Thread 106341664

How can one person be so insufferable?

>>106341664 (OP)
How's the PhD level GPT-5 working for you? LLMs are a meme, they are not the way to general intelligence because the training is inherently flawed.

wtf is an energy based model

Notice how he pitches these alternative architectures but doesn't provide any proof of concepts

>>106341711
I'm sure Yann has a working prototype which he'll share with the world any day now.

>>106341721
Because he has nothing, it's all conceptual,
>search the solution with different methods
>irrelevant pic

>>106341711
Boltzmann machines

>>106341721
>>106341726
so what does he do all day?
sit around theorizing?

>>106341810
That and raping kids minds and talking to his WEF and censorship masters
Don’t blame him they pay him way too much for this, it’s an offer he can’t refuse

>>106341810
sell you lies

>>106341664 (OP)
He forgot an "of" between favor and joint-embedding. Should I send him an email?

>>106342991
You have to have the smoothest of brains to think that LLMs extend to AGI, much less ASI. Lecunn is absolutely right - that kind of stuff is going to come from elsewhere. If you asked him I'd sure he'd tell you he's not sure what the answer is, specifically.

(You) believe in LLMs because your mind is simple enough to be defeated by highly statistically accurate word prediction. You are too simple to grasp this basic fact.

>>106343455
bot

>>106343591
Bot. Right. You know how boomers are just completely fooled by AI generated images? They don't the context to piece together that what they're seeing is artificial. You're the same way, but only a tiny notch higher. You're fooled by clever text.

>>106341664 (OP)
I hate Yann LeCunt. He has that smelly Frenchman smugness about him.

>>106343455
>>106343612
Where the fuck did that poster say that LLM was leading to AGI you fucking sperg?

>>106343455
>LLMs
They are already not anymore what they were in the beginning. There's talent and incentives to throw at them to expand them into increasingly more capable AI systems. The original LLMs don't lead to AGI, that's true.

>>106341721
He's one of the top 10 most published AI researcher.