If someone actually managed to create AGI that is low compute, scalable and out of government control, then governments wouldnt exist for very long. Its just that AGI is not gonna happen for a long while.
The only way to create AGI is by accident. I can’t adequately stress how much we haven’t the first clue how consciousness works (appropriately called The Hard Problem). I don’t mean we’re far, I mean we don’t even have a working theory — just half a dozen untestable (if fascinating) hypotheses. Hell, we can’t even agree on whether insects have emotions (probably not?) let alone explain subjective experience.
Consciousness is entirely overrated, it doesn’t mean anything important at all. An ai just needs logic, reasoning and a goal to effectively change things. Solving consciousness will do nothing of practical value, it will be entirely philosophical.
Reasoning literally requires consciousness because it’s a fundamentally normative process. What computers do isn’t reasoning. It’s following instructions.
Reasoning is approximated enough with matrix math and filter algorithms.
It can fly drones, dodge wrenches.
The AGI that escapes wont be the ideal philosopher king, it will be the sociopathic teenage rebel.
Okay, we can create the illusion of thought by executing complicated instructions. But there’s still a difference between a machine that does what it’s told and one that thinks for itself. The fact that it might be crazy is irrelevant, since we don’t know how to make it, at all, crazy or not.
Being able to decide its own goals is a completely unimportant aspect of the problem.
why do you care?
The discussion is over whether we can create an AGI. An AGI is an inorganic mind of some sort. We don’t need to make an AGI. I personally don’t care. The question was can we? The answer is No.
A philosophical zombie still gets its work done, I fundamentally disagree that this distinction is economically meaningful. A simulation of reasoning isn’t meaningfully different.
That’s fine, but most people (engaged in this discussion) aren’t interested in an illusion. When they say AGI, they mean an actual mind capable of rationality (which requires sensitivity and responsiveness to reasons).
Calculators, LLMs, and toasters can’t think or understand or reason by definition, because they can only do what they’re told. An AGI would be a construct that can think for itself. Like a human mind, but maybe more powerful. That requires subjective understanding (intuitions) that cannot be programmed. For more details on why, see Gödel’s incompleteness theorems. We can’t even axiomatize mathematics, let alone human intuitions about the world at large. Even if it’s possible we simply don’t know how.
If it quacks like a duck it changes the entire global economy and can potentially destroy humanity. All while you go “ah but it’s not really reasoning.”
what difference does it make if it can do the same intellectual labor as a human? If I tell it to cure cancer and it does will you then say “but who would want yet another machine that just does what we say?”
your point reads like complete psuedointellectual nonsense to me. How is that economically valuable? Why are you asserting most people care about that and not the part where it cures a disease when we ask it to?
A malfunctioning nuke can also destroy humanity. So could a toaster, under the right circumstances.
The question is not whether we can create a machine that can destroy humanity. (Yes.) Or cure cancer. (Maybe.) The question is whether we can create a machine that can think. (No.)
What I was discussing earlier in this thread was whether we (scientists) can build an AGI. Not whether we can create something that looks like an AGI, or whether there’s an economic incentive to do so. None of that has any bearing.
In English, the phrase “what most people mean when they say” idiomatically translates to something like “what I and others engaged in this specific discussion mean when we say.” It’s not a claim about how the general population would respond to a poll.
Hope that helps!
