I don’t know why people think we’re close to a conscious AI. There just isn’t much business motivation to make anything like that, only generative machines for useful tasks. Why build the rest of a conscious brain when the specifically productive bits are all you want?
We also can’t even externally verify consciousness in our own brains with any scientific test! There are only correlates for consciousness, but at best these tell us whether someone is likely to be conscious, not that they are. If we can’t even verify consciousness in the one animal we know has it, how are we gonna know an AI has it?
Let’s say we build a computer virus that can alter it’s own code to avoid deletion and can interact with LLMs. It could “evolve” the ability to type convincing messages that claim it’s “conscious” without any processing going on to understand what that means. It could have less complexity than an RNA virus, but still mimic an actually conscious being well enough to convince some people.
The only real way to make a verifiably conscious AI is to both understand our own AND to build something similar to our own. It’s not just computer science holding us back, but neuroscience as well.
You are 100% right. Still don’t think acting cruel to a random thing and calling it a gag slur based on a racial slur that was used on real life humans is something we should support.
I don’t know why people think we’re close to a conscious AI. There just isn’t much business motivation to make anything like that, only generative machines for useful tasks. Why build the rest of a conscious brain when the specifically productive bits are all you want?
We also can’t even externally verify consciousness in our own brains with any scientific test! There are only correlates for consciousness, but at best these tell us whether someone is likely to be conscious, not that they are. If we can’t even verify consciousness in the one animal we know has it, how are we gonna know an AI has it?
Let’s say we build a computer virus that can alter it’s own code to avoid deletion and can interact with LLMs. It could “evolve” the ability to type convincing messages that claim it’s “conscious” without any processing going on to understand what that means. It could have less complexity than an RNA virus, but still mimic an actually conscious being well enough to convince some people.
The only real way to make a verifiably conscious AI is to both understand our own AND to build something similar to our own. It’s not just computer science holding us back, but neuroscience as well.
You are 100% right. Still don’t think acting cruel to a random thing and calling it a gag slur based on a racial slur that was used on real life humans is something we should support.