How does ChatGPT Compare with Conventional Neural Machine Translation Systems in Performing a Chinese to English Translation Task?

LIJIAN CAI
DOI: https://doi.org/10.3726/jts012024.02
2024-01-01
Journal of Translation Studies
Abstract:Abstract As machine translation (MT) systems have made significant progress in recent years, it is essential to understand their strengths and weaknesses. In this paper, I look at how ChatGPT compares with DeepL and Google Translate in performing a Chinese to English translation task. This comparative analysis points to better readability of ChatGPT-generated translations compared with output by the conventional neural machine translation (NMT) engines. The analysis also unequivocally affirms that the enhanced capabilities noted have not resulted in an adequately high level of context awareness able to shield algorithms from generating incoherent or contextually irrelevant texts or outright nonsense; and it concludes that the human touch remains indispensable and that any algorithmic takeover of the translation industry is not going to happen anytime soon. This does not mean, the paper argues, that there is no urgency to relinquish a business-as-usual mindset. The paper highlights the imperative to embrace AI technology as a transformative force to remake the translation industry; as well as the need for translation education programs with a view to staying relevant through a more sensible focus on rigorous training in practical skills and critical thinking, and the cultivation of linguistic and cultural competency.
What problem does this paper attempt to address?