AI Advancements Placing Human Translators in Potential Decline?
In a recent warning, the French Translators' Society (SFT) has raised concerns about the increasing use of Artificial Intelligence (AI) in language translation, highlighting several potential issues across literary, professional, and legal domains.
The key concerns revolve around the creative and literary limitations of AI, economic and professional threats to translators, quality and contextual accuracy issues, limitations in real-time and conversational translation, and cultural and legal implications.
Firstly, AI translation models tend to simplify language, lacking the ability to capture playfulness, nuance, cultural allusions, and emotional depth essential to literary texts. This results in a loss of literary flow and creative fidelity.
Secondly, translators face economic harm as AI-generated translations often force them into post-editing tasks that are more complex and time-consuming but compensated at much lower rates. Furthermore, translators are frequently denied authorship rights and royalties for post-editing AI work, impacting their professional recognition and income.
Thirdly, AI often fails to ensure contextual accuracy, terminological consistency, and legal nuance, particularly in specialized domains such as legal translation. This necessitates sustained human oversight and expertise in post-editing to preserve the integrity of translated content.
Fourthly, AI struggles with rapid, natural conversational translations and research requiring nuanced semantic decisions. It particularly struggles with less-known texts and marginal languages, creating barriers for users needing real-time comprehension and decision-making in multilingual contexts.
Fifthly, AI's inability to fully grasp embedded legal culture and comparative law nuances poses challenges to the translation of legal documents in the EU. This introduces fears about the protection of legal cultures, as AI may threaten the subtleties preserved by human expertise.
The broader implications of integrating AI in translation include a shift toward a posthumanism view, challenging the traditional human-centered approach but also raising ethical and professional questions about the future role of human translators. Large corporations leveraging AI for translation risk severely disrupting the translation job market and diminishing the role of human translators in dominant industries.
To address these concerns, the SFT recommends respecting the human factor, highlighting the added value of translators, and promoting greater transparency of the origin of textual contents. Public authorities are encouraged to take action for better sharing of the economic value of a translation, deliberate use of human translators, and stopping funding for AI solutions in translation.
A notable example of a botched translation by AI is the Washington Post article translated by AI in "Libération," which betrays the original thought and dulls it. Translation, being a widespread practice involving various actors such as companies, public services, citizens, and legal entities, requires careful consideration and responsible use of technology to preserve its quality, cultural nuance, and the economic stability of translators.
References: [1] The Local, (2021). French translators fear for their future as AI-powered translation tools become more common. Retrieved from https://www.thelocal.fr/20210429/french-translators-fear-for-their-future-as-ai-powered-translation-tools-become-more-common
[2] Le Monde, (2021). Les traducteurs font face à l'ère de l'intelligence artificielle. Retrieved from https://www.lemonde.fr/les-decodeurs/article/2021/04/28/les-traducteurs-font-face-a-l-ere-de-l-intelligence-artificielle_6080625_1643812.html
[3] The Guardian, (2021). AI translators are a threat to journalism, says French translators’ union. Retrieved from https://www.theguardian.com/world/2021/apr/29/ai-translators-are-a-threat-to-journalism-says-french-translators-union
[4] The Conversation, (2021). The hidden vices of AI in translation. Retrieved from https://theconversation.com/the-hidden-vices-of-ai-in-translation-160817
Artificial Intelligence (AI) may struggle to accurately and faithfully translate literary texts due to its inability to capture nuance, cultural allusions, and emotional depth, resulting in a loss of literary flow and creative fidelity.
AI-generated translations, often requiring post-editing, pose economic threats to translators by offering lower compensation rates for more complex and time-consuming tasks, and by denying authorship rights and royalties.
