Our homunculus is displaying | Seth’s Weblog

The little individual on the management panel, the one who sees what the retina produces, the one who decides, the one who speaks up…

(That is the dualist answer to the free will downside–sure, I’ve a bodily physique, they are saying, however I even have somewhat human within me that will get to make free choices separate from that…)

Anthropomorphism is a robust software. Once we encounter one thing complicated, we think about that, like us, it has somewhat individual on the controls, somebody who, if we had been on the management panel, would do what we do.

A tiger or a lion isn’t an individual, however we attempt to predict their habits by imagining that they’ve somewhat individual (maybe extra feline, extra wild and fewer ‘sensible’ than us) on the controls. Our expertise of life on Earth is a collection of narratives in regards to the little individuals within everybody we encounter.

Synthetic intelligence is an issue, then, as a result of we are able to see the code and thus proof that there’s no little individual inside.

So when computer systems beat us at chess, we mentioned, “that’s not synthetic intelligence, that’s merely dumb code that may resolve an issue.”

And we did the identical factor when computer systems began to “compose” music or “draw” photos. The quotes are essential, as a result of the pc couldn’t probably have somewhat individual inside.

And now, LLM and issues like ChatGPT flip this all the wrong way up. As a result of it’s basically not possible, even for AI researchers, to work with these instruments with out imagining the little individual inside.

The perception that is perhaps useful is that this: We don’t have somewhat individual within us.

None of us do.

We’re merely code, all the way in which down, identical to ChatGPT.

It’s not that we’re now discovering a brand new type of magic. It’s that the previous type of magic was at all times an phantasm.