Re:French AI startup Mistral releases ‘unmoderated’ chatbot ...
by sixothree on Saturday September 30, 2023 @11:34AM
(#37716418)
Would that not be the actual result of training an llm with real data?
Re:French AI startup Mistral releases ‘unmoderated’ chatbot ...
by thaumasiotes on Saturday September 30, 2023 @07:59PM
(#37721160)
Your wish has been granted: https://languagelog.ldc.upenn.edu/nll/?p=60536
In a somewhat surreal turn of events, the author of that post concludes that "Hobbled by such constraints and limitations, China's chatbots are certain to be inferior to those of free societies."
(He is prone to slamming the CCP regardless of whether he can make a logical case, which is odd since it's not like it's hard to make justified criticisms. He is also known for his habit of deleting comments if he doesn't like them, which is why that post has only the one comment on it.)
Re:French AI startup Mistral releases ‘unmoderated’ chatbot ...
by dragonwriter on Sunday October 01, 2023 @05:28PM
(#37731041)
> Might be then people behind LLMs in Meta and OpenAI will see how laughtable and awful this childish censorship is.
They are doing it to curry PR and political favor, which can be useful to secure favorable regulation, not because they think it. Whether it is “childish” is irrelevant to the calculations which support doing it.
Might be then people behind LLMs in Meta and OpenAI will see how laughtable and awful this childish censorship is.