in rhetoric/oratory, both sides (or just the speaker) are trying to convince the audience, so the situation is different. In that case, very practically, the burden of proof is on the person who wants to convince the audience (in a reductionist way, we could say the audience accepts the null hypothesis by default).

In most internet discussions, the situation is different. Telling someone "the burden of proof is on you" is just a lazy/casual way of saying, "I don't believe you."

Good job, you figured it out. I support your search for continuing to figure it out.

If that were a serious question, you would familiarize yourself with the debate and research on the topic dating back to Descartes. But I think you lied, it wasn't a serious question.

Burden of proof is something that is assigned in a court of law.

Back in the real world, the burden of proof is on the one who wants to know the answer. If no one proves it, then we won't know.

In the court of law, the burden of proof is assigned to one side or the other.

In the real world, the burden of proof is on the person who cares about the answer. If no one proves it (true or false) then the best we can say is "we don't know" or more likely, "X is true with n% probability"

Why? Even if I told you, you wouldn't understand. You've proven yourself unable to understand the main point.

LOL that's kind of hilarious.

That's a really good question but we haven't experimented with it much because it's so expensive. Modern NNs are heavily biased towards gradient descent.

There's a hypothesis, but we don't know because they haven't surpassed human language proficiency. Not even close.

An RNN doesn't have backwards linking synapses. You can't use gradient descent in a neural network with backwards linking synapses because then it's not differentiable and training the network takes significantly more compute power.

Again, you fail to find the main point.

It's not false.

Every sentence you wrote is wrong. Everyone here is literally dumber for having read your comment.

What countries did they come from?

I can believe that it took you many books to learn to read, since you are rather bad at understanding the main point of a post. I will repost it for your convenience.

LLM use significantly more data to learn how to read than humans do.