Post #4173529

If they could somehow solve every single bad thing connected to large language models, both perceived and demonstrable, I still wouldn't want to talk to one by voice.