Can Translations Be Done Using AI? Complexities & Problems In Using AI.
Artificial Intelligence has become exceptionally popular today because of the many benefits it offers. It performs tasks within seconds, saving a significant amount of time, and is capable of scanning vast amounts of information on the internet to provide detailed insights; something that would take an ordinary person several days to accomplish. Although AI can mimic certain human actions and reasoning patterns, it remains artificial intelligence. It is not entirely error-free and carries several limitations that must be considered before relying on it.
For instance, while translating Sanskrit texts into English, AI often makes mistakes in sandhi-vigraha and samāsa-vigraha, which are crucial for producing accurate translations. Even a small error in vigraha can alter the meaning of a sentence, leading to incorrect interpretations.
Moreover, Sanskrit grammar is vast, and its literature is like an endless ocean filled with diverse texts and treatises. Accurate translation of any classical work requires deep expertise; not only in the text itself but also in its allied disciplines. For example, while translating a rare Tantric Jyotiṣa text, I had to consult the Manu Smṛti for traditional units of measurement used in certain remedial procedures.
Image Source: Pinterest.
Similarly, while working on Bṛhat Jātaka, one has to refer to Praśna, Muhūrta, Āyurveda, and other Horā texts to ensure that the translation aligns with the foundational principles of Jyotiṣa.
Unfortunately, AI falls short in this regard. It still cannot distinguish foundational doctrines from later inventions. For instance, it does not yet recognize that divisional charts (D-charts) are not mentioned in the classical Jyotiṣa canon and are a modern construct.
AI struggles with adhikāra, prakaraṇa, and the traditional context in which verses were written. This can lead to the translation of philosophical verses literally when metaphor is intended or misreading technical verses as poetic or vice-versa. Texts often omit words that are assumed to be understood by scholars (arthalopa). AI may supply incorrect words or misinterpret the implied meaning. There are several other problems like difficulties in identifying the correct meaning for polysemous words, errors in handling metaphors, problems in meter and syntax, etc.
Thus, while AI can certainly assist in the translation process, it cannot replace the nuanced understanding and interpretative skill of a human expert well-versed in the subject.
Comments
Post a Comment