Humankind is finally embarking on a magical story of digital creation, but will it ruin literature? Or will the master-slave relationship between humans and machines ultimately be reversed in a single stroke?
Which industries will soon be completely disrupted by artificial intelligence, commonly called robots?
From Deep Blue, which defeated the world chess champion Gary Kasparov, to AlphaGo, the “master” version of which recently and handily slaughtered the best go players, a new lay of the land has become a foregone conclusion. The rules of the game have been completely rewritten: sure, games like chess and go can continue to exist, but professional competitions will no longer represent the highest level of play. The experts will now be demoted to amateur status; their play is a healthy diversion, dancing in the plaza is for old ladies. The most brilliant games will be reserved for the robots and the teams of researchers who program them – people who may never have played chess in their lives.
Translation is another field that appears to be subject to imminent disruption. The first translation machines were nothing amazing, and the awkward, patchwork translations they produced resembled homework assignments by slacker students. But even so, I’ve felt obliged – but never dared – to go to the foreign language institutes and announce: the good times won’t last much longer. In late 2016, Google introduced a new generation of machine translation with a neural network algorithm that improved error rates by sixty percent. Microsoft and other firms are hot in pursuit with related research and development, leading many scientists to predict that one of the five most anticipated technological achievements of 2017 will be that “it is no longer necessary to study foreign languages” (as printed in Russia’s Komsomolskaya Pravda, 28 December 2016). The state of affairs seems to be that, aside from literary translation, which is somewhat thorny, all business, government, media, and travel organizations will soon be using phone apps to handle ordinary foreign language work in spoken and written forms. Who will still need to employ professional translators?
What about education and medicine? And accounting, law, advertising, finance, quality assurance, engineering, the stock market--what about those industries?
The American scholar Kevin Kelly is among the optimists. Citing massive non-profit, communal achievements such as Wikipedia, he looks forward to a “digital socialism”. Jack Ma of the Alibaba Group also believes that “big data can revive the planned economy concept”. But what they haven’t mentioned is how robots are already wiping out large swaths of white- and blue-collar jobs. The arrival of big data and cloud computing has led to huge breakthroughs in the capacities of robots in terms of recognition, memory, research, computation, planning, and learning. In these realms, they have already far outpaced the vast majority of humans, and they are model workers in terms of precision, durability, and other advantages. These new comrades also come with noble hearts of silicon: ATMs are never corrupt, face-scanners never slack off, self-driving systems never clamour for a raise, and the claims-processing machines at insurance companies and the news-writing machines at news companies never get tired, unless you cut off their electricity.
Some have boldly predicted that ninety-nine percent of human mental labour will be replaced by artificial intelligence (Global Times, 6 January 2017), and even the most conservative estimates are above forty-five percent. These projections do not exactly sound like good news. The Israeli scholar Yuval Noah Harari recently predicted that the vast majority of people will eventually be reduced to a “global useless class”, and he suggests that, when the stratification of organisms through gene technology is also considered, “we may be preparing to create a maximally unequal society”! (Harari, Sapiens: A Brief History of Humankind, and Homo Deus: A Brief History of Tomorrow) Indeed, hints of the future already begun to appear. After the “dark factory” – robots do not need light to work – the next step will be the “dark office”. And when even shopkeepers and peddlers are replaced by vending machines, and re-employment roles like sanitation work and security are “darkened” by robots, then what will the huge army of unemployed people do? Will they go tanning, play mah-jongg, run marathons, or travel around the world? As soon as the employment crisis reaches ninety-nine percent of working-age people, or even half that many, there is certain to be a complete collapse of economic life. In these circumstances, the endless vacation days will feel more like Judgment Day, and forget about socialism – none of the ‘isms’ will provide any succour. What kind of political or societal structure will resist disintegration? Digital socialism may actually be digital oligarchy – but let’s put that aside for the moment and address it again later.
As a literature enthusiast, I cannot help but imagine what will happen to literature. It may not be the most important matter, but it will affect a great many people who work in the humanities, as well as the audience for literature.
We might as well first consider two poems:
From the western window comes the sounds of the tides,
A nimble ship on a long journey on the seas.
At dusk, a misty rain approaches through the autumn air,
Alone, I drunkenly dream of white clouds.
The crescent moon hangs above the western river bank,
The damp mists rise to meet the sky.
The island and the white sand become one in the darkness,
The only light comes from the fishing boats.
These two poems were written by the Song Dynasty poet Qin Guan (秦观) and by Oude (偶得), an app designed by IBM, respectively. The question is, how many people could determine at a glance which poem was written by “him”, and which one, by “it”? When I applied this test to graduate students of literature at a certain university, these experienced readers and expert appreciators were left scratching their heads in doubt. When I cleared the screen and mixed the human poem in with several others by the Oude app, even fewer students were able to pick out the work of Master Qin of the Wanyue School (婉约派).
"Master Oude" is just a plaything with a very ordinary database and set of algorithms, but it nonetheless is capable of confounding the experts to a certain degree. It is also far superior to a human poet in both speed and breadth of subject material, striking a profound blow to the self-respect of many poets. The poems flow forth with ease, all of them worthy, and for Oude, it’s no big deal. Even nonsense is no match for Oude: enter in the Chinese characters for nonsense (胡说八道), or even put them in backwards (道八说胡), and Oude will still produce a steady stream of clever, unexpected verses of great variety, transforming four meaningless characters into one hundred elegant styles: “Fruit pits (胡儿) should not be left beside the flowers, for it’s said (说) that they’re better than orchid sprouts for planting lotuses. The August (八月) evening light reflects in the wine, and the sages (道人) inadvertently appear like spring mists.” Or: “The sages (道人) open their eyes and come forth from the mountains, their hair as white as eighty-year-olds (八十). Say (说) they’re older than the fishers and the woodcutters, so why (胡为) do they leave me to take watch at the railings?” These high-production batches of eloquence are indeed absurd, but are the fashionably dressed celebrities among us any better at erudition? Compare the works of Oude to the present prevalence in pop music of lyrics that imitate the classics, including the styles of Tang and Song poetry, and sell out the national culture. These lyrics, crooned forth by pop stars until whole arenas are rocking with classical pastiche, hardly seem superior to the works of Oude. Meanwhile, parrot-like imitations of slogans, political jargon, it-girl slang or old-cadre lingo, and chicken-soup platitudes pervade our newspapers. And the poems derived from a quick glance at Liweng Duiyun (笠翁对韵, a Qing Dynasty primer) that end up under discussion at the Writer’s Associations – are they any better than Oude’s “nonsense”?
Poetry isn’t the only field under assault. Novels, essays, criticism, and film and television scripts are also facing savage threats from robots. In the 1960s, Bell Laboratories in the United States was already experimenting with robot writers. In the decades since, thanks to the internet and big data, this ambitious search has come a long way: the caterpillars are finally becoming butterflies. In May 2016, Japan’s TV Asahi reported that a novel written by an artificial intelligence program created by a team of students at Future University Hakodate had made the short list for a literary prize in a field of 1450 submissions, much to the shock of readers. To say that the story had been written solely by a robot is not exactly true. The relevant code in the program was written by humans, and the details, plots, dialogues, roles, descriptions of setting, and other “components” in its database were also created and stockpiled by humans. What the robot had to do was simply follow orders to conduct a series of tasks: sorting, composing, deducing, grammatical testing, and polishing. Nonetheless, this victory of machine over man feels like day one of a literary revolution. The first step has been taken, and now, as the algorithms continue to develop and the datasets continue to expand, the flourishing of a robot literary culture seems inevitable and imminent. The robots have arrived, the experts have gathered, and I fear that sooner or later they will be seized by the urge to establish their own writer’s association, issue regulations, and nominate a chairperson.
When that day comes, perhaps readers will merely have to enter their order into a dialogue box:
Male Protagonist: Stylish Gentleman. Female Protagonist: Wild Girl. Supporting Roles: No Preference. Genre: Love / Suspense. Setting: Island / City. Primary Mood: Anxious. Religious Taboos: None. Main Story: Beloved Dog / Leukaemia / Meteorite Strikes the Earth. Tone: No Preference...
And so on.
And immediately afterwards, our reader receives an orderly, vivid, and dramatic story, or even a series. The author may be a robot, a person, or some crude mixture of AI and HI (human intelligence), much like the “writing software” that is already available on Taobao for fifteen yuan, usually little more than low-price plagiarism assistants. These programs have already become the second half – perhaps the better half – of many internet writers. It’s not impossible to imagine that a certain well-known and popular literary giant, perhaps the award-winning Mr. Ma or Ms. Helen or some fellow who has given many acceptance speeches and made many generous donations, will years later be exposed in one fell stroke as not a human but rather a silicon chip, a hard disk, and an internet connection: a virus-like electronic ghost.
When the French intellectual Roland Barthes published The Death of the Author in 1968, he seems to have hinted at the current crisis. But when the death of the author ultimately arrives, what will she be replaced by? Will the next Qu Yuans (屈原), Du Fus (杜甫), Shakespeares, Tolstoys, Cao Xueqins (曹雪芹), and Kafkas be mass-produced in some place like Silicon Valley or Zhongguancun, leaving people overwhelmed and disgusted by more literature than they can digest? Or will humanities workers gradually be defeated and dispersed by the tyranny of technology? Will all the nascent Qu Yuans, Du Fus, Shakespeares, Tolstoys, Cao Xueqins, and Kafkas come to a premature end, driven insane or smothered to death by robots?
Perhaps this what the techno-utopianists already have in mind.