I do not have any intentions to delegate my agency.

Qualia Journal

I don't use ChatGPT and other LLMs for having a psychological tête-à-tête about my life.

I predominantly use LLMs for interacting about science and technology. Recent favorite topics include quantum mechanics and information thermodynamics.

In general, I think LLMs are great at compressing and analyzing information. I don't appreciate their agency too much.

That's why I am not panicking about Moltbook. I rather might panic about people who panic about Moltbook. I am observing OpenClaw with interest, but do not have any intentions to delegate my agency.

LLMs might exhibit cognitive supremacy, but judgments and choices would remain human, for the time being. 

Even when you are consulting about your life, it would be great to use LLMs for information compression and analysis. Communications between humans would be central even in the era of LLMs.

We don't know natural language yet. Qualia Journal

Elon Musk predicts that we will not need natural language any more due to advancements in artificial intelligence and interfaces such as neuralink.

That may well come to pass, as anything predicted by the genius entrepreneur. However, for the time being, it appears that natural language has gained more power because of LLMs.

The general idea used to be that natural languages were inferior to mathematical languages, because they are soft and ambiguous. However, LLMs, based on next token prediction, have shown how powerful natural languages are in terms of establishing concepts and world models.

In a sense, natural languages might be superior to mathematical language.

Interestingly, we do not know exactly why this is the case.

We don't know natural language yet.