David Kyle Johnson

And again, it seems hard to imagine that without consciousness.

And then finally, I was writing about a current study that says that consciousness may allow for faster adaptation and learning.

So there are very specific evolutionary adaptations that would benefit from consciousness or may require consciousness.

So I think using as a premise that you need consciousness in order to be able to talk like an LLM, otherwise we wouldn't have evolved consciousness, is not a valid argument.

The bottom line is he doesn't present a coherent argument for Claude being conscious or that it needs to be conscious.

He misunderstands Turing, I think, misunderstands how LLMs work.

But I also think there's another layer here.

The final layer I'll talk about is that the illusion is so powerful because of human vanity.

And what I'm referring to is that we all think that humans are creative and original and

And that when we think of something, we're brilliant.

You know, that we all think we're being witty when we make the same joke that a million people have made.

You know what I mean?

And the fact is that human speech is mostly derivative.

Even things that we think are new or creative, whatever, it's mostly derivative.

And so LLMs have a really easy task mimicking human thought through speech because they're derivative too.

Mm-hmm.

It's actually a lot easier than we may think, right, to mimic human thought because most of what we say is derivative.

And if everything LLMs write or say are derivative of their training data, it looks real to us.

It looks human to us because it is human to be derivative.

And you also have the history.