4 ms·
>it shouldn't be a stretch to believe that sentience might produce answers that people prefer Even if that were true, there's no reason to believe that trainin
by root_axis 1y ago
>it shouldn't be a stretch to believe that sentience might produce answers that people prefer
Even if that were true, there's no reason to believe that training LLMs to produce answers people prefer leads it towards sentience.