I might get tomatoes thrown at me for sharing this, given the recent hostility to AI many here have shown, but I have created a custom GPT to help tutor Pali. I have been using it to help myself parse out phrases and understand the origin of certain words and have found it very helpful.
Here is it if you want to use it (note you need a ChatGPT subscription): Pali Language Tutor
Here’s an example of a chat session I had with it today while studying lesson three from G&K (all can view this link, I think): Parsing Lesson III, Part I
Thanks for posting these guidelines, Ven. @Snowbird.
Personally, I was quite shocked and horrified by this whole discussion, and would strongly prefer it if students in my Pāli class did not use AI to assist them with their homework exercises. Aren’t you taking this class to learn how to translate and parse Pāli sentences on your own? That’s the whole point.
I hope you can feel comfortable in asking people in this forum for help with your Pali homework. No need to feel isolated and have to talk to a machine.
Thank you so very much. I know how kind and helpful many people here, especially you and @johnk, are. I really appreciate your generosity with the beautiful dana of knowledge that you both have been giving us all.
Regarding AI, I don’t like machines and technology. I’m old and love ‘pencil and paper’.
As a teacher and professional translator, I’ve been trying to learn the advantages and disadvantages of computer-assisted teaching and translation. I believe that nothing in the world is absolute - not absolutely good nor absolutely bad. We humans need to be knowledgeable and wise enough to know how much we can rely on computer/technology.
I’ve promised Ajahn Sujato that I will try to render a new version of Thai translation of some of the suttas on SuttaCentral, based on his translation. Hence, my need to learn Pali – not to enable me to master the language so as to translate the suttas, but to help me make educated/informed judgement ‘when’ and ‘where’ to ask Ajahn why he rendered the translation the way he did (when his interpretation is different from Bhikkhu Bodhi and/or existing Thai versions).
To work on such a big project, I think someone (something) at my ‘beck and call’ to help with parsing the sentences could be ‘handy’.
I think the issue is that the AI tools are still very bad at Pali when you really pin them down. The things it does get right are amazing. But we aren’t using it for party tricks. And by design it will sound very confident with itself (despite all the “Oh, I’m just a poor AI bot, I make mistakes”) As learners, by definition, we won’t know what’s right and what’s wrong.
But of course I know that translation is very, very difficult and it’s understandable that folks will want to use whatever tools are available. I just don’t think these bots are close to providing real help.
If you read the AI’s work that @JStewart is working on, you can see that it does more than normal AI. I copied and pasted from two AI programs: one is ChatGPT 3.5 (free) and the other is Copilot (part of our Microsoft 365) in order to compare its work with normal AI programs.
If one has no knowledge at all, one tends to rely too heavily on a source. But if a learner or a translator uses Stewart’s customised AI, it may help them to a certain extent – that is my two cents.
Not only AI but also the teachers I highly respect like Ajahn Brahm, Aj Brahmali and Aj Sujato don’t automatically get a ‘Yes, I believe you’ from me. That’s why I’m learning Pali – so that I can ask intelligent questions when I am in doubt.
AI as well as internet is like fire: dangerous but useful. Young children should not play with it, but as responsible and wiser adults, we know that children needs to be trained to use it skilfully and made aware of its dangerous side.
Would it be more appropriate for the mini-thread started by JStewart:
to be moved wholesale and live under the “Miscellaneous” Pali Course category?
For several months, beginning with Bhante Sujato’s OP, many people have dedicated precious time to curate this list. I consider it a gift to the larger community. Given the AI content guideline, I wouldn’t want someone to read the AI recommendations that now live on this thread and inadvertently get caught up in them/it. To me, this de-values the Resources thread.
Well, James, I do have to say, that I’ve just now had a close look at the output you shared from your custom Pāli GPT, and my reaction is “Wow!”
I am very, very impressed with your programming skills in what you have been able to do with this AI tool! I say this as someone who, in his “real” job (pre-retirement) made a career in computer programming and systems and database designing, starting back in 1973(!)
On the other hand, I do agree with @Snowbird that while what AI can do and get right is truly amazing, the danger is that it can occasionally be supremely confident when in actual fact it is plain wrong.
I know how kind and helpful many people here, especially you and 
