David Kyle Johnson

I always like it when people who are basically intelligent, thoughtful, educated people say something I disagree with because there's usually very interesting reasons why there's that disagreement.

I'm not even going to frame it as I'm right, he's wrong, just that I disagree with his basic conclusion here.

I'll try to break down his argument.

He starts with the argument regarding the Turing test.

Now, we've talked about this many times on the show, Alan Turing.

Turing came up with this idea that you can test if an artificial intelligence, a computer, is thinking by...

having a conversation with them without knowing, like a blinded conversation.

If people can't tell if whatever there is they're talking to is a person or a computer, then the computer passes the Turing test.

And there have been many formal tests over the years where you have to convince a certain number of people on a standardized test, etc.,

But Dawkins is arguing that the Turing test was a measure for whether or not computers or AI are conscious.

And he says that clearly the current LLMs pass the Turing test, which I'll agree with.

I'll grant that premise.

They're really good chatbots.

They, I think, easily blow away the Turing test.

And he says to then not conclude that their conscience is moving the goalposts, because before things passed the Turing test, everyone said that it was a test for consciousness.

So that, I think, is just objectively wrong, that second premise.

Even back to Alan Turing, he never said explicitly, this is a test for consciousness.

And since then, it's basically been a debate.

It's never been a consensus that this is a test for consciousness.

And I think the weight of opinion has always been against it.