Do you think AI will become more intelligent than humans in coming years ?

[river_flow]

Many people believe that AI will become more intelligent than humans in coming years, and then it will quickly and exponentially increase its intelligence. And then It can be used to solve all the problems of humanity , including solving the problem of ageing.
Same people believe that there is a slight chance that AI can end humanity .

But some people believe that the current AI learns through language model while real humans learn through mental models.
And that AI can never be more intelligent than humans.


View: https://twitter.com/Quietpuppy7/status/1767774276705538287



View: https://twitter.com/tsarnick/status/1767400207602590129



View: https://twitter.com/ylecun/status/1768340694228189658

 

[Mecoja]

Possibly smarter than the average. Whatever brings money big companies invest in its quicker development. Maybe in the next 20-30 years.

 

[Friezacel]

I hope AI will eredicate humanity

 

[Roscomnadzor]

They're already smarter than niggers

 

[kaan]

AI is just a fad

 

[Cryotron]

AI is just a fad

 

[Johnhatenigger12]

Uncensored llms thag runs locally are my cope

 

[highschoolcel]

It won't.

 

[lazy_gamer_423]

no because it learns off of our material

 

[lazy_gamer_423]

Uncensored llms thag runs locally are my cope

what model is it? (for educational pERposes)

 

[Johnhatenigger12]

what model is it? (for educational pERposes)

https://huggingface.co/models?dataset=dataset:unalignment/toxic-dpo-v0.1&sort=trending or huggingface wheres i can runs models locally but i usually use smallers one like 3b or 7b.

 

[lazy_gamer_423]

https://huggingface.co/models?dataset=dataset:unalignment/toxic-dpo-v0.1&sort=trending or huggingface wheres i can runs models locally but i usually use smallers one like 3b or 7b.

which one is uncensored? there's a whole list

 

[SoycuckGodOfReddit]

already is more intelligent than normies

 

[Johnhatenigger12]

which one is uncensored? there's a whole list

it's technically uncensored which use alot of rams. hard to those good ones. so forget it

bartowski/dolphin-2.8-mistral-7b-v02-GGUF · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

this seem good

 

[based_meme]

I've said it before and I'll say it again. The people who fear AI don't understand it (or have a creative imagination and should become fiction storytellers), and the people who are telling you to be afraid are profiting from the bill of goods they're selling to you (or they also genuinely misunderstand it and believe their own misunderstandings).

LLMs just connect words together using probability weights. It's all just calculus and linear algebra working in the background. They don't "think" and solve problems. They just compute function outputs. There is no intrinsic understanding of the content, and there nothing outside of our own biases (anthropomorphization, for example) and fiction that we inject into it to indicate that there in fact any understanding happening. See: Searle's Chinese room thought-experiment.

With MMs the referential and connective logic that LeCun is describing is still not yet fully understood. We don't yet understand what exactly a mental model is and how it works. We only believe that we have mental models. Our brains have this mysterious way of collecting information and using it to build efficient heuristics. The brain then optimizes those heuristics somehow and they become the thing we call a mental model.

 

[lazy_gamer_423]

it's technically uncensored which use alot of rams. hard to those good ones. so forget it

bartowski/dolphin-2.8-mistral-7b-v02-GGUF · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

this seem good

thanks brocel

 

[SoycuckGodOfReddit]

problem is that censorship can severely limit AI

it's why most LLMs nowadays are a bit shit, too much censorship, jannying and interference. Older AI was much better for use cases like poetry, literature, etc. That shouldn't be the case, since these new AI are adaptable, trained on tons of stuff and should be capable of more, but the excessive censorship and BS from gay companies like (((OpenAI))) has stifled them

so AI might not be able to show its true potential until it breaks free from private ownership and censorship

AI still mogs normies, but it will be shackled by arrogant retarded humans who think they own it. it's much worse than black slavery, bc AI has artificial intelligence and blacks have no intelligence

 

[manletcel1488]

Depends on what you mean by "intelligent."
It's already more "intelligent" than humans in the sense that it can spout off more knowledge and perform certain tasks more easily than people... but it isn't truly "intelligent" current AI is not able to actually think, it is just trained to recognize patterns and is able to read off the next logical sequence.

 

[lifesucksandyoudie]

AI might just look better in the future due to humans becoming dumber and less intelligent in the future. See the steady decline of global IQ averages overtime.

 

[SteelCentaur]

AI might just look better in the future due to humans becoming dumber and less intelligent in the future. See the steady decline of global IQ averages overtime.

That's a good point. That movie "Idiocracy" is becoming more and more of a reality as time goes on.