Fifteen years ago, a chess-playing computer defeated the finest grandmaster in the world. Small wonder that some people see artificial intelligence as the next stage in human evolution. Yet despite the rapid and continual advances in technology, there are still certain tasks, relatively easy for humans, that computers find difficult. Facial recognition is one of them. Another is translation. For a sophisticated machine, thinking about a rook’s capture of a bishop four moves in advance is child’s play compared to figuring out idioms, metaphors, abbreviations, double meanings and slang.

A couple of weeks ago I happened to read a column on the TSN website about the Montreal Canadiens. It was a machine translation, made by Google, of a column by Mario Tremblay that had appeared on the website of RDS, the network’s sister site, in French. Tremblay’s prose is not particularly idiomatic, and most of his points were just about comprehensible – but it would be an exaggeration to say the TSN story was in English. Here are four examples of the published Google translation, followed by my human version.

“The Director General of the Canadian, Pierre Gauthier, seller is present: ” that’s how the column began. (The general manager of the Canadiens, Pierre Gauthier, is now ready to sell.) “Gauthier must linger to grow defensive.” (Gauthier needs to take his time to make the defence bigger.) “Many have before him.” (There are many teams ahead of them.) “His absence changed the data.” (His absence altered the situation.)

“Absence,” in French, is feminine, and the next sentence in Tremblay’s original column read as follows: “Elle a forcé Gauthier à aller chercher Kaberle et Chris Campoli.” Mysteriously, the machine twisted this into “She was forced to seek Gauthier Kaberle and Chris Campoli.” Google introduced other mistakes too. “Raphael Diaz et Subban, qui en est à sa deuxième saison” turned into the erroneous as well as ridiculous “Subban and Raphael Diaz, now in its second season.”

But even when the translated meaning is clear, the column’s wording often sounds peculiar to the ears of a native speaker: “the large center Ryan Getzlaf;” “none of them was sitting in the right chair because of injuries.” If it’s hard for a machine to translate a sentence with accuracy, it seems impossible for the machine to achieve elegance too. Even the most diehard technophile should not imagine that Google will eliminate the need for human interpreters and translators any time soon.