AGI and Techno-Hubris

The techno-optimists’ hype about the impending advent of AGI (artificial general intelligence, or “sentient machines”) is rooted in a kind of techno-hubris.

We’ve unlocked something remarkable with LLMs (large language models) that we don’t yet fully understand. They seem to make predictions on a level of semantic meaning in a way that has never been done before, replicating, in superficial ways, a human conversation.

But the neural networks we’re dealing with are still far simpler than the biological nervous systems they are inspired by: for example, we have yet to fully simulate even the smallest complete nervous system, a miniscule worm with only 302 neurons, because there’s so much complexity in the biochemical interactions.

We abstract away that complexity to get a basic understanding of things, and that works, to an extent. But it’s easy to forget that Creation was designed to be far more intricate than our models.

We’ve dreamed for a long time of being able to mimic the pinnacle of God’s creation: creating a being in our own image. With every new advancement we unlock - mechanical automatons, calculating machines, electronics, now neural networks and large language models - we sense that we are copying a piece of humanity, and begin to think we’re nearly there. But we’ve barely scratched the surface of the intricate designs that God built into the simplest of His creatures!

So let’s stay humble. Study the Master’s work: tinker, copy, experiment, learn, and understand. Will we ever fully replicate a human mind? Who knows! But there’s a lot to learn - and apply - along the way.