We totally do emulate organisms on that scale. The real challenge is simulating the sensory inputs and the feedback loop between the outputs, the environment as the body acts, then new inputs.
Disembodied simulations of nerual networks don't work. They are part of a body, an environment, and all the feedback loops that come with it.
It sounds like you really just want to see a ML algorithm have a body to learn in. Why we ever expect AGI to happen without letting an ML algorithm learn by interacting with a "real" reality seems strange to me. By all means, keep making glorified optic nerve and expecting them to "wake up".
You don't need sensors, you just need a virtual room.
> We totally do emulate organisms on that scale. The
There is no evidence those emulations actually emulates those organisms. They just built a neural net in the same structure and assumes the cells doesn't matter. But cells are really smart and can navigate environments on their own, they are intelligent beings in their own right, and building a flea using a thousand of those is very plausible compared to doing it using neural net of similar size.
And yes, in order to prove that we actually emulated those you need to show that it does the same things in the same scenarios. You don't even need to do everything, just a simple thing like being able to move around, gather material and build a home in a physics engine would be huge.
> You don’t need sensors, you just need a virtual room.
While technically true, I actually think this is way more difficult than it sounds, bordering on practical impossibility.
I think the other commenter was making a really important point. The simulated environment would need to be incredibly rich, to a point as to almost defy imagination.
Consider what happens to a human mind when confined in a box (prison) with limited opportunities for stimulation. There’s a room, a gym, other people with which to socialize, food, walls, an outdoors enclosure... And yet someone who spends their entire life in this type of environment will certainly be facing serious neurodevelopmental issues.
For human/mammal order of AI, I would even argue that simulating adequate inputs might actually be a more difficult problem than building the AI that responds to them!
Also note, organisms not only navigate the environment, they interact with it, handle their own capture and consumption of energy from it,and reproduce. Autonomously.
We totally do emulate organisms on that scale. The real challenge is simulating the sensory inputs and the feedback loop between the outputs, the environment as the body acts, then new inputs.
Disembodied simulations of nerual networks don't work. They are part of a body, an environment, and all the feedback loops that come with it.
It sounds like you really just want to see a ML algorithm have a body to learn in. Why we ever expect AGI to happen without letting an ML algorithm learn by interacting with a "real" reality seems strange to me. By all means, keep making glorified optic nerve and expecting them to "wake up".