Zhou Haizhong, a mathematician and linguist in China, once pointed out in the article Fifty Years of Machine Translation: To improve the translation quality of machine translation, the first thing to be solved is the language itself rather than the programming problem; It is certainly impossible to improve the translation quality of machine translation by relying on several programs to make a machine translation system. At the same time, he also pointed out that machine translation can't reach the level of "faithfulness and elegance" until human beings understand how the brain makes fuzzy recognition and logical judgment on language. I'm afraid this view just illustrates the bottleneck that restricts the quality of translation.
It is worth mentioning that Ray Coswell, an American inventor and futurist, predicted in an interview with The Huffington Post that by 2029, the quality of machine translation will reach the level of human translation. There are still many controversies about this assertion in academic circles.
In any case, people are most optimistic about machine translation at present, and this worry is based on an objective understanding and rational thinking. We also have reason to believe that with the joint efforts of computer experts, linguists, psychologists, logicians and mathematicians, the bottleneck problem of machine translation will be solved.