The contest passage was taken from an article about artificial intelligence that appeared in an online Japanese government technical journal. Despite the article’s source, it was written in accessible and engaging lay language. The author, Satoshi Kurihara, a Keio University AI expert, scrupulously avoided technical jargon. The article describes the AI in general, gives several examples of AI applications, discusses people’s fears of AI, and ponders the future of AI. People now have plenty of reasons to view AI as a threat. There have already been “flash crashes” on the stock market fueled by AI stock-trading software and lethal autonomous weapon systems (LAWS) are already a reality. Computer recreations of deceased celebrities have often been unsettling. However, the author envisages a future where people will have a symbiotic relationship with AI that will allow humanity to ascend to a higher level. In this future, AI will even support and enhance human creativity. (The author himself was involved in an AI-assisted manga writing project, Tezuka 2020, attempting to create more manga in the manner of Osamu Tezuka.)
The contest passage was a fascinating discussion of the nature of automation and autonomy that segued into a discussion of concerns over AI weapons. Like the article overall, it was free of overly technical jargon and painted a picture that was easy for the layperson to see. Translation therefore had to reflect this informal register. That in itself is a challenge as the process of translation involves researching the subject matter and finding exact terminology. But the translations had to be toned down a bit to reflect the original text.
In addition to finding the right conversational register, challenges in the text included seemingly small things:
*Scanning the text for English words to see that 目的is referred to as “goal” (not “purpose”).
*Finding that 共生 really did map to “symbiosis,” but that would require reading the entire 10-page article, which seems like a bit of bait-and-switch for a translation contest billed as requiring one page of translation.
*Identifying the entire device/organism as the system.
*Sentence 6: “外界からの情報に対してどの振る舞いを発動させるのか、の関係も極めて複雑であるところにある” could be translated a number of ways with or without the “関係.” In it most literal form, appearing in one translation, it was a relationship “between” only one item (although the second item could be guessed). “The relationship between which behaviour is triggered in response to information from the outside world is extremely complex.” (between behavior and … [what?])
*Translating “意志をもって” which describes Roombas and AIBO. Elsewhere in the article the author says that these machines seem to be conscious (“意識を持ち”), but “意志をもって” probably does not mean they are conscious.
*Putting everyone in the right place in long Sentence 12 about who’s lagging behind whom.
*Relating deepfakes and LAWS to the rest of Sentence 13 (done by few).
Each of the translations by the contest finalists had its own strengths and it was difficult to rank one over another. The translations were performed in a highly conscientious manner (or showed intense concentration for a brief period of time). They also showed, by turns, insightful understanding of the text and creative use of English. Each contestant showed great potential as a translator, although not every person was at the same level of development. A summary of comments on each translation appears below followed by more detailed versions.
Scott conveyed the information in the contest passage accurately with only minor exceptions. The translation was generally free of glaringly odd word choices and stilted constructions typical of translationese. While this translation did not often soar to the rhetorical heights of some others, it marched steadily forward with very few major distracting or misleading missteps.
Like Robin and Catherine, Scott begins by capturing the reader’s attention with “take insects for example” for “昆虫の場合,” presenting the topic up front and starting to divide up an otherwise long and meandering sentence. Scott also used the word “goal” instead of “purpose.” (The author defines the Japanese word 目的 as the English word “goal” in the context of task completion.) In Sentence 2, “autonomy” and “automatic” are clearly distinguished as two separate characteristics.
Scott actively rearranged the text to be palatable to the English reader, going so far as combining and redividing sentences on two occasions.
The rhetorical device “Now, humans” clearly signals that the topic has switched from insects to humans in Sentence 3. In the series of things people do in Sentence 4, the expression “not to get yelled at” maintains the informal tone “怒られるのがイヤだ.” Long Sentence 8 is divided up into palatable portion suitable for the English reader.
The contest screeners pointed out that Sentence 12, the “Japan is lagging behind” sentence, was particularly challenging to contestants. In that sentence, Scott retaining the original wording with a forceful “possible for Japan to secure a solid position”. Scott retained the “abominable war becomes even more abominable” sequence of the Japanese in the last sentence.
Scott made one fairly serious error in the second “difference” in Sentence 6, which may be saying proofreading error or failure to proofread. In “the links between those actions and which are activated in response are extremely complex” (外界からの情報に対してどの振る舞いを発動させるのか、の関係も極めて複雑であるところ), “information from the outside world” is translated as ”actions” and “which are activated” looks garbled unless a “those” is added ([those] “which are activated”). In a similar instance, with the issue of LAWS in Sentence 13, “as we've seen before with the issue of deep fakes arising along with machine learning, such has the issue of LAWS (Lethal Autonomous Weapon Systems) been identified along with autonomous AI technology”, the apparent typo of “such” for “so” would be extremely disorienting to the reader. Scott also said the organisms and devices have systems when they appear to be the systems.
However, other errors tended to be minor and not a major distraction to the reader.
Robin’s translation was also excellent. It contained many of the best translations of individual words or expressions and, unquestionably, many of the clearest sentences in the contest. It was also highly accurate. This made ranking the contestants very difficult. However, there were a few places that may have confused the reader.
Like Scott, Robin starts with the construction “consider an insect” to focus the reader's attention on the topic and begin dividing up this rather long sentence. Robin used “live” and “goal”, the more suitable renderings here.
In Sentence 5, “boils down to” for “元を正せば” and “the most apt course of action for a given situation” for “状況に応じて適切と思われる行動” are the best translations of those terms in the contest. In Sentence 6, “and determining which course of action to take can be incredibly complicated” does away with the often superfluous “関係” and emphases the “を発動させるのか” in “外界からの情報に対してどの振る舞いを発動させるのか、の関 係も極めて複雑であるところ.” In Sentence 7, “adept” is indeed an adept translation for “巧妙”. ”to have a will of their own” is an accurate translation of “意志をもっているように” in Sentence 10. In Sentence 12, ““when it comes to autonomy” retains the conversational style of the article.
However, Robin chose to use the singular “the insect” and “the human” when plurals may have fit better. (The author refers to “we humans”.) In Sentence 2, “autonomous” and “automatic” sound like different degrees of the same characteristic in “The insect can thus be seen as an autonomous, quite nearly automatic living organism.” In Sentence 5, “perpetual system” sounds conspicuously out of place in this article. In Sentence 8, “反応の仕方” is translated as “response system” when “how we respond” may be more appropriate.
Overall, Ayami’s translation was accurate and extremely pleasant to read. However, there was a high number of usage errors (usage errors are only a problem when they sound strange to the reader) and the register of the English was a bit too high.
Ayami was one of the few contestants to translate Sentence 6 as a comparative statement. In Sentence 7, Ayami also used “adept” for “巧妙” and used “systems” as the subject, which accounts for all involved [wo]man, beast, and machine. A particularly concise translation was provided for Sentence 12 about “lagging behind.” Ayami said that organisms and devices were the systems, not that they had the systems.
Catherine’s English was particularly appropriate for the Japanese text. Catherine also began with the strong statement “Consider insects, for instance.”
She identified whole organism/device as the system and faithfully stuck with “secure a solid position.” Catherine also attempted to address the connection between “指摘されるように” and “アクティブに進められている” in Sentence 13 in a very logical manner. Taking quite a bit of license she used field-specific terminology for LAWS. In Sentence 9, “that distinguishes” in “It is perhaps this ability … that distinguishes a system as having autonomy” conveys the emphatic nature of “と呼べるのではないだろうか”. In Sentence 11, Catherine was one of the few contestants to use “symbiotically” for “共生” to be consistent with the author’s theme. In Sentence 13, Catherine attempted to address the connection between “指摘されるように” and “アクティブに進められている”. In Sentence 14, “make the final decision to pull the trigger” is exactly the right English expression to use.
Unfortunately there were large meaning-transfer errors when classifying insects in sentence 10 and concerning problems with LAWS in Sentence 13. There were also quite a few grammatical and usage errors and language that was in the wrong register.
Mizuki’s translation is pleasantly written with some vivid and eloquent phrasing. Mizuki uses “alive” and “living” instead of “survival” and “goal” instead of “purpose.”
In Sentence 4, “we do not want to get in trouble” is exactly the same register as the Japanese, and “we go somewhere in order to meet someone” doesn’t subject the English reader to some contorted usage of “move”. In Sentence 6, the differences are correctly cast as comparisons. In Sentence 12, the use of “we” in “I don't think we are behind” reminds the reader that the author is Japanese and is writing from a Japanese perspective. In Sentence 13, Mizuki gives the best translation of the finalists for the seemingly haphazard connection between “問題が指摘されるように” and “進められている” with “Lethal Autonomous Weapons Systems (LAWS) [are] causing controversy in the field of autonomous AI, as deepfake[s] is in that of machine learning, and it shows that autonomous AI is actively being researched and developed around the world.
Mizuki’s liberal recasting of sentences, while generally good in J-E translation, might have made some sentences unclear or changed the meaning. There were some article misusage, pluralization problems, trouble with some English expressions, and a few awkward spots. However, the problems were not that severe and the text doesn’t really sound translated.
Ciaran’s English was very pleasant in a conscientious translation that included an excellent rendition of “元を正せば” as “”it ultimately comes back to” in Sentence 5. Sentence 13 was extremely well written and Ciaran tied together “問題が指摘されるように” and “進められている”. The translation of the final sentence was forceful, faithful, and eloquent.
Valiant attempts to provide glue or context might have led away from the meaning as in “Only two factors separate our two species” in Sentence 6. Unfortunately, there were significant or major transfer errors that obscured the meaning to the reader, involving “移動” in Sentence 4, “巧妙に” in Sentence 8, and who’s lagging behind in Sentence 12.
(Continued to Commentaries from the Judges 3-2)