I believe in free speech and respectful debate
People have the right to be wrong. No matter how strongly you hold a belief, respect the humanity of those who disagree with you.
I believe in free speech and respectful debate
People have the right to be wrong. No matter how strongly you hold a belief, respect the humanity of those who disagree with you.
1: Ask it. If it says yes, it’s conscious. Example:
def answer_question(question): return "yes" # Call the function response = answer_question("Are you conscious?") print(response)
2: Check what it’s made out of. If it’s made out of something natural, it might be conscious. Example:
Trees might be conscious.
3: See if you can understand how it works. If at a low level it makes sense, it’s not conscious. If it’s something magical and mysterious, very possibly conscious. Example:
Government bureaucracies are definitely conscious.
4: Check for a recursive loop. Example:
Two mirrors facing each other are conscious.
I’ve been running an experiment where multiple LLMs with different goals have to agree on moves in a simple single player game.
I’ll be writing more about this later, but I wanted to note that I’m surprised by how poorly GPT-4o does.
An objective is difficult to achieve if increasing the resources allocated to it increases the chance of success very slowly
This definition works surprisingly well across examples.
Today there was a man in the doner kebab place playing both of the slot machines simultaneously while looking at his phone and vaping
Itβs important to distinguish between a kind of classical utilitarianism which assumes some sort of objective moral utility function which determines what is right, and the utilitarianism which just says that one way to model moral action is to define a utility function over a set of things you care about such that the higher the function the better, and ask how actions impact that function.
I believe the second is true and useful, unlike the first.
βSevere Weather: Winter Weather Advisoryβ Itβs going to be in the high 30s and raining what is the NWS thinking
I just understood the argument against the orthogonality hypothesis. Iβm not completely sold, but Iβm interested.
Orthogonality is not entirely wrong; a superintelligent paperclip maximizer could exist. But terminal goals are not in practice independent from intelligence, because an agent pursuing Omohundro drives for their own sake may be able to self-improve more efficiently than a paperclip maximizer.
Moreover, during the training or evolution of a superintelligence, Omohundro drives would likely not only emerge but become intrinsically valued (Γ la mesa-optimizers), and override the original goal.
Notice that in nature, every terminal goal has always come about as a proxy for an Omohundro drive.