3 ms·

I'm open to the possibility of achieving an artificial intelligence, but for other reasons. Most of the people talking about AGI now are assuming it is possibl

by GTP 3y ago

I'm open to the possibility of achieving an artificial intelligence, but for other reasons. Most of the people talking about AGI now are assuming it is possible to achieve that with today's computer architectures, maybe just more powerful than what we have now. But to be fair, the fact that our body is intelligent says very little about the possibility that also a PC could be intelligent, as their inner workings are extremely different. To be sure that a computer can be intelligent, you have either to build a computer AGI and show that it is intelligent or you have to prove that intelligence is computable.

burnished 3y ago
I make no claims about the hardware it would run on, simply that our bodies are evidence that intelligence can be evoked from a system that fits largely inside the human skull, certainly no larger than the human body. That alone is sufficient evidence that the goal of creating a recognizably similar intelligence is not impossible. Also, most of the people talking about AI and intelligence don't know jack from shit. Other people have wondered if the differences between computer hardware traits and brain traits (one has relatively few but highly reliable and completely integrated units, the other is the inverse) is crucial and have been working on neuromorphic hardware. Other groups have focused on modeling and simulating at various levels with the goal of creating medically relevant models (one day), with stunning results in modeling specific neuron's firing behavior, and interesting if not at all medically plausible simulations of the behavior of groups of neurons interacting and performing tasks. Talking about how smart a computer chip can be is irrelevant when the hardware is instead designed to function like neurons do. A simulation of intelligence that performs by doing what a brain does in the way that a brain does it has a strong argument for being akin to intelligence - especially if you construct the simulation such that it must deal with human experiences, such as persistent embodiment in a stimulating environment with multiple competing needs.

by GTP 3y ago

burnished 3y ago

I make no claims about the hardware it would run on, simply that our bodies are evidence that intelligence can be evoked from a system that fits largely inside the human skull, certainly no larger than the human body. That alone is sufficient evidence that the goal of creating a recognizably similar intelligence is not impossible. Also, most of the people talking about AI and intelligence don't know jack from shit. Other people have wondered if the differences between computer hardware traits and brain traits (one has relatively few but highly reliable and completely integrated units, the other is the inverse) is crucial and have been working on neuromorphic hardware. Other groups have focused on modeling and simulating at various levels with the goal of creating medically relevant models (one day), with stunning results in modeling specific neuron's firing behavior, and interesting if not at all medically plausible simulations of the behavior of groups of neurons interacting and performing tasks. Talking about how smart a computer chip can be is irrelevant when the hardware is instead designed to function like neurons do. A simulation of intelligence that performs by doing what a brain does in the way that a brain does it has a strong argument for being akin to intelligence - especially if you construct the simulation such that it must deal with human experiences, such as persistent embodiment in a stimulating environment with multiple competing needs.