Re:ChatGPT and Wolfram Is Insane
by danans on Friday March 24, 2023 @04:20PM
(#35294893)
> Kids submit GPT homework and it’s either correct (rendering the skill the homework was testing worthless)
LLMs are trained entirely on content produced by people. If people stop developing the skills required to produce well thought out content, the models will stagnate and even decline. They are completely dependent on humans for content input, so the skill of creating it will always be worth something if the language model is worth something.
Re:ChatGPT and Wolfram Is Insane
by RC_ITR on Friday March 24, 2023 @04:43PM
(#35295176)
We agree. Creativity will continue to be more valued, while formulaic synthesis will lose value (in the same way recall has declined in value as databases get better).
LLMs are trained entirely on content produced by people. If people stop developing the skills required to produce well thought out content, the models will stagnate and even decline. They are completely dependent on humans for content input, so the skill of creating it will always be worth something if the language model is worth something.