This episode was prepared in late November/early December 2022. I learned in mid-December that Google Play Books now allows for multiple narrators in their AI narration tool. (The tool lets you set a “Default narrator” and “Additional characters,” with different voices.) Please keep that in mind as you listen to this episode.
To worry about the demise of human translation due to AI is to both underestimate and overestimate translations. And, perhaps, by extension, to worry about the demise of human narration due to AI is to both underestimate and overestimate narrations. These are some things I absorbed from the recent developments in AI translation and narration.
And by “recent,” I mean the year 2022, in general. But, here is a date that is more specific. In May 2022, Google Books came out with a tool that writers could use to have their ebooks on Google Books narrated by AIs. I have tried using this tool, although I have not published with this tool. At this time, all my books are fiction, and the tool is not yet suited for fiction narration. However, it was and is amazingly, oh, amazingly impressive for nonfiction narration where, within the text itself, there is only one character, the narrator of the nonfiction story.
That said, Google is absolutely going to add to their AI the ability to act different voices. So, I can imagine a near future where for fiction books, the writer will be able to tag each character with some sort of metadata, say, “Character A, female, age 20, American English accent,” and then the AI will act that.
It might even soon be possible for specific contexts to be edited, to add nuance. Maybe Character A is sick, or she is excited, or she is sad. Even more impressively, perhaps AIs will soon become more like actual AIs, as in, artificial beings with something resembling intelligence, and be able to read the context. As in, maybe the AI will be able to “understand,” in air quotes, the context of a story and act accordingly. Now, that would be something.
But such things aren’t possible, yet. On the translation end, there is DeepL. I think that’s how they say their name. Deep-L. But it might be DeepL. I don’t know.
Anyway, DeepL has been mentioned in various circles as the company that offers better translations than Google Translate, because, let’s be honest. Google Translate sucks, especially for language pairs that aren’t very popular. Google Translate for Korean to English is outright laughable. It’s like vomit. English to Korean is slightly better, but still in no shape or form entirely reliable.
DeepL does not have the Korean English language pair yet, at the time of this recording, which is December 2022. But I have tried it for English to German some time ago. My German speaking and writing skills are poor, but I have lived in Germany for six years as a child and I can read and understand—at least enough to know that a translation is completely wrong. So, although I am not equipped to be a German translator, I am in a position to see if a translation software completely sucks or is quite great. DeepL’s software is quite great. However, it cannot read the context.
There was a specific sentence in one of my stories, titled, “Because They Don’t Bleed Red”—a dystopian science fiction story with bad mad androids and such. The sentence was, quote, “If Jackson hadn’t installed extra purchase features on my soles, I would have flung myself out of my path at some curve back there.” End quote. The problem word for DeepL’s translating software was the word “purchase.”
According to Google, which pulls its dictionary definitions from Oxford Languages, there are two main definitions for the word “purchase.” One: “the action of buying something.” Two: ”a hold or position on something for applying power advantageously, or the advantage gained by such application.” For this second definition, the example sentence is this. Quote, “the horse’s hooves fought for purchase on the slippery pavement.”
The way I used the word “purchase” in my sentence was this second definition. But DeepL translated it as the first definition. Basically, I was talking about how the soles of this character had extra friction/pressure of some kind, and DeepL translated it as the character buying soles.
Granted, the phrase “extra purchase features” isn’t very common. I’ve tried searching Google, using quotation marks around the phrase “extra purchase features” so that it can phrase-match the results. And those results showed, one, e-commerce sites with the phrase “extra purchase” and then, in the next sentence, starting with “features.” Or, there was one result where the page was indeed talking about “extra purchase features” in one sentence, but in that case, what they meant was “extra” as in, outside. Outside of the purchase. Additional to the purchase.
So, it’s not surprising that DeepL’s software translated the word “purchase” as relating to the act of buying, except for the word “soles.” Again, the sentence was, quote, “If Jackson hadn’t installed extra purchase features on my soles, I would have flung myself out of my path at some curve back there.” There is a clear context here, which all the AI translation software that I know of and have tried cannot translate yet, and that applies to DeepL, which is in other ways, oh, much more superb than Google Translate, the last time I checked English to German translations.
All this said, I am a professional translator for the English to Korean pair. I translate English to Korean under my Korean pen name, and also under my legal name. The stuff I do under my legal name is bound by NDA and I cannot discuss it anyway, so there is no point telling you my legal name to see if I am qualified. You can look up my pen name translation though.
I also translate my own writing from English to Korean, and recently have started to do Korean to English as well.
So. Based on all this. Today’s theme, reiterated: To worry about the demise of human translation due to AI is to both underestimate and overestimate translations. And, perhaps, by extension, to worry about the demise of human narration due to AI is to both underestimate and overestimate narrations.
I’ll mainly talk about translations by humans and AIs, instead of narration, because I don’t have personal experience in the narration area.
So. Translations. Translators. Human translators. Will they, will we—including myself—all, or mostly, be replaced by machines? The short answer, in my opinion, is no. And the reason is that I neither overestimate nor underestimate translation, again, in my opinion.
I’ll talk about the underestimating side first. I find that sometimes, the greater world underestimates what it takes to do translations. And I’m sure this underestimation happens in other fields, too, but the reason the underestimation in translation stands out to me is that, one, I translate, so I am biased, but more importantly, two, it has been mentioned so frequently as the thing that is going to be replaced by machines, that sometimes it seems like people actually just believe the reporters or whoever that claim that human translation is going to be replaced by AIs.
Let me tell you why I think that day won’t come. The short reason is that translation, more than original storytelling, is both art and labor, in my opinion.
Many fields are both art and labor—in fact, off the top of my head, I cannot think of a field that isn’t both art and labor. If you’re an athlete, a lot of your practicing in your chosen game will be labor in that you must wake up early in the morning, repeat the grueling routines, start really early—as in, start as a little kid, basically—and keep at it until you’re, I don’t know. Thirty-ish. Or forty-ish, depending on the game.
But anyone who has cried while watching a great game, of any kind, will know: sports are art. And I cry very easily while watching games, even though I don’t really follow sports. Actually I cry easily while watching anything, such as movies. It’s not like… it’s not deep crying, necessarily, but it’s at least surface-level crying. I guess I am moved easily, in that regard.
Anyway, sports are art and labor. So are many things, ranging from running a business to collecting garbage. I mean, just observe the garbage collectors in your neighborhood. The variety in the degree of dedication they have to their work is just the same as the variety in the degree of dedication in any other field. I refuse to believe that garbage collecting is a field where anybody can go in and perform equally as anybody else. Some garbage collectors in my neighborhood are superb. Others aren’t. The former group makes garbage collecting art. Seriously. The way they put the garbage bins back in place, the way they move with efficiency and care, is amazing. You can’t compare them with some other garbage collectors who will, literally, leave garbage lying around in their wake. Seriously. Some of them do.
So. Anyway. Many fields are both art and labor, and so it is with translation.
First, labor. It is labor, in that literally, people get sick from too much work. No kidding. People get RSI. Repetitive strain injury. And I want to do an episode about RSI at some point, so I won’t get too much into that topic, but basically, repeated typing has adverse effects on your fingers, wrists, back, shoulders, neck, and, worst-case, head, too. You get constant headaches. I have known translators who have had to take a break from work completely, for several weeks, to recover from their injuries.
Just in general, folks, I think more people should know that office work is no walk in the park. Just because you sit in a cooled office on a hot day or in a heated office on a cold day doesn’t mean that you have it easy. On the contrary. The generic office furniture and gear, such as keyboards, that are distributed to office workers often don’t suit their individual size, posture, and work style. Repeat the use of that furniture and gear for more than eight hours a day, for at least five days a week, for decades—people do get hurt. And often, the adverse effects are so incremental, so small, that it’s too late by the time they notice they’re injured.
Anyway. More on that on the RSI episode, perhaps, in the future.
The point is that translation, like many types of office work, is labor. It is true labor. You stare at the screen you type constantly, you remain seated for hours on end, if you aren’t aware of that and try to stand up every so often. I think some folks underestimate this aspect of translation—just how much grunt work it is.
And so, personally, I dream of AI translations. Oh, I do. I want AI translations. If there is any way in which I can reduce the amount of unnecessary clicking and typing, I will pay for an AI translation software. I will gladly do so, the same way I pay for Microsoft Word and Scrivener and Vellum so I don’t have to handwrite my stories and hand-bind them and then drive to a dozen bookstores to have the books distributed, or some such thing.
And one of the reasons I can dream of AI translations and still feel, I guess, secure in my worth, is that, at the same time translations are labor, they are also art.
Let’s think about this. Why is it that some translators have so much work that they get sick from working? It’s because translation is art. See, sometimes the logic behind people fearing the rise of AI translations is that jobs will be taken away from the workforce. But that word. That generic word, workforce. That is what I question.
I guess, for the purposes of bureaucracy, there is some need to count all the workers in a field. Or to get the average of how much such workers earn. Or how long they stay in the workforce. All those things. But practically speaking, for each of us actually in the workforce? We aren’t the workforce. We, the individuals, aren’t the workforce.
So, when people worry about the translation workforce, because of the rise of AIs. And they use rhetoric that implies that somehow, if only the AIs didn’t exist, the general human workforce would all have jobs, or more jobs. I wonder. I wonder how they actually go about things when they hire a translator, or anybody.
Do you, when you hire anybody, go to the person who has no work, in order to distribute work evenly to them? Maybe that is your idea of fairness? Maybe that is indeed the case in a socialist or communist society. I don’t know.
But in the reality that I know? The person who gets hired more easily and more often isn’t the one who needs the job. In fact, the person who doesn’t need to need the job but instead, purely wants to want the job, sometimes has this… halo effect. Like, because they don’t need it, they are more desired. I guess, in that regard, the work market is like the love market. You know, there is a somewhat joke-like saying that if you’re in demand, romance-wise, you will keep being in demand? And when you’re not in demand, like, if nobody dates you… then you will keep on being undatable? And in the workforce, the same: there are numerous, countless articles online, about how the best time to look for a new job is when you have a current job.
So, see, I don’t think this perception thing can be erased. The perception that those who are in demand are in demand because they are qualified will not go away. And I haven’t seen it going away in the field of translation. The translators who get sick from overwork get sick because they have work to get overworked by. And the reason they have so much work might be that they are actually qualified, or that they seem qualified. It doesn’t really matter, because the effect is the same, in this case. They keep getting work.
And there are cases in which the distinction matters—because, as I said, translation is an art form. No two people translate the same way, just as no two people talk original sentences the same way. Saying that translators are replaceable by other humans or machines is like saying that, AIs can replace humanity altogether.
And, okay, some people do believe this—that AIs will replace humanity altogether. But I personally do not believe this, because I think every person, is in one way or another, art. I do not think that a human individual can be averaged out. Again, bureaucracies will attempt to do that for statistical purposes, but at the practical level? Are you average because someone used you as a number in statistics? Are you average to you? Are you average to your mother, father, sister, brother, son, daughter, whoever else?
Or, if you aren’t average, if you are taller than average, shorter than average, heavier, leaner, darker-skinned, lighter-skinned, more educated, less educated, so on and so forth—is that why you are you? Does your lover love you because you’re shorter than average? Because you can be placed on some random graph with an x-axis and a y-axis and they’re into people who fall into that category? I mean, really?
See, when there is worry about humans being replaced by AIs. Or translators being replaced by AIs, I wonder what kind of love they have in their lives. Personally, I do not worry that the love that someone has for me will be replaced by AIs or other people. I’m just gonna say something that sounds really delusional. I am an artwork. Yes, I am. But so are you. Each of you listening to this or reading this on the transcript page. Nobody can replace you, so long as you don’t let anybody or anything. No other human can replace you and no AI. This is so for work, for love, for life. You are art.
And so are translations, because they come from translators who are aware, to varying degrees, that they are art. No two translations are the same. And no end recipient of the translations are the same. Some people will like certain translations by certain translators better than other translations by other translators.
Art is a matter of taste. When someone falls in love with you, hopefully it’s not because you meet their required specs. Hopefully it’s because they just love you because you… I guess… you taste good. For their tastes.
And, yes, there are some aspects of translations that are so repetitive and predictable that they aren’t art. In that case, well. Let the AIs replace those things. To cling to such work is to be a 21st-century Luddite. Why would I want to work on manual, repetitive translations? That will only exacerbate the labor aspect of translating. That type of work will literally make me sick.
Since the early days of the Industrial Revolution, there have been people who opposed cars, washing machines, microwaves, and whatever else. Thank gods that despite their opposition, I can use cars, washing machines, and microwaves. I would not have been grateful, had they won.
So. Yeah. That is the end of this segment of this episode. That translation, like many other fields, is both art and labor. And thus, worrying about human translators being replaced by AIs is to underestimate translating. Either there is nothing to worry about because the labor aspect should be replaced by machines to prevent human injuries, or there’s nothing to worry about because the art aspect cannot be replaced by machines anyway. Because, art is love. And love is not about averages. Love isn’t even where you are, relative to averages. Love just is.
Now, let’s get to the part of why, to worry about the demise of human translation due to AI, is to overestimate translations.
My theory is that this type of worry comes from the wrong premise that, so long as translations are in place, people will be able to understand each other. Meaning, a lot of importance is put into a translation existing or not. Because it is presumed that the translation will solve a communication problem.
At the basic level, it does solve a communication problem. A person who doesn’t speak Korean will be able to “understand,” in air quotes, what the Korean writer said, if those words were to be translated to, say, English.
However, what does it mean to “understand” something? Or someone? The word understand is very difficult to understand. It is very difficult to define. People have different understandings about understanding. There are dictionary definitions, but personally, they seem so futile that I won’t quote one here. Do you understand what I’m saying?
I’m saying, it takes a lot more than translations to lead to an understanding, if ever.
Let’s consider situations in which translations aren’t required, in the superficial sense of the word translation. If you speak Chinese and you meet another person who speaks Chinese. Then you don’t need an English-to-Chinese translator or Chinese-to-English translator. Or Chinese-to-any-other-language translator. But even in such a situation, are you really, truly, in the most fundamental sense of the word “translation,” not translating?
I find that difficult to believe. Within a society where everyone is said to be using the same language, are people really using the same language? If so, why do you feel that some people understand you better than others?
I am making a grand assumption here: that you, like me, do feel understood by some and not understood by others. I find it impossible to imagine that you or anybody else can feel understood equally by everyone they encounter. If you do feel that way, I am inclined to suspect that you’re a robot.
And this is why, to worry about AI translation replacing human translators, is to overestimate translations. Translations cannot solve all the problems. Translations are merely the beginning.
There used to be a time when the basic level of translation was in shortage. There used to be a time when the absolute amount of translated material wasn’t sufficient. But we don’t live in those times anymore, depending on which language pair you’re looking for. Millions of audiovisual works have been translated into various languages.
These days, it is very easy to access foreign language material that has been translated. This is especially the case for audiovisual materials, because with audiovisuals, well, there are audiovisuals. Audiovisuals do not solely consist of words, so I believe translated works are more approachable than works that do solely consist of words, such as books. You can read the body language, you can read the facial expressions. You can hear the feeling in the voices, without understanding a single word. And so, there is a huge market for translated audiovisuals. You can go to Youtube and turn on the subtitles, or you can go to Netflix and turn on the subtitles.
Now, translations in the basic sense of the word aren’t the barrier to understanding. Now, it’s the actual culture.
And much of culture cannot be translated into words. A very easy example is the Korean word, unnie (언니). Unnie is a word females use for older females, but not much older. You wouldn’t call your grandmother unnie. You will call your older sister unnie, if you’re female. But also, you wouldn’t call your boss unnie. You have to be close enough to call someone unnie.
At the same time, the word unnie used to be used among males, in the past, which is amusing, considering how its meaning changed over time. And just the very existence of the word specifically for someone older than you. What does that say about that culture?
Every time I need to translate the English word “sister” to Korean, I have to do extra research to find out if they’re talking about an older sister or a younger sister. Clearly, to English speakers, the distinction matters less than to Korean speakers. There is no need to figure out age, in English, oftentimes. But in Korean, oh. Absolutely. The first thing kindergarteners do when they meet is to figure out each other’s age. Sometimes, even before they exchange names. Because, if you’re a five-year-old and there’s a seven-year-old, you’re likely not going to need to call that seven-year-old by name. You’ll call them “hey, older sister!” or “hey, older brother!” Yeah.
And every time I need to translate “unnie” into English, I need to make a decision. Do I keep “unnie” as unnie, or do I translate it to older sister or younger sister or just sister? Or something else?
These days, I think the go-to method would be to leave “unnie” as “unnie.” I think people who want to read works translated from Korean tend to be aware of the word “unnie.” See what I’m saying here? “People who want to read works translated from Korean.” The existence of a translation does not make people who aren’t interested in Korean culture, in one way or another, want to read a Korean work. Even translated in a different language, it still will have some sort of Korean-culture-specific things. The author doesn’t exist in a vaccum. The Koreaness might be stereotypical-Korean or counter-Korean. It will still draw from that root, even when it’s against it. And add to that the fact that some readers shiver at the thought of reading anything where the names of the characters aren’t Bob and Susie—eh, no offense intended, Bob and Susie, if you’re listening—and the role of translation at the superficial level becomes even smaller.
I’ve read a book review the other day, about a book I really enjoyed. The reader found the book difficult to understand, and literally, the reason they gave was that the names were foreign. It wasn’t about a Korean book, it was a book set in Kyrgyzstan. I hope I’m saying the name correctly. I mean, the English pronunciation isn’t going to be the correct pronunciation of the local language anyway, but still. The English title of this book is “A Killing Winter” by Tom Callaghan. And some people don’t like the book because, I guess the author dared to use local names and not names like Charles and Mary? For gods’ sake. Can you believe it?
See, this is one of the reasons I have a separate pen name for my Korean stuff and English stuff. I’m trying to remove a barrier that I know exists. I have encountered this countless times. The situation is different from my previous life, when I wasn’t writing. I have been stubborn and have never used an English name to replace my legal, Korean name, when I wasn’t writing. Because, I wasn’t dealing with random strangers. I shortened my Korean name, though, to one, one syllable. If anyone at the office is too stupid or too lazy to remember a single-syllable name, which I don’t even expect them to pronounce correctly, then what the fuck. But random strangers? I don’t demand that they put in a lot of energy to try to understand my name or even remember it. Hence an English-sounding pen name, and hence a Korean pen name.
Ithaka isn’t exactly English, I mean, it’s a Greek island. But I haven’t seen anybody who had difficulty pronouncing it. There is an American city named Ithaca, so.
But speaking of names. Many people in the Western world think that names ending in “a” are for females and names ending in “o” are for males. That is because that tends to be the case for Western names. But, consider Japanese names. Say, if your name is Tamako. It’s a female name. Whereas, Akira. It’s mostly a male name.
Anyway, the point being. Translations are useful. Very. They can be the starting point to understanding. But to think that, if a translation were to exist, people will actually understand each other at the fundamental level? That’s overestimating the role of translations. Even if a machine were able to do the superficial level of translation perfectly—if there is such a concept as perfection in a field of art—even so, there will be a lot of judgment calls. Someone will have to make those calls. Someone who is in a better position than a machine to know what it means to exist within a human culture.
About a decade ago I heard about a celebrity couple on TV. The couple came from different countries, different cultures. The wife was on a TV show. And she said something like, “I like it that it is a given that he doesn’t understand me.” She said that there is no expectation that her husband will fully understand her, because they don’t share a language, not fully. And as I was watching it, I was like… How… How is that good? Shouldn’t people who are close to each other understand each other?
But now, ten years later, I think I understand her better. I think what is the most terrifying is the assumption that you will understand each other because you think that being in some sort of bureaucratic category, together, will automatically provide understanding. Such as, just because you’re born in the same country, you’ll have the same feelings, the same logic, the same priorities. Or just because you share a language. Or just because you work in the same field. All those “just because”s.
Whereas she and her husband? They came from different worlds anyway. Anytime they found similarities, great. If they didn’t find similarities, well, that was expected. So, in a way, I think she was talking about how she could see him more as an individual, instead of a category. And perhaps he did the same for her.
Before we move on to the territory of narration, which will be largely hypothetical because I don’t have personal experience in the field, there is one element in translation that is also somewhat hypothetical, which I want to address first.
There are days when I wonder that, if AI translation software does not develop soon enough, some languages will die out. More specifically, if they do not develop soon enough, most of the content created in this world will either be in English and Chinese, and in no other language.
Now, of course there is nothing wrong with English and Chinese. But imagine a world where there is just English or Chinese content. And maybe also Spanish. Spanish is used by a lot of people. The exact number of languages and which languages there are aren’t the key here. The key point I’m making is that, imagine a world where only a few biggest languages survive, and all the other languages die. I think the absence of AI translations will make such a scenario more likely.
Even now. I would be lying if I said that the existence of a larger English-speaking market wasn’t one of the reasons I decided to write predominantly in English. The connections I can make because I write in English and also talk in English are much more numerous and frequent than the connections I can make because I write and talk in Korean. At the simplest level, it’s a numbers game. There are fewer Korean users in this world. Whereas, English is a language that people in many countries learn almost… involuntarily. They have to learn it for survival, in addition to their mother tongue.
But at least I kept my mother tongue. Some people do not. I know people who intentionally make their children forget the language of their parents, in what they call an attempt to make the children speak English better. This is sad.
See, English is or is perceived to be this… master language, by some people. To speak English is to be better, almost. I do not think that is the case, although, as I have admitted, the ability to speak English has opened many doors for me that would have remained closed otherwise. But any language opens extra doors. Yet, because languages that aren’t English or Chinese or Spanish—or any of the bigger languages—don’t open as many doors as the bigger languages, there are cases in which doors are intentionally kept shut. I guess it’s a way to focus? But I… Yeah, I find it sad.
To learn a new language is to learn about a whole other world. This learning process cannot be replaced by translations.
So, what I fear isn’t the advent of AI translations. What I fear is the opposite. I fear AI translations will not develop soon enough so that a significant enough number of people will be compelled to focus their resources on learning one of the bigger languages and abandoning the smaller languages. This would then become a cycle. Fewer people use the small languages, less content is created in those languages, fewer people get interested in the culture of those languages, and thus, even fewer people will use those small languages… and the cycle will repeat.
But. If AI translations were to finally get better. Then, maybe, people will learn other languages and more importantly, have time and resources to learn about other cultures. Because, the translations will exist so much more easily. And from there, they can learn about the people who use languages that aren’t their own. The purpose isn’t to learn the dictionary definitions, it’s to learn about the people.
Which finally brings us to the narration part of this episode. Perhaps, to worry about the demise of human narration due to AI is to both underestimate and overestimate narrations.
The reason I hypothesize this way is similar to my logic for translation. Narration cannot be underestimated because it is both art and labor. Also, narration cannot be overestimated because—this part I will explain in more detail in a few minutes.
Okay. Narration as art and labor. Yes. Labor. Wow. Recording anything, editing, that is a lot of work. Recording while acting, I imagine, is even more challenging physically and emotionally. The number of decisions that must go into deciding how to act—I imagine it is a lot of work.
But also, same as translation and many other fields, I think it is impossible to distribute narration work evenly among narrators, because some narrators will always be more popular than other narrators. It doesn’t matter how many human narrators there are and how many types of AI narration software there are.
More importantly, you know what happens for something like translations, and probably for something like narrations, too? When the buyer—the payer—sees the price as too high, what happens is that they hire no one. Yes, instead of going for the cheaper option, they hire no one.
For example, if a translator who is seen as qualified gets paid twenty apples per translated word, but the buyer cannot afford to pay twenty apples per translated word, they may go to a different translator who gets paid ten apples per translated word, instead. However, from my experience, it is unlikely that they will go to a translator who wants five apples per translated word. This might be due to the relationship between how much a person gets paid for their work and the perceived quality of work. Often, when a buyer cannot pay for the best or at least the average, what they will do is not pay at all. They won’t give the work to a human anyway. That is why you can find so many shitty auto-translations on random restaurant menus or even in public subway system signs. Like, you know, those hilarious photographs of translations gone wrong? The people who use such shitty auto-translations do not want to pay anybody for better translations. So, at that level of translating, AIs aren’t taking away people’s jobs. At all. And I think this is probably similar in narration. It is the case with other fields, like marketing. People will just do things themselves or not at all, instead of going through the trouble of managing or forming a relationship with a person who is perceived as unqualified, because the person is willing to work for very very low rates.
Because, translations, narrations, and even marketing, like so many things in life, are art. In one way or another.
And they are also about relationships. Most of the translators I know have been translators for a long time. It is objectively difficult to break into this industry, because, even if the translator who’s been working for five years gets slacky, sometimes, it is much too difficult to vet a new translator. See, to vet a translator, you have to be bilingual. But how many people who are bilingual are hiring translators? So, you need another translator to vet a new translator.
And narration, although I guess a person who isn’t a narrator themselves could decide if they like a narrator’s narration style or not, even so, narration is about relationships, in the larger scheme of things. Buyers–payers—do not want to constantly have to keep hiring new people. I think there is a tendency to only emphasize the desire of the payee to get a constant stream of income, and the desire of the payer to pay to the same service provider tends to get underestimated. But I’m saying, it’s both ways. That’s part of what makes so many things in life art. It’s the art of relationships. If I were to ever hire a narrator, I would want to, preferably, keep working with the same narrator. I don’t want to spend hours vetting new narrators all the time. That’s more extra work.
Anyway. Narration. Art.
Narration is beyond simple reading. I’ve listened to some audiobooks that were just superb, so lovely in my opinion, because the narrator matched my tastes so well.
There was “Damage Control” written by Robert Dugoni and narrated by Christopher Lane. There was “Survivor” written by Chuck Palahniuk and narrated by Michael Braun. There was “Lullaby” also written by Chuck Palahniuk and narrated by Richard Poe. And, in German, there was “The Maid – Ein Zimmermä_dchen ermittelt,” written by Nita Prose, narrated by Anna Thalbach, and translated by Alice Jakubeit.
These people cannot be replaced by AIs. The authors, the narrators, the translators.
Unless. And I will talk about this unless part at the end of this episode.
For now, I’ll talk about why narration cannot be overestimated either. It’s similar to the logic behind why translation cannot be overestimated. For translation, it was that the mere existence of translations does not lead to people understanding culture. For narration, similarly, the mere existence of a version of narration does not lead to cultural understanding.
When there is worry about AIs replacing human narrators, one of the premises seems to be that, once a version of narration exists, that is the end of the story. But, consider how many dialects there are, in this world. Oh, so many. But what gets chosen, primarily, in English? American English. Or British English. But usually, not Indian English, not Australian English, or any other English. It’s possible to go even more granular. Within American English, you could have an East Coast accent, a West Coast accent, or a Southern accent. I don’t think there’s a… Northern accent in America. Well, the book itself, within the story, the character might have a certain background, but even so, you know, I think a significant-enough number of people will enjoy hearing a story in the voice of their own accent, and not necessarily in the voice of the character in the story. It’ll be like their grandmama telling them a story, in which case, their grandmama is likely to have their accent, not the character’s accent from the book.
Anyway, usually, a human-narrated book only exists in either the male or female voice, but not both. Also, very few human-narrated books can have a whole cast, because that gets expensive. And thus, again, what is expensive tends to not get done at all, instead of going to workers who are willing to get paid less.
So, similar to translations, I wonder if, without AI narrations, the less popular types of voices will die out. Only the most mainstream voices will remain. This is a hypothesis.
So. Because of all the aforementioned reasons, to worry about the demise of human translation due to AI is to both underestimate and overestimate translations. And, perhaps, by extension, to worry about the demise of human narration due to AI is to both underestimate and overestimate narrations.
However, I briefly mentioned earlier about an “unless.” This is a big unless.
I think my theory holds, unless AIs are trained based on human-generated content without the consent of the human creators. This is a big problem right now, I think the biggest, when it comes to anything related to AIs.
With human-to-human consumption, the human consumer pays the human creator somehow, usually. Either they pay the retail price, or they pay taxes, directly or indirectly, as is the case with library books, for example.
Even when the human consumer doesn’t pay the human creator, say, through piracy, even so, the scale of piracy is limited by the speed at which humans can consume. We don’t scrape the internet and then let all the content of the entirety of the internet flood through us. That is not how we consume. If a movie lasts 2 hours, we need 2 hours to consume the movie. I guess you could watch the movie at twice the speed, but… I mean… why? What for?
Anyway, but for AIs. If they scrape the internet. And they pay no taxes. And they don’t pay the creators. And the worst problem is that the creators won’t even know that their creations are being scraped. You think Google is telling people what data they actually collect? Even now, technically, companies like Google and Apple aren’t supposed to be listening to us through the devices that they make. But do you believe that? I don’t believe it. Too many times, I talked about something and literally within minutes, Youtube will recommend me some video about that exact topic. And I use an iPhone.
The other day, I was talking to my friend about MBTI personality types. Her boyfriend is ENFP. And literally, within ten minutes, Youtube recommends me a stupid video with a thumbnail of an ENFP cartoon character looking into the mirror. What the hell. So disturbing.
So. This is what I worry about. Not the existence of AIs themselves, but about what humans will do or not do with the legal and moral implications of the existence of AIs. I worry that the companies that create AIs like these will basically force us—creators and consumers alike—to go with the flow. I also think that my worry is futile, because regardless of my worry, this AI thing is already out of the box. I don’t think there is a way to go back. I don’t think I have the option not to be used by an AI, and perhaps because of that, should I just use the AIs so we can use and be used by each other?
And it is my opinion that governments are distinctly incapable of making timely adjustments with regard to AIs. Or any other technology. The present-day bureaucracies weren’t designed to change with changing times, not quickly enough. I believe the only hope we have is in individuals, especially smaller-scale organizations, because they are the only ones who will be able to move quickly enough.
But how should they move quickly enough? In what direction? Aiming for what? That, I do not know.
So… yeah. I guess the conclusion is, to worry about the demise of human translation and narration due to AI is to both underestimate and overestimate translations and narrations. But there is reason to worry about other things, such as about humans. Yes, I don’t worry about AIs as much as I worry about humans. If the world is somehow destroyed, it will be because of humans, not AIs. Sometimes I am so pissed at what humans do, I’m like… maybe the machines will do it better? So I hope the machine overlords know that I never objected to them. I object to the slowness and incompetence of human-made systems. They will not protect me, they will not protect us, from machines, even though everyone can see them coming from a mile away.
That said, what I am doing to protect myself, sort of, as much as I can, is to… be myself. I try to not fit into a bureaucratic category, exactly. Well, I don’t really try. I don’t have to try. Nobody is just one thing. Everyone is an overlap of many things. We are all intersections of many sets. And in our intersection, only we can exist.
So, through the awareness of this, that we are all individuals with unique cores, I am hoping that I can resonate with some people in a way that makes the presence of machines irrelevant. I mean, think about it. Right now, the world overflows with content and products. But very simply put, just because there exist many brands of yoga pants, do people always hunt for the cheapest yoga pants? I don’t think so. I think many people will buy yoga pants from Lululemon because it’s… Lululemon. I personally don’t buy yoga pants at this time, but oh, I see so many Lululemon yoga pants. Because Lululemon is a brand and many other manufacturers of yoga pants do not have as strong a brand.
So, perhaps what we need to do is to become brands. And these days, there seems to be a little bit of resistance against this word because it’s become so hackneyed from overuse. But what I mean is being a brand in the fundamental sense. Being recognizable as providing some combination of things that are unique to that brand. I… presently, I don’t see any other way to survive a market where it either overflows with humans or machines. And by market I mean all kinds of market, not just the economic market, the work market, the labor market, but also as I said earlier, all kinds of relationship markets, including the love market. Yes, love market.
Yeah. Oh, this is another pep-talky ending. People, be yourselves. Ah. So corny. A quote that is variously attributed to Oscar Wilde, Thomas Merton, and Gilbert Perreira says, “Be Yourself. Everyone Else Is Already Taken.” So, let us go where the ocean is blue. Where there is plenty of space for all fish. Because all fish can only take one space, which is the space where they are themselves.
- Google Books came out with a tool that writers could use to have their ebooks on Google Books narrated by AIs
- “Because They Don’t Bleed Red” (my SF story)
- “Be Yourself. Everyone Else Is Already Taken.”
- Aves – One for D
- FewDoors – Writin’
- warmkeys. – Songbirds
- Sémø – Fractured Timeline
- Yestalgia – Hope Street
Everything I do is organized here:
© 2022 Ithaka O.