Reflections on the newsworthiness of machine translation
|Jeffrey Ding||Apr 5||2|
Greetings from a world where…
ChinAI is now three years old, which means it should be able to walk in a straight line, but it rarely does. Thanks to all the readers, especially the 170+ paying subscribers who support ChinAI under a Guardian/Wikipedia-style tipping model. If you’ve been meaning to subscribe but just haven’t gotten around to it, no time like the present.
As always, the searchable archive of all past issues is here.
Reflections - Making Translation Newsworthy
What is newsworthy? This question should haunt everyone with a platform.
Last month, Stanford HAI published the AI Index Report 2021, a 222-page report on the state of AI, put together by an all-star team supported by a lot of data and strong connections to technical experts. What was newsworthy in this report? According to The Verge, “Artificial intelligence research continues to grow as China overtakes U.S. in AI journal citations.” In fact, the article takes its cue from what the report authors themselves deemed significant, given that “China overtakes the U.S. in AI journal citations” features as one of the report’s eight key takeaways.
Dig deeper into the data, however, and you’ll uncover alternative takeaways. Look at the cross-national statistics on average field-weighted citation impact (FWCI) of AI authors, for example, which gives a sense of the quality of the average AI publication from a region. Interestingly enough, the U.S. actually increased its relative lead in FWCI over China over the past couple years. According to the 2019 version of the AI Index, the FWCI of US publications was about 1.5 times greater than China’s; in 2021, that gap has widened to almost 2 times greater (p. 24).
So, working off the same materials as released in the AI index, here’s another way one could have distilled key takeaways: “The U.S. increases its lead over China in average impact of AI publications.” Or, if you wanted to be cheeky: “China lags behind Turkey in average impact of AI publications.” Just as newsworthy, in my opinion.
However, what I found most newsworthy about the AI Index went beyond horserace reporting about “who’s winning the AI race?!”Instead, I was most intrigued by the rise of commercially available machine translation (MT) systems, covered on page 64. According to data from Intento, a startup that assesses MT services, there are now 28 cloud MT systems with pre-trained models that are commercially available — an increase from just 8 in 2017. But wait … there’s more: Intento also reports an incredible spike in MT language coverage, with 16,000+ language pairs supported by at least one MT provider (slide 33 of Intento’s “State of Machine Translation” report).
Thanks to the State of MT report, we can also get a sense of the greater significance of translation advances. Let’s take Microsoft’s neural machine translation (NMT) service, priced at $10 per million symbols, for a ride. Recall that Microsoft achieved a milestone in 2018 for the first MT system reach human parity for translating news articles from Chinese to English. For $10 we can translate about 200,000 words, using an average of 4.79 symbols per word. How does one even grasp the potential of 200,000 words? That’s four times the number of Chinese characters in Wang Shuo’s great novella <<动物凶猛>>, one of so many Chinese language classics that have not yet been translated to English. These are, of course, very rough calculations — NMT works much better for news than novels, and the process would require a lot of post-editing — but they do open a window into the possibilities of NMT.
Somehow, these incredible advances in translation are not relevant to the effect of AI on U.S.-China relations, at least based on existing discussions. Compare the complete dearth of Twitter discussions centered on the following keywords: U.S., China, and “machine translation” against what you get when you replace “machine translation” with “facial recognition.” Consider another reference point, the recently published 756-page report by the National Security Commission on Artificial Intelligence (NSCAI). Sixty-two of those pages mention the word “weapon” at least once. Only nine pages mention the word “translation,” and most do not substantively discuss translation (e.g. the word appears in a bibliographic reference for a translated text).
Yet, I could make a convincing case that translation is more significant than targeting for U.S. national security. Think about the potential of improved translation capabilities for the intelligence community. Another obvious vector is the effect of translation on diplomacy. To the NSCAI’s credit, it notes how advances in MT could “transform the way we communicate across geographic and cultural barriers, enabling business, diplomacy, and free exchange of ideas” (p. 36). Consider also how MT could increase economic interdependence, which could have a pacifying effect on conflict escalation. Professor Steve Weber explores this in “The 2020s political economy of machine translation,” a working paper which analogizes MT to the container shipping revolution. Finally, recall the possibility of more translated books, which could also have security implications. After all, empirical studies have found that students have reduced threat perceptions of their host country after studying abroad, and what is reading, if not a study abroad experience.
So, how will AI advances affect U.S. national security and U.S.-China relations? Are the U.S. and China DECOUPLING in strategic technologies like AI? The maddeningly beautiful thing about studying AI + politics is that no one really knows. If you take translation as the key frame of reference for AI as opposed to weapons targeting or facial recognition, the answer to these questions is very different (#coupledforlife). This is made all the more complicated by the fact that AI is a general-purpose technology, with potential applications that far exceed translation, facial recognition, and weapons targeting. So, if you really wanted the true answer to the above questions, you would have to somehow predict the net impact of the distribution of all still-evolving AI applications.
Our answers to these questions and our determination of what is newsworthy about China’s development of AI, therefore, say more about us — our biases, our interests, our social influences — than they do about the Truth of the matter. I am no exception. Unlike journalists, translators do not have to pretend to be neutral. ChinAI is a product of my own biases, interests, and social influences.
That includes my fears. I vigorously pushback against overhyping China’s technological prowess in part because I fear how threat inflation could affect Asian Americans like me. I am quick to rebut claims of decoupling in part because I worry about what it means for my family’s ability to visit loved ones in China.
Newsworthiness doesn’t just have to be driven by our fears. And so, ChinAI is also a product of my dreams. Here’s one: One day, my kids will learn to read Wang Shuo’s 动物凶猛 in Spanish.
ChinAI Links (Four to Forward)
My four favorite issues of ChinAI from year 3 were all collaborations with contributors:
ChinAI #98: Techlore - The Historical Rise of TSMC & Samsung in Semiconductors — Joy Dantong Ma found an epic poem of sorts about the history of Taiwan Semiconductor Manufacturing Company.
ChinAI #104: Tencent 2020 AI White Paper — Caroline Meinhardt, a GovAI summer fellow, helped shed light on how Tencent is thinking about AI applications, including deepfakes.
ChinAI #112: The Human Cost of 30-min Food Deliveries — Gabriele, San, and many other anonymous contributors helped translate a 15,000-word longform article on how drivers get trapped by food delivery system algorithms
ChinAI #129: An Emotional Mess — Shazeda Ahmed guest-edited an issue about emotion recognition in China, featuring a translated article about backlash to emotion recognition applications in classrooms and her report (co-written with Vidushi Marda) on China’s emotion recognition landscape
Thank you for reading and engaging.
These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is a PhD candidate in International Relations at the University of Oxford and a Predoctoral Fellow at Stanford’s Center for International Security and Cooperation, sponsored by Stanford’s Institute for Human-Centered Artificial Intelligence.
Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).
Any suggestions or feedback? Let me know at email@example.com or on Twitter at @jjding99