If you go to the following site you can find all of the commentaries translated into English. Just select English as your language: https://tipitaka.paauksociety.org/index.php
8 Likes
I canāt seem to find an āaboutā section on the site to explain who has done the translations, etc.
For folks who want to see the Pali in bold text, you can use a browser extension like Stylus and add
.Pbodytext {
font-weight:bold
}
The site itself lets you modify the font, however bolding is not an option.
Thanks for sharing this! Would love to know more about the creators.
5 Likes
I strongly suspect itās machine translation plus human editing thatās been done.
Iāve gone through the SÄ«lakkhandhavagga, and there is a glaring lack of typos. This is almost impossible to achieve for the size of the document (800 pages all). Personally, I canāt go through a forum post without a couple. 
The style, syntax is inhumanly consistent - just comparing to a random collection of, say, Bhikkhu Bodhi, Sujato or Charles Patton.
(But there is the sentence āThe rest is as stated everywhere elseā repeated twice in a row verbatim. Thatās an interesting artifact.)
Thereās some interesting stylistic choices that explain implied verbs/objects in asterixs or brackets that a machine is unlikely to do, but even ChatGPT could be prompted to include things like that.
Given the sudden appearance and the crazy scope of this project (Chinese & English Full Commentaries), as well as the complete lack of attribution - Iād wager that these are machine translations with human editors.
I donāt mean this as an accusation - but if this sort of thing is a deal breaker for you, well, thereās good reasons to suspect machines here.
Edit: Heh, Yamakavaggo in SN Commentaries has this interesting artifact:
- Corrected Translation: The Pairs
The document has the same āCorrected Translationā artifact in two other places.
8 Likes
Itās the difficult words and expressions to spot check, not the common ones. Because thatās where machine translation fails - where humans are uncertain or the material simply hasnāt been translated before.
3 Likes
Even the straightforward stuff is a bit distorted. I started skimming the Vinaya Commentary and line 16 for TherÄ ca bhikkhÅ« navamajjhimÄ ca it translated as āthe elders and monks, both new and intermediateā Right words and kinda there but as if through a funhouse mirror.
5 Likes
So, not very promising then. The programmer hasnāt given it the semantic sophistication to identify the necessary parsing as:
therÄ bhikkhÅ« ca, navÄ bhikkhÅ« ca, majjhimÄ bhikkhÅ« ca.
3 Likes
Indeed. As I said in a previous thread, even AI optimists should not want AI translations statically hosted like this.
3 Likes
Skynet strikes again
1 Like
I mean, I think that would be worse - because then the final output has no human oversight, and weāre back to the āmachine said thisā kind of place.
Properly using AI, to provide a very rough draft, just so a qualified human expert can get in and do the real work a little easier - that would be better I think. Though Charles once mentioned how it could dull the translator. And Iām not well versed in commentarial language to assess the fidelity of this current work.
Currently, weāre in a very weird spot. Thereās a proper interface that could be done to help translators and empower them with AI - think something like Bilara equipped with AI, that goes sentence by sentence, and also caches your edits (as well as being able to mark when a source text has a typo so that the machine doesnāt suddenly think itās okay to keep translating it like that - something the Chinese translations would definitely need) to make sure the next sentences are loyal to your earlier choices. Such a tool would greatly streamline human efforts, diminishing error margins, helping keep a consistent syntax throughout etc.
However, people whoāre serious about translations, donāt even attempt to use AI at all; and people who use AI donāt seem to be interested in sophistications like that. So despite the technical ease of building such applications, it really doesnāt exist in reality. 
I think you are assuming this had human oversight.
Itās a shame that they donāt explain the process that generated the translation. Or eve give a contact email.
2 Likes
I think artifacts like āCorrected Translationā point in that general direction, but yes, itās still a working assumption.
It does say quite clearly in the News page that these are AI translations
3 Likes
Good spot there. Thanks.
The thing thatās absurd to me about it is that there is a kind of random number generator nature to machine translations. If a text is run through a model repeatedly, it would spit out variations that are all a little different (or very different, depending on how the ādiceā roll). Running the texts through a model once and then posting the results is a little arbitrary.
Of course, the proper way to go about it is for a competent person to carefully edit the output. But the idea is to skip the time and effort and just throw something online. And a competent person who attempts such editing will usually come to the conclusion that itās just as fast to translate it from scratch because of the inconsistent translations of terms throughout, etc.
7 Likes
As everyone agrees with the evidence that these are AI translated, I (moderator) recommend locking and hiding this thread. Please let us know of any objections via this thread or in a PM before we do so.
Personally I donāt think hiding the thread is appropriate. The AI rule is about posting AI content directly on the forum, no? From Q39
Links to sites with predominantly AI generated content will only be tolerated within the Watercooler.
Seems like everything is in accordance with policy.
At most the OP should have a warning added that the site linked to contains AI translations.
We need to discuss projects like this as they evolve.
12 Likes
I agree. The critiques on this thread are important.
5 Likes
Precisely. Meanwhile real EBT experts such as yourself have to waste time debunking this thoughtless slop.
But perhaps thatās nothing new, as @Bodhipaksa can confirm 
Just a shame to see someone who obviously cares about the PÄįø·i texts (as this siteās author doubtless does) fall for the AI marketing and (imo) hinder their own cause.
5 Likes