The artificial general intelligence crew gets this completely wrong too.

Just add more compute power and you'll get intelligence when we don't really know what it is

underneath that makes this creative and allows us to come with good explanations.

People talk a lot about GPT-3, the text matching engine that OpenAI put out, which is very

impressive piece of software.

But they say, hey, I can use GPT-3 to generate great tweet.

Well, that's because first, as a human, you're selecting which tweets out of all the garbage

that it generates are good.

Second, it's using some combination of plagiarism and synonym matching and so on to come up with

plausible sounding stuff.

But the easiest way to see that what's generating doesn't actually make any sense is just ask

it to follow a question.

Take a GPT-3 generated output and ask it why?

Why is that the case or make a prediction based on that and watch it completely fall apart?

Because there's no underlying explanation.

It's parroting.

It's a brilliant, Bayesian reasoning.

It's dating from what it already sees out there generated by humans on the web.

It doesn't have an underlying model of reality that can explain the scene in terms of the

unseen.

And I think that's critical.

That is what humans do uniquely that no other creature, no other computer, no other intelligence,

...