News for hackers. Stuff that matters.
Re:Stop Calling Everything AI, Machine-Learning Pioneer Says by m00x on Thursday January 19, 2023 @12:25AM (#34437005)
> One of the challenges faced by ChatGPT was in seeing it produce conversational-style output instead of simply the most common words that it saw.

Do you have a source for that? I've implemented GPT from scratch and I don't see anywhere that doesn't just take the current output + input as attention and produces the next word. The loss being if it correctly guessed what would be the next word according to context and position.

Re:Stop Calling Everything AI, Machine-Learning Pioneer Says by randomdata on Thursday January 19, 2023 @12:31AM (#34437036)
There were some papers linked here yesterday. I'm not good at memorizing URLs. Sorry.
Re:Stop Calling Everything AI, Machine-Learning Pioneer Says by fiso64 on Thursday January 19, 2023 @03:14AM (#34437921)
ChatGPT was also trained with reinforcement learning to produce outputs human raters preferred
Re:Stop Calling Everything AI, Machine-Learning Pioneer Says by m00x on Monday January 23, 2023 @04:57PM (#34495679)
It's definitely getting closer in this case, but still chatGPT still look for any truth in its response.