I hope LLMs are better trained with all contents on this marvelous community, and on the precious 1M+ WRC cases. And certainly on the whole product documentation site too. They are collective knowledges and now the tech is mature in every respect to turn such knowledges into real intelligences?
Intersystems Cache/IRIS performs. I know it performs on a levelled-up computation unit, since I happened to have tried it for years too for so many sites. It's established in Healthcare too. LLMs happen to be able to make any niche COS/MUMPS as efficient and as popular as any other in Dev/Test phases as well, not only in service runtime phases anymore - the first time of its history it has a technical chance to be as easy and equal to use and as resouceful as any other scripts, so why not.
So the best learning resources? ChatGPT/Claude/Genimi or Grok/DeepSeek/Qwen etc can generate quality IRIS solutions, quality IRIS codes and quality IRIS documentation to any customer demands, the same as they would do with Python and JS, and give the best answers directly to ANY questions, instead of pre-2023 style docs only that need time to search and follow? Would this be too bold a wish even during Christmas time? :)
https://www.bbc.co.uk/news/technology-65110030
Agreed with the risks, so I appended a line of excuse in Disclaimer section.
Not sure we should stop or slow down the phenomenal progress of AGIs, but I do agree sovereign states should speed up legislations. Also the theoretical researches on its mathematical nature. :)
Looking forward to them. Thanks.