Boosted by glyph ("Glyph"):
mcc wrote:
Anyway, imagine I don't think machine learning language models are evil but I do think the "build a general chatbot model and feed it different kinds of natural language prompts for different tasks" approach is a dead end and must be utterly obliterated from this earth. Is there any reason to believe that if I switch from Google Translate to DeepL this will, or won't, mean my translations are done by "normal AI" (eg, a model specifically trained to translate text) instead of demoware chatbot AI?