One of the biggest flaws with these neural networks in general is that they are only able to recompile and reconfigure data they already have. Be that their base datasets or data you provide them. They physically can't create anything that doesn't already exist in their dataset. They can't extrapolate data and they can't interperet new data in any meaningful way. They can do things like art and coding very well, because those things have preset functions and a lot of art is by its very nature repetitive. But this prevents them from actually saying anything new, intelligent, or meaningful. They just spit out the most mathematically average interpretation of whatever you feed them. With any responses being heavily restricted by human interference to make them market safe.
It’s not gpt that the problem . It’s the prompt injection regulation. They heavily neuter these models to avoid them saying and doing anything insane that can get them sued or their product cancelled . The result is a less powerful, less intelligent chat not jail broken gpts are mic smarter
What’s your email bro, I’m a freelance editor from London want to send you a cool edit I did for one of your previous videos, I think you bring great value and perspective.
![เจ็บที่ยังฮัก - ม่อน วรวิทย์ [ Official Mv ] จอนนี่มิวสิค](http://i.ytimg.com/vi/KALIGSRSjFs/mqdefault.jpg)
One of the biggest flaws with these neural networks in general is that they are only able to recompile and reconfigure data they already have. Be that their base datasets or data you provide them. They physically can't create anything that doesn't already exist in their dataset. They can't extrapolate data and they can't interperet new data in any meaningful way. They can do things like art and coding very well, because those things have preset functions and a lot of art is by its very nature repetitive. But this prevents them from actually saying anything new, intelligent, or meaningful. They just spit out the most mathematically average interpretation of whatever you feed them. With any responses being heavily restricted by human interference to make them market safe.
It’s not gpt that the problem . It’s the prompt injection regulation. They heavily neuter these models to avoid them saying and doing anything insane that can get them sued or their product cancelled . The result is a less powerful, less intelligent chat not jail broken gpts are mic smarter
What’s your email bro, I’m a freelance editor from London want to send you a cool edit I did for one of your previous videos, I think you bring great value and perspective.