I agree completely. A doctor losing a patient’s life due to a mistake is an unfortunate occurrence, but ultimately that is just one life.
A mistake in a translation or interpretation could affect countless of individuals over countless of lives.
For that reason, I have decided I do not want to participate in these forums in matters relating to interpretation. I have realised even participating in discussions relating to translation can lead to lots of misunderstandings (on my part as well as others). The danger of these misunderstandings is that they can viewed by others and in turn influence others. I do not wish to carry the burden of prolonging others suffering.
I do maintain a website documenting my own journey, and I publish my Pali textbook and translations there. I warn readers that most of the content on the website are for me, and not an exhortation to others.
Which reminds me, I need to get back to my translation of Kaccāyana and my Pali textbook. They are secular works, so I am more comfortable that my output is of low criticality. I have been very busy this year due to my commitment teaching a few masters courses, but the reward from that is the joy and satisfaction of seeing young minds blossoming.
However, Smalley said there are concerns that post-editing can create “a lot more work” for translators, who have to carefully compare texts to catch misreadings and “poor or unidiomatic” style. “Colleagues who have done this kind of post-editing work say that it requires a far higher degree of attention, because the AI generated text often reads so plausibly,” she said.
I am one of those who don’t like post-editing. And this is independent from the kind of text I am translating. I tried it a few times in the past with DeepL, I think, on texts that were not suttas, and I found it just doesn’t write in the style I like. I felt it took me a lot of time to think about how I would express such a sentence if I were to express it myself, and pretty soon I decided that I prefer to just do it on my own.
A while ago (and I’ve mentioned this elsewhere in this discussion already, but I don’t remember where) I was asked to complete the work of another human translator who had started translating a text, but couldn’t finish it. I found myself unable to write in the style of another person, but at the same time wanted to honor their work as much as possible while still delivering a consistent piece of text. That took me some sweat!
I won’t ever be a large-scale translator, I believe. I like it to take each text I work on as an individual piece of work that deserves my full attention in every detail. (Just to be clear: I am not a professional translator.)
But with sutta translation there is still another dimension. While working on a text, I inevitably also engage with the meaning, and what it means for my own life and situation. This is helping my personal understanding of the suttas much more than just reading them ever could. This is the greatest personal benefit I gain from this work, and speed is the last thing that matters here for me; except for the fact that my life span is of course limited, and I am no longer in the early phase of it.
Not sure whether this falls under “religious emotions”, and if it’s worth talking about or not …
Adding:
This doesn’t mean that such a deepening of understanding cannot happen for someone who uses pre-translations of any sort. I even know a few people with whom I can see this. But with me, it simply doesn’t work that way.
To err is human. No one single unenlightened human being is perfect.
Working as a team, not only in the same physical space and time but also across physical locations and time, is a way to better serve the dhamma. I hope someone knowledgeable would have a friendly academic discussion with Ajahn Sujato about some of his translation that they don’t agree with and they both come up with a better version instead of saying “I disagree with him” and “This is what I think”.
And on a side note, even an arahant is perfect, they are not perfectly capable of rendering the Buddha’s teachings to all languages accurately.
As a fee-earning translator, I know for a fact that the threat is real and the fear among many translators is real and well-founded.
Personally my ego says that I don’t care. If a client doesn’t want to invest a lot of money for high-quality translation, they are entitled to be satisfied with AI work, just like poor people have to accept low-quality products because they can’t afford better quality ones.
I totally get your point. I guess there is also a difference if we engage in the work because of spiritual/personal interest, or if we do it to earn a living.
I studied Buddhism academically for more than 10 years and for me, translation at some point became just a step in accessing sources, with little “sacredness” involved. And while for Pali Suttas the necessity for translation is not that great anymore, only very small sections of the Chinese, Tibetan, or even Sanskrit texts available are translated into western languages. However, there is a strong need for people to skim through these texts in a timely manner, to find topics, get related passages etc. so machine translation is going to become a game-changer for the research community eventually. Whether that will happen this year or in five years is a bit difficult to say, but the technology is improving rapidly.
I don’t think this technology will “replace” human-made translations at any point, but I can also imagine a future where we won’t do translation at all anymore. I wouldn’t be surprised if in 5-10 years from now we will have plugins for the webbrowser, mobile phone that translate any given Buddhist text with pretty high reliabity into any of our favorite language. Who needs a human translation anymore by then? The researcher will be happy to get to the goal quickly. The non-academic person, interested in the Suttas, is likely to go for reading human-written books anyway – they are not the target audience of a machine translation system.
Not sure what the translators of these texts will do. What I see for now is a strong generational divide: People older than 40-50 are usually mildly to strongly opposed to machine translation, while the younger folks generally welcome it. I have seen this divide in reactions talking to translators working on all major source languages, i.e. Tibetan, Pali, Sanskrit, Chinese.
My armchair hypothesis on why we see this divide is: People younger than 40 grew up with the internet as a major technological disruption, while the generation before that only knew the TV, and not much changed in their lifetime. They know the internet, but they haven’t been forced to change their life in such a fundamental way as the younger generations. The TV still works! The younger folks have the traumatic experience that technology can totally change the way we do things completely, while the older generation might claim that things are as they always have been and shake their heads at the younger generation that uses mobile phones from morning till evening.
I am not claiming they are any more correct, thats my ‘armchair hypothesis’, and if its wrong, fine!
Alas, age is a bad argument especially in relation to wisdom; what I presented above is an observation, and an interesting one; I don’t expect you to agree with my speculations about possible reasons, that’s just one among many. What other options do we have? Your idea that young people are less wise?
I was working for a CCRC/IRC quite some time ago now who produced the second generation of the most powerful modelling system in the world at that time in Engineering, and was working on further generations, so I spent quite some time with PhDs who were tinkering with fuzzy logic and training algorithms for fuzzy logic and such to try and mesh this business with the probability models. As well, they were generating fuzzy logic applications to warm problems like labour productivity, underwriting/insurance for non-historical problems, etc. The other thing they were working on was distributed modelling and they had access to US military stuff for that.
Don’t ask me about it though, I would have to refer. One of my good helper PhDs handed me a book on fuzzy logic that he thought would be helpful after I asked him “anything not in math.” He was standing there as I excitedly began to flip through it until I hit the third chapter, which was all math, the fourth, fifth … I am pretty sure I started up from my desk to back away and then controlled myself, but I couldn’t do anything about the stark horror showing in my face.
I know a linguist who spent most of her early-mid career on a single particle in Japanese (ni) who is also a professional translator and teaches translation. I do believe she would share wholeheartedly in your appreciation for what you obtain and can share out of your very considered work. Experts with that knack for both meaning and for translating that meaning, the sense of it, to others in ways that invisibly make the difficult appear easy are so rare and valuable.
I remember one time as a teen I “mastered” some books of my own choosing off my dad’s bookshelf and came strutting into the kitchen to announce my ingenious, individual, startlingly brilliant, new discover to my dad. He was washing the dishes and so was half ear to me until my brilliant announcement came out full stream. He grabbed the dish towel, turned around, stared at me and pointed his finger with soap bubbles still on it and said, “Yes, very good. But do you believe you’re the first person to have thought that. You don’t think this isn’t part of our collective knowledge”? He was peeved. Such a rare thing. I remember it well.
Yeah, I suspect it’s a Hard Problem because we don’t even know what we want.
We build exact knowledge systems (e.g. Wolfram Alpha) which are capable of precise inference but often misunderstand our queries and have to be painstakingly filled with precisely tagged data OR we build big probabilistic models which can automatically churn through data and capture a lot of nuance but can’t provide any guarantees of correctness… Smarter people than I are working on bridging that gap, but I wouldn’t be surprised if a theorem is proven in the next few years that it can’t be bridged… though even that might take a clearer definition of what we want than we are able to articulate… But maybe this is just “the alignment problem” restated…
No. The answer is here in this great sky rolling across the golden endless prairie. The light practically lifts a person up into it.
I think I mentioned before a long time ago that they discovered the clock went backwards on the Monte Carlo simulation engine.
And yes, the reason why they were doing this was because they were working under a tripartite funding structure that involved the honking corporations in the oil patch who had cost overruns in the multi-billions on megaprojects (non-historical) and were striving for cost-productivity models to handle that.
Fuzzy logic offers some advantages when there is no historical data but the teams have access to expert knowledge. They spent a lot of time developing surveys and things to figure out how to translate expert knowledge into fuzzy models.
It is also less (though better to say more or less) precise than probability models but faster. Similar to the “hunch” that experts decide about whether or not to pursue (often in their own heads).
Sooner or later, the makers gotta put something down though and that always involves risk, hence some of the connections to other stuff (i.e. underwriting/insuring projects).
We encountered massive ethics and other problems in producing methods to respond to some of the warm problems, however.
Well Abraham, my helpful PhD who is still a very good friend told me way back then that that 2gen modelling engine could bring construction of massive projects down to the precision of I think he said 2 cm. And they were modelling in 5-D, oh yeh, something else they were working on. He thought this person was God.
For my work, those are certainly the primary target audience. Or no, it’s probably even more accurate to say that I myself am my primary target audience. The mere fact of doing this work gives me so much joy and satisfaction that nobody could ever give me by paying me for it. And it still surprises me when I get feedback from others that they enjoy my translations.
Well in that case I fall a bit outside of all categories.
I am 65 and have had very little exposure to TV in my life. The first look I had into a TV was at the age of 10 when a neighbor invited us kids from the street to come into her living room and watch the astronautes do their first steps on the moon. My parents didn’t have their own TV until much later, I guess this was only after I had moved out. But from the age of 10 on, my grandparents were living in the same house with us, and we kids were allowed to watch certain things at their TV. An allowance of which I made much less use than my sister and brother.
As an adult, I owned my own TV for perhaps 5 or 6 years, then I found it doesn’t do me good, pulled the plug, and gave it away.
The first time I used the internet was at about 50, and I only did it because I had to (for work). I learned how to write emails and such simple things. And then came quite a steep learning curve! Via the internet I found Dhamma connections I would never have encountered otherwise, and finally ended up using it all the time for working with Bilara, and even develop the Voice website family together with Karl …
But I still don’t personally use machines for translation. Which doesn’t mean I categorically condemn others who do it, considering the context and the way it is done.
I also wouldn’t rely on machine proofreading alone, the result of which I can sadly see every morning in my dad’s local newspaper!
Thank you, Sabbamitta, for sharing your story about technology! I’m 63 and know the old world and the new one is certainly more complicated, with the benefits traded for losses. All things considered, technology isn’t the panacea.
On a higher level, developing humanoid androids with the ability to have awareness and consciousness, and instilling the ethics of the Dhamma into them may be quite an important and monumental task. What if you, in your next life, were reborn as a mind-aware android, with hundreds of beta levels of Buddhist programming in your cybernetic matrix? Would you consider yourself lucky if the team that built you were all Arhats and high level Practitioners and Bhikkhus?
Life comes from life. Fifty years ago the level of computing power of today seemed impossible. Will it take another fifty or one hundred years to build a human like android? Or one day, Lieutenant Commander Data, anyone?
But maybe this is just “the alignment problem” restated…
The mere fact of doing this work gives me so much joy and satisfaction that nobody could ever give me by paying me for it. And it still surprises me when I get feedback from others that they enjoy my translations. 
