Yep, that is Friston’s free energy principle in a nutshell. Interesting to think of circuits as tiny simulations though. Then the brain is forced to develop smarter and smarter simulations until we can say something like “that guy knows how the world works”, meaning he’s good at predicting his own inputs.
And what works for Samuel L Jackson won’t work for the rest of us, so it’s still tied to our own actions and abilities, ie our “outputs”.
For more human-like AI, yes. But the easier path in the short term is just to have AI agents that “live” on the internet. All their input senses and output abilities would be internet-based, so it really would be like they live in a digital world reflecting the human world. Like Plato’s Cave, they would imagine they understand the world, but wouldn’t know what they don’t know.
Yep, that is Friston’s free energy principle in a nutshell. Interesting to think of circuits as tiny simulations though. Then the brain is forced to develop smarter and smarter simulations until we can say something like “that guy knows how the world works”, meaning he’s good at predicting his own inputs.
And what works for Samuel L Jackson won’t work for the rest of us, so it’s still tied to our own actions and abilities, ie our “outputs”.
For more human-like AI, yes. But the easier path in the short term is just to have AI agents that “live” on the internet. All their input senses and output abilities would be internet-based, so it really would be like they live in a digital world reflecting the human world. Like Plato’s Cave, they would imagine they understand the world, but wouldn’t know what they don’t know.