Analyzing Translation Processes in AI-Generated Educational Videos: A Cognitive Load Human–AI Study of English–Arabic Content

Abstract

The study looks at the methods which the translator and the Machines may use for the translation of AI (artificial intelligence) videos injected in education. The cognitive motivation behind our comparative study of human and AI English-to-Arabic translation processes. This research employs a process-oriented design. This study draws on knowledge from Cognitive Load Theory, Translation Process Research and Multimodal Learning Theory to investigate multimodal instruction’s cognitive effort distribution. The most commonly used indicators to assess cognitive load are (a) the NASA Task Load Index (b)translation time process indicators (c) revision frequency; and (d) reformulation density.  According to the findings, human translators are highly cognitively loaded. It is interesting that the cognitive load with the task is more when it is multimodal and heavily contextualised by instruction. The human translator makes use of cognitive mediation. Additionally, in this case, using AI to translate something doesn't work quite well as the proper time to do the same does not happen and also it is not efficient. However, it certainly cannot respond to multimodal instruction. The study shows how translation is of a cognitively and pedagogically significant activity in education and AI. Consequently, this follows.