Hacker News Reader

Comment by binarymax | Hacker News Reader

Ugh. You can’t say “strides towards consciousness” until you define consciousness AND have a metric that can provide objective measure of such things.

View on HN

I don't see why this is true. Just because a thing is difficult to precisely define, doesn't mean it doesn't exist, nor does it mean you can't get closer to it.

I could say that I'm "taking strides towards becoming a better person" and that could still have meaning, despite the lack of any objective measure and, even, with the understanding that "a better person" could change over time.

I would argue that "taking strides towards becoming a better person" is a totally meaningless statement. If I were to ask you "how so" then you would probably respond by listing definitive actions that demonstrate attributes of goodness, like generosity or kindness.

I disagree. You could also characterize it as, for instance, having better knowledge of yourself (which, interestingly, seems to be a characteristic of consciousness as well).

Having better knowledge of yourself is a definitive action, no? I would call it self-awareness.

Replace "consciousness" with "gobbledygook". Does it exist? Can you get closer to it? Can't apply science to it.

I think you are saying we have a rough idea what it means, but then it's a sliding scale as far as ability to get closer to it.

I don’t see how consciousness is at all hard to define. A system that is aware of itself.

Checking for consciousness is harder, but the rule of thumb should be anything that makes a highly subjective case it’s conscious probably is.

In philosophy of mind, self-consciousness is separate from qualia, although some might argue that to be self-conscious, a being must have qualia. Qualia meaning what it's like to be in some mental state. Seeing a color has a color quale, which means there is an experience of color. Or better yet, we experience a colored-in world when we see (or dream, remember, visualize). Does any current AI have some mental experience?

For us, there's something it's like to be self-conscious, just like there's some experience that goes with our thoughts, such as inner dialog (hearing your thoughts). For some, that might be mental imagery instead of internal sound.

At any rate, it's the various feels that make up our experience of being a self in the world, or a self reflecting on itself.

>I don’t see how consciousness is at all hard to define. A system that is aware of itself.

A thermostat that is aware of the temperature inside its own housing is "aware of itself". It's not conscious however.

Nailing down exactly what consciousness means is remarkably difficult.

Honestly I think this is largely because humans want the conclusion that only they are consious, or at least that only they and some animals are, and are trying really hard to carve out a definition that makes that work. Why would we do that? Well obviously because we create AI to do work for us and we don't want to have any ethical considerations get in the way of treating our new slaves however we want.

Yup this is my thinking also. It isn’t a hard concept imo people just like theologizing about why they’re special

I would think that...

1) awareness, even without a sense of self, would still count as consciousness, and

2) describing consciousness in terms of awareness doesn't help to define it, since awareness is a subjective experience that cannot be shared directly

I think consciousness probably requires pleasure/pain, and no, just setting negative/positive values in training does not mean pleasure/pain is involved.

The best a neural network could is pretend it is conscious (I'd have no trouble turning it off). Would artificial consciousness be useful outside of some kind of companionship app (eg. replace my dog)?

Consciousness means you know you exist. This is a super simple definition and GPT does not, nor do any of these AI models, meet this very simple definition.

Yea, but then you have to define what "knowing you exist" means... Does it mean you have a model of the world and you know your place in it, meaning that you can predict what would happen if you would do this-or-that in your current state? That doesn't seem impossible for an AI and you could argue some of the current reinforcement AIs have a model like that and act according to that but it's not as accurate and complex as humans have. So maybe it's a spectrum depending on how accurate your model is and we just don't have a human level consciousness just as we don't have human level intelligence yet.

Not really, knowing yourself to be an entity is pretty straightforward unless you want it to be a philosophical argument.

my program that prints "hello world" to the screen is then conscious?

Does it make a convincing case?

"My infant child is making strides toward consciousness"

Completely uncontroversial, yet we don't have a metric and we haven't defined it precisely.

>Completely uncontroversial, yet we don't have a metric and we haven't defined it precisely.

but we have witnessed countless infants grow up and qualify as what we consider to be 'conscious'.

we haven't yet seen that jump in AI, thus we have no way of knowing the distance to there from where we're at.

I and many others strongly disagree with the statement’s assumption that infants are not conscious, so it’s not a completely uncontroversial statement. Sure, they have very limited I/O, but I understand that as orthagonal to consciousness.

Replace "infant" with fetus then...

Still not uncontroversial: https://pubmed.ncbi.nlm.nih.gov/31922622/

It's obviously not controversial. Your linked article doesn't claim that a 0-week old Fetus is conscious. It claims you can define consciousness in such a way that a 24 week old fetus is conscious. That 100% completely reinforces my point, which is that we are in agreement that a 0-week fetus moves toward consciousness (your article claims it achieves it by 24 weeks). We are able to agree on that without agreeing on the definition of consciousness, so the lack of a unanimous definition doesn't prevent us from pointing at AI and saying "this is moving toward consciousness".

By your logic, it would be wrong to say that OpenAI is closer to consciousness than a piece of lumber.

However, your comment is a good reminder that non-rigid thought patterns are not a prerequisite for consciousness. (How apropos your username!)

Which I wonder if we'll ever have as consciousness is a strictly subjective thing.

I've always liked the definition of making sense to ask what it's like to be something as an indicator of consciousness.

For example, you can try to imagine what it's like to be a dog, but it doesn't make much sense to ask what it's like to be a toaster.

What's it like to be an artificial neural network? I don't think that's valid, nor is there any good evidence that we're making progress.

> consciousness is a strictly subjective thing.

I’m wondering whether that can be taken as an objective truth.

Reply on news.ycombinator.com