@metapsyche literally every observable process is a flow of energy, so the headline and the conclusion basically admit they know nothing. They can just barely tell if you are passed out or not sometimes.
@freemo Isn't thought the defining part of consciousness? Boiling your statement down to "we are programmed to think that we can think"... well then we are programmed right... how is that an illusion?

Why would anyone attempt to decipher the brain of a human being before even being able to do it for a much simpler organisms? I guess "look guys I modelled the nervous system of this worm in its entirety" is not as appealing as "look guys I figured it all out, cancel philosophy, check mate atheists!".
https://github.com/openworm/OpenWorm

@namark

Well yes thought = consciousness, but it has the same problems. If thought just means you can reason about things, then I'd say even basic computer AI is conscious by that definition.

Most people like to think computer AI's doing facial recognition or solving other problems arent conscious and humans are. I'd say in terms of consciousness there is no real difference, one is simply more complex than the other and is genetically programmed to think it has some special thing called "consciousness" that other computers, organic or not, might lack.

@metapsyche

@freemo subtle difference between
consciousness = think
and
consciousness = think that can think

The latter makes more sense to me and is clearly a correct thought. If you think something, then you can indeed think, so that something being "I can think" is just correct, and not an illusion.

The thought itself is a mystery, and is what makes consciousness a mystery as well. Calling it ability to reason smells like another circular definition. Neither mathematics nor physics have a conclusive answer on weather it is or can be equivalent to a machine in principle.
Pretty much the only thing we know for sure about it is that brain chemistry affects it, so that's our lead, and simpler forms of life are our best clues.

@metapsyche

@namark
So if a chat bot says to you "I can think" (something that wouldnt be unusual for a chat bot) is it therefore conscious?

@metapsyche

@freemo no, that is just sophisticated parroting, which unfortunately again draws way more attention that actual work on AI.

If the chat bot would develop it's own language, and teach it to me, and then convince me that it is suffering from the same condition that I call consciousness, I will have to assume it is indeed the same. Still I would not know what it is (I guess I would if I wrote it, but lets face it, at this stage it would just be a neural net, and I would have no clue, and it would be just as big a mystery for me as my own brain).

@metapsyche

@namark

You've never created your own language and then taught it to me to convince me you suffer from consciousness. By that measure I have to assume you arent conscious either and just capable of sophisticated parroting :)

@metapsyche

@freemo in case of me (or any other, whom you believe to be a human) you have preemptively assumed it, cause you had no reason to doubt. Just like you don't constantly worry that the ground is going to fall under your feet when you walk, or that the next object you drop is going to fly up. You can not definitively prove to yourself that I can think, it's just your best bet and only hope. @metapsyche

@namark

But I dont preemptively assume it. I dont even think **I** have consciousness (just the illusion of it) so why should I assume any other human does?

Now if you provide some sort of a test (like the one you just mentioned for the chat bot) I can happily apply that to you and other creatures/inventions. Though I'm not sure your "test" has anything to do with consciousness, its just a test for what creatures are capable of developing language and communicate, nothing more.

@metapsyche

@freemo You assume thought process that is similar to yours. Otherwise all you communication is pointless. I already tried to explain why consciousness is not an illusion. Which part do you disagree with?

I guess from your assertion, consciousness = thought (which I disagree with), and consciousness is illusion, follows thought is illusion? That seems pointless, whether it is an illusion or not, it is what it is. I don't know what it is, so there is no conclusive test I can devise. I simply brought an example of what would make the chat bot mysterious enough for me, to justify the assumption that it can also think. The own language part was to exclude random success by parroting, the convince part is more relevant fundamentally.

@metapsyche

@namark

The problem is that consciousness is a nebulous term that isnt well defned in any meaningful way and certainly has no way of being objectively verified.

Thought is different, that is a viable term and if we go by that definition I have no reason to think a chat bot doesnt have thought, just of a simpler nature

As for assuming you have thought just because I have thought.. Well if you leave it to me I'd say anything that can speak, even something that can just parrot things, but do so in a way that is reactionary to its environment has some level of thought, so it really doesnt take us very far. But just because we can communicate doesnt really imply you are any more or less capable of thought than me either. Its entirely possible the way I think is entirely unique and the only thing we truly have in common is the way we communicate.

@metapsyche

@freemo I though the definition of consciousness as "thinking that I can think" was satisfactory. It is still nebulous though by the virtue of thought being nebulous. I have never seen a definition of thought that would not be in some sense circular. In my opinion that is where the actual mystery is. You of course can conjecture that this seemingly common notion of thought that we have is just a complex mechanical computation, but you still can't define the thought itself unless you somehow demonstrate said complex computation, not just the simpler version. You can make generalization in theorems such as "anything that is more complex than this is such and such" but that is still not a definition. Just like saying that an elephant is made of atoms, doesn't really define the elephant, not in any useful way at least.

And yes, we might think completely differently and just appear same to each other in communication, but that is not the best bet, or the most reasonable guess. We tend to prefer consistency in our assumptions.

@metapsyche

@namark

"Thinking that I can think" is not so much adequate as a definition as it is simply something we can actually discuss and pick apart

Simply put, if I cant tell what someone thinks, which Ic ant, then I also cant tell if someone thinks they can think.

Communication isnt a prerequsite, but even if they can communicate I have no idea if their words are valid or not. Just because you claim that you think you can think doesnt mean you actually do, as you said you might just be parroting those words.

In fact by that definition the only thing I can say is that I am conscious because I seem to perceive an illusion of thought in my own brain. But even if i beleive an illusion then I still "think i can think", it just, isnt all that useful really once you break it down.

@metapsyche

@freemo Yes, but still don't get why you call it an illusion, but I guess that doesn't really matter much.

@metapsyche

@namark

Well being able to to "think that I think" is little more than being able to observe ones owns thoughts. You can effectively get that by adding a recurrent layer to an ANN, its own thoughts are fed back into its own perception. Such a network just isnt always complex enough to express that, but that isnt part of the criteria anyway.

The illusion comes from the fact that we are genetically wired to perceive that as something special in the first place, as if it gives us some unique almost spiritual quality we call consciousness. Its literally just intermediate processing data wired back into the input (perception).

@metapsyche

@freemo We don't exactly perceive or observe every single thought though, right? I would imagine it would be much easier to define them if we did (or at the very least we would know our clock speed... do we have a clock speed? I think the article in OP had something about clock speed... I believe it yes... of course... hard science). On the other hand to observe one's thought would be an infinite loop wouldn't it? The machine would be stuck continuously observing itself observing itself observing itself... though I guess it's a machine and it can cheat, accumulate several thought and then observe them with one thought (hmm, one thought, sure, that's a thing for us too, no doubt). Or observe a thought and do something else in one thought. Still it will be forever behind itself, unaware of its current thinking.
@metapsyche

@namark

Except that in both neural networks and biological ones there is a loss of information each time through the loop. Sure I can perceive myself thinking, and I can even perceive the fact that I perceive myself thinking but just like in a recurrent network the more layers down the rabit hole we go the more abstract it gets until we get to a point where we just recognize this very fact (As we are doing right now).

@metapsyche

@freemo I was sarcastic claiming to be sure that human mind has a clock speed, a discrete measure of thought and a loop. If you are convinced in that, then well, I guess that's a most reasonable assumption of sorts. Sounds a little overzealous though.

As far as I know the Information loss/reduction is not through the loop, it is through the layers, meaning each step is reducing not the loop itself. That reduction doesn't matter cause each step itself (the way I understand your analogy) is a whole new thought that is in no way reduced. It will need to be fed in as input entirely, or reduced by some other means. Maybe another network? If yes does the operation of that network constitute a thought? What's going to reduce that? In other words, what would you define as a thought of this machine?

@metapsyche