 Methodology
The research is built around two hypotheses. Hypothesis 1. Artificial Intelligence/ChatGPT becomes a powerful competitor for the profession of translators in terms of economic profitability. In order to discuss this hypothesis, we identified a series of relevant parameters in defining the translator profession and its functioning in a competitive market: source language, target language, type of document, content subject, delivery date, the volume of text to be translated, the competence of the translator, availability of the translator, capability to learn, costs, correctness and risk of errors. We evaluated comparatively HI and AI performances in relation to these parameters to establish their limited or unlimited capabilities and values.
Hypothesis 2: AI is a significant competitor for HI in terms of the quality of the generated texts. In order to verify this hypothesis, we selected, among specialized languages, a representative sample of 50 texts in Romanian, belonging to the economic field, more precisely the subfields of finance and accounting, marketing, management, international affairs, and the economy of trade, tourism, and services. Regarding the types of documents that are very used and suitable for translations, the range is extensive, including text files, emails, notes, commercial letters, reports, business plans, product specifications, contracts, agreements or even web pages. The texts chosen are representative of the investigated domain and were selected based on the typology analysis of the analytical programs offered by faculties specializing in translation within the University of Oradea [21] and the Bucharest University of Economic Studies [22]. The texts were labelled according to their domains, accompanied by their sources, and recorded with the chosen foreign language variants. A set of five languages was selected to encompass internationally recognized languages (English and French), languages with Latin (Hungarian) or Cyrillic (Russian) alphabets from neighboring countries, and a language spoken by a reduced number of people (Albanian). Excerpts (10–20 lines each) were chosen from the mentioned domains: formalized documents, university manuals, specialized books, and other specialized documents. ChatGPT was utilized to translate the texts into five foreign languages (English, French, Hungarian, Russian, and Albanian). These translations were then evaluated for acceptability with the input of native speakers experienced in economics in the mentioned languages.
5 Results and Discussion
To verify Hypothesis 1, we have approached translations from an interdisciplinary perspective, combining philological and economic viewpoints. We have analyzed the key elements that contribute to the dynamics of supply and demand in the translation sector: source language, target language, type of document, content subject, delivery date, volume of text to be translated, translator's competence, translator’s availability, capability to learn, and costs.
According to the market supply and the specializations of the faculties, a human translator works, on average, with two or three foreign languages; exceptionally, the human translator might master four or more foreign languages while AI-assisted translation does not encounter such limitations. A text can be effortlessly processed through nearly any number of languages, and the accuracy of the resulting product is astonishing.
Another advantage of using AI is cost-effectiveness, a significant aspect for both clients’ budgets and translation service providers. Various factors influence the cost of translation, but human translations are significantly more expensive than automated ones. Every client desires low costs and high quality, and this can be achieved when a translator works with highly effective translation applications that automate a substantial part of the process. However, even with machines, the most professional translations are provided by paid programs, inherently involving certain costs. Additionally, the training aspect of those who use and benefit from these software tools adds to the consideration.
Another strength of AI is that there's no risk of errors due to fatigue or lapses in attention, which sometimes occur with humans (e.g., skipping text fragments or accidental deletions). The translator’s availability, physical endurance, workload, and deadlines are subjected to human limitations, whereas mechanized translation generates text almost instantaneously. In the case of traditional translation, a single translator is often insufficient; large projects sometimes require teamwork, including a project manager. Hence, it can be stated that anyone who claims they can achieve quality translation standards without the assistance of translation software is naive. AI has become a formidable competitor due to its advantages over the human factor. The results demonstrated that HI has limited capabilities in relation to these parameters, while AI has unlimited capabilities.
Hypothesis 2 was examined through the precise investigation of economic texts based on the above-mentioned criteria. There are numerous types of texts within the mentioned domains, which may require professional or personal translation.
After translating the selected economic texts into the targeted languages, we have systematically observed that AI can translate just as quickly and effectively from any source language to any target language. While we chose only five languages, our simulations show that ChatGPT can efficiently work with unlimited languages, regardless of the alphabet, language family, or prevalence. The AI-provided translated versions of the selected documents were outstanding, with no noticeable errors or hesitations concerning economic terminology, simple or complex financial concepts.
Regarding accuracy, AI translation has significantly improved and continues to do so day by day due to the incredible pace of AI learning. While human translators continuously learn, they will never be able to match the speed at which AI develops its capabilities. We anticipate that language models like ChatGPT or its equivalents will soon provide consistently high-quality translations compared with human translations, ensuring consistency across large volumes of content, compared to human translators who might introduce slight variations in their translations based on individual style.
In terms of meeting deadlines, translation through ChatGPT is faster and more scalable than human translation. It can handle large volumes of text within a short time frame, which is advantageous for companies with tight deadlines or high translation demands. ChatGPT-based translation relies on AI algorithms to automatically translate text from one language to another. These translations were obtained for free, whereas human translation would have involved hiring professional translators with expertise in the economic field, fluent in both source and target languages. The cost of human translation would have varied significantly based on source and target languages, complexity, and subject matter.
However, there are still some critical differences between AI-based and human translation in economic texts. Human translators, especially those with experience in economics and finance, can currently better understand the context and provide precise and contextually relevant translations. AI delivers excellent results, but as ChatGPT itself warns when asked about its competence in economic translations, the AI-generated translation versions may have hesitations, ambiguities, or inaccuracies. ChatGPT recommends that users verify and review automated translations before using them in critical or professional contexts. This is a topic that will need further exploration in the future. Table 1 depicts comparisons between HI and AI.
Artificial Intelligence is a formidable competitor to human intelligence. While HI (Human Intelligence) has limited resources in terms of source language, target language, type of document, content subject, delivery date, volume of text to be translated, capability to learn, competence of the translator, and availability of the translator, AI has unlimited capabilities in these aspects. Regarding the risk of errors, both humans and machines can make errors, whether small, medium, or large. It all depends on the competence of the translator and should be viewed in light of the astonishing ability of models like Chat-GPT to learn. Human translators must continuously learn to ensure quality services without mistakes. However, a significant difference can be observed in terms of costs. Human translation comes with a cost, as the perceived prices can be significant, and sometimes even very high. On the other hand, Artificial Intelligence generates high-quality materials for free or at low prices, leading to dramatic cost changes. In terms of the accuracy of the versions, both HI and AI generally produce good or very good translations, but sometimes also poor ones. However, it is crucial to understand that at this moment, competent HI is needed to evaluate AI results. And HI must learn how to turn AI into an ally and use its capabilities in its own interest.
6 Conclusions
The translation versions generated by AI, without undergoing a thorough philological analysis, prove to be of sufficient quality and have enough advantages (cost, wide range of languages, speed, etc.) to prompt clients (companies, publishers, individuals, institutions) to request human post-editing services rather than traditional human translation. Nevertheless, human translation still remains the preferred option for critical or complex content, creative materials, or highly specialized domains due to its accuracy, contextual awareness, and ability to convey emotions and tones effectively. Therefore, human translators still play a crucial role and are generally considered superior in certain situations. However, in the future, due to the emergence of increasingly sophisticated language models, the focus in the translation market will likely shift from translation to post-editing. All these aspects serve as a wake-up call for the stakeholders involved in the translation process: educational institutions (philology or applied modern languages departments), service providers, and recipients of the finished translated product. This research opens up new perspectives by testing specific content from different domains (medical, legal, ethnic, literary) or considering ethical or technological performance aspects and the need of training/retraining of professionals in different fields.
References
Gaspari F, Almaghout H, Doherty S (2015) A survey of machine translation competences: Insights for translation technology educators and practitioners. Perspectives, 23, 333–358. https://doi.org/10.1080/0907676X.2014.979842
Vieira LN (2020) Machine translation in the news: A framing analysis of the written press. Translation Spaces, Volume 9, Issue 1, 98–122. https://doi.org/10.1075/ts.00023.nun
Bowker L (2021) Translation technology and ethics. In: Koskinen, K., Pokorn, N.K. (EDS) The Routledge Handbook of Translation and Ethics, London, Routledge, pp. 262–78
Bowker L, Buitrago Ciro J (2019) Machine Translation and Global Research: Towards Improved Machine Translation Literacy in the Scholarly Community. Bingley, Emerald Group Publishing
Hunt M, O’Brien S, Cadwell P, O’Mathúna DP (2019) Ethics at the intersection of crisis translation and humanitarian innovation. Journal of Humanitarian Affairs 1: 23–32. https://doi.org/10.7227/JHA.022
Neubert A (2000) Competence in Language. In Languages and in Translation. In Schäffner C, Adab B (eds) Developing Translation Competence, pp. 3–18. John Benjamins, Amsterdam. https://doi.org/10.1075/btl.38
Lihua Z (2022) The Relationship between Machine Translation and Human Translation under the Influence of Artificial Intelligence Machine Translation, Mobile Information Systems, vol. 2022, https://doi.org/10.1155/2022/9121636
Zaretskaya A, Pastor C, Seghiri GM (2015) Integration of Machine Translation in CAT tools: State of the art, evaluation and user attitudes. In: Skase Journal of Translation and Interpretation 8: 76–88. Corpus ID: 219634109
Ramírez-Polo L, Vargas-Sierra C (2023) Translation Technology and Ethical Competence: An Analysis and Proposal for Translators’ Training. In: Languages; 8(2):93. https://doi.org/10.3390/languages8020093
Pym A (2013) Translation Skill-Sets in a Machine-Translation Age. In: Meta: Translators’ Journal 58: 487–503. https://doi.org/10.7202/1025047ar
Tavares C, Tallone L, Oliveira L, Ribeiro S (2023) The Challenges of Teaching and Assessing Technical Translation in an Era of Neural Machine Translation. In: Education Sciences, https://www.mdpi.com/2227-7102/13/6/541
Chen TJ (2023) ChatGPT and other artificial intelligence applications speed up scientific writing. In: Journal of the Chinese Medical Association, 86(4), 351–353. https://doi.org/10.1097/JCMA.0000000000000900
Gordijn B, Have Ht (2023) ChatGPT: evolution or revolution? Med Health Care and Philos 26, 1–2. https://doi.org/10.1007/s11019-023-10136-0
King MR (2023) The future of AI in medicine: a perspective from a chatbot. Ann Biomed Eng 51, 291–295. https://doi.org/10.1007/s10439-022-03121-w
Stokel-Walker C (2022) AI bot ChatGPT writes smart essays—should professors worry? In: Nature 2022. https://doi.org/10.1038/d41586-022-04397-7
https://www.intertranslations.co.uk/what-are-the-5-most-common-types-of-translation
I. SO 17100:2015 Translation services—Requirements for translation services, https://www.iso.org/standard/59149.html, last accessed 20.07.2023
ISO 18587:2017 Translation services—Post-editing of machine translation output. Requirements, https://www.iso.org/standard/62970.html, last accessed 20.07.2023
Tavares C, Oliveira L, Duarte P, Moreira Da Silva M (2023) Artificial Intelligence: A Blessing or a Threat for Language Service Providers in Portugal. Preprints.org., https://doi.org/10.20944/preprints202307.0163.v1
Loney N (2023) Continuing our commitment to translate everything with the Trados Studio 2022 service release (SR1), available online at https://www.trados.com/blog/translate-everything-with-sr1-for-trados-studio-2022.html
https://fisadisciplina.ase.ro/SelectiePlan.aspx?pp=19, last accessed 20.07.2023
http://steconomiceuoradea.ro/wp/archives/1819, last accessed 20.07.2023
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Copyright information
© 2024 The Author(s)
About this paper
Cite this paper
Constantin, F., Pop, AM., Sim, MA. (2024). Human Intelligence and Artificial Intelligence in Professional Translations — Redesigning the Translator Profession. In: Kavoura, A., Borges-Tiago, T., Tiago, F. (eds) Strategic Innovative Marketing and Tourism. ICSIMAT 2023. Springer Proceedings in Business and Economics. Springer, Cham. https://doi.org/10.1007/978-3-031-51038-0_27
Download citation
DOI: https://doi.org/10.1007/978-3-031-51038-0_27
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-51037-3
Online ISBN: 978-3-031-51038-0
eBook Packages: Business and ManagementBusiness and Management (R0)