Lady Gaga accepts role of CTO for top MT provider

Lady Gaga接受TopMT提供商CTO的角色

2021-04-01 19:25 multilingual

本文共521个字,阅读需6分钟

阅读模式 切换至中文

In a new career shift, Lady Gaga, née Stefani Germanotta, has announced that she will be taking on the role of chief technology officer (CTO) of one of the language industry’s top providers of machine translation (MT). The former singer-actress said the shift “came from a place of deep boredom after a decade-plus of trying to out-weird other artists.” After making a household name for herself by courting controversy in various forms — for example, by wearing a dress made of raw beef to the 2010 MTV Video Music Awards — Gaga said she “is tired of competing with the way these younger singers like Cardi B and Lil Nas X are trolling everyone.” For the 11-time Grammy and one-time Oscar winner, the obvious move was to computational linguistics. “I write songs, and that’s almost the same thing,” said Gaga. “Besides, tech leadership is really about how big your imagination is. If you can imagine it, someone else can do the hard work of actually making it happen.” Gaga has always been interested in the marriage of man and machine necessary for this kind of work. “You can see it in all my music videos. The humor error, the glitchy stuff, the iconoclasm of futuristic tech. That’s the monster we all fear and long for. Going beyond art to actually making it happen in real life, that’s a natural progression for me.” Gaga played coy as to which MT provider was welcoming her. “They’re thrilled, but they want to make some internal changes before they announce it themselves. Just in case it backfires and I can’t actually deliver an innovative solution to zero-shot translation for all the world’s languages.” Traditionally, MT was trained by feeding an algorithm examples of how one language (say, English) had been translated by a human being into another specific language (say, French). Zero-shot translation is the ability of a system to translate one language into another without having specifically been trained to do so. This could be particularly difficult for languages that have very little training data available, such as Indigenous and minority languages, as well as languages with somewhat unique linguistic features. Gaga isn’t worried, however. “Most people would say MT for Finnish to Navajo is a bad romance, but I think it’s a match made in heaven.” In her new job, Gaga has no intention of compromising her cutting-edge approach. “They’re going to be surprised,” she stated. “If I can become a rock star and a living legend — which is the most competitive job on the planet — I can certainly figure out how to automatically and accurately translate any language into any other language.” The key, she hints, lies in her musical background. “You code so much unspoken data in music. Emotion. Innuendo. The audience gets it in an instant. It surpasses language and becomes its own thing. Machine translation just needs a music guru.” By the time she’s done, says Gaga, Trados will accept “rah rah ooh ma ma, ga ga oo la la” in its termbase.
Lady Gaga,Née Stefani Germanotta,宣布将担任语言行业顶级机器翻译(MT)提供商之一的首席技术官(CTO)。 这位前歌手兼演员说,这一转变“来自于十多年来试图超越其他艺人的极度厌倦”。在以各种形式引起争议--比如在2010年MTV音乐录影带颁奖礼上穿着生牛肉做的裙子--使自己家喻户晓之后,Gaga说,她“厌倦了与Cardi B和Lil Nas X等年轻歌手的方式竞争。” 对于这位11次格莱美奖和一次奥斯卡奖得主来说,显而易见的举措是转向了计算语言学。“我写歌,这几乎是一回事,”Gaga说。“此外,科技领导力实际上取决于你的想象力有多大。如果你能想象到,别人也能做真正实现它的艰苦工作,“ Gaga一直对这类工作所必需的人机联姻感兴趣。“你可以在我所有的音乐录影带里看到它。幽默的错误,令人毛骨悚然的东西,对未来科技的破坏。那是我们都害怕和渴望的怪物。对我来说,这是一个自然而然的过程,超越了艺术的范畴,让它发生在现实生活中。“ 对于哪个MT提供商欢迎她,Gaga表现得含糊其辞。“他们很兴奋,但是他们想在宣布之前做一些内部的改变。以防万一事与愿违,我无法为世界上所有语言提供一个创新的零点翻译解决方案,“ 传统上,MT是通过给算法输入一种语言(比如英语)如何被人类翻译成另一种特定语言(比如法语)的例子来训练的。零点翻译是指系统无需经过专门训练就能将一种语言翻译成另一种语言的能力。对于培训数据很少的语言,例如土著语言和少数民族语言,以及具有某种独特语言特征的语言来说,这可能特别困难。不过,Gaga并不担心。“大多数人会说MT从芬兰语到纳瓦霍语是一段糟糕的爱情,但我认为这是天作之合。” 在新的工作岗位上,Gaga无意妥协她的尖端做法。“他们会很惊讶的,”她说。“如果我能成为摇滚明星和活着的传奇人物--这是这个星球上竞争最激烈的工作--我一定能琢磨出如何自动准确地将任何一种语言翻译成其他任何一种语言。” 她暗示,关键在于她的音乐背景。“你在音乐中编码了那么多未说出的数据。感情。含沙射影。观众一瞬间就得到了。它超越了语言,成为自己的东西。机器翻译只是需要一个音乐大师,“ Gaga说,到她完成任务的时候,Trados将会接受“rah rah ooh ma ma,ga ga ooh la la”这个词。

以上中文文本为机器翻译,存在不同程度偏差和错误,请理解并参考英文原文阅读。

阅读原文