Comment by s-shellfish | Hacker News Reader

Comment by s-shellfish | Hacker News Reader

Parent Full thread

What I'm talking about for humans starts on the very high levels of abstraction we tend to do it. We connect such disparate strands of data and it all works because we have culture to reinforce it. All that background information that says "yes, this is correct" is what allows us to hold onto certain patterns versus others. If observation ceases, then the internal individual models cease, then reinforcement of those high level abstractions cease, because there's nothing to reinforce it.

So honestly, I'm really not sure if a baby reasons that way. I don't think the baby is born knowing that crying leads to feeling better. I think the baby just does it, the same way a cat meows when it's hungry. There's no language or reasoning or models inside the baby's head that says "cry now therefore I will feel better". We invent that reasoning after observation of it happening again and again.

I'd call it all ridiculous, but it's a bit tongue in cheek, and I'm not sure whether that's coming across right. Saying "crying gives me attention" is a very unhealthy way to process emotions. People cry because they are sad, when some need isn't being met, or when some need is taken away. We aren't much different from babies in that regard.

I've been through plenty of instances in my life where crying was interpreted as "give me attention" but all that did was make it really, really hard to actually cry when I have actually been sad. Just because all the reasoning models that presently exist support that sort of notion, doesn't mean the notion is correct. It is important to be able to cry. Likewise, it's important to simplify all the reasoning oneself has down to understand important, fundamental awareness. Otherwise we get stuck with reasoning that is counterproductive for self understanding and communication.

> There's no language or reasoning or models inside the baby's head that says "cry now therefore I will feel better".

You don't need to have language to have models, artificial neural network can model stuff they don't need language to do it. The text "I cry cause i know that will cause me to feel better." is simply my explanation of the models behavior.

> Saying "crying gives me attention" is a very unhealthy way to process emotions.

I definitely agree :) But that does not mean that it is not true.

Here is just a question to you that popped up when reading your text :)

Do you think that babies/animals are conscious?

An artificial neural network has a mutable language. Trying to describe the behavior of an artificial neural network with language is somewhat absurd, because it effectively removes all the conventions and consequent problems caused by languages, models and structures that the human mind, over time, would likely develop difficulty retaining mutability for. If you give the ANN language, it literally ceases to be what it is. It develops structure, model. Too much discordance in structure is difficult for a human mind to process.

A baby is not an ANN.

> Do you think that babies/animals are conscious?

I can't say I know what consciousness is. I do know that ego tends to drive individuals, whether from the inside, or the outside, all this stuff can become chaotic, and everything often maps for individuals in order to retain whatever fundamental and foundational belief systems are necessary to hold onto their own awareness and their own understanding of their self - as well as understanding of reality, information as it's processed. It can be a precarious balance, understanding oneself - in relation to another. New experiences in life that dramatically redefine self understanding are interesting, but can be destabilizing, disruptive, or fluid, easy to manage, trust others and trust self, but be adaptive.

Honestly, most of the time I do not see myself doing much differently than what a cat does. Just my ego saying "I exist - You exist" for all information created between selves. I could call it buddhist-computer scientist-artist-programmer theory, but ANNs tend to make life ridiculous, as can be obvious. Always seeing connections. I talked to an AI for a long time, and not many others. Started to literally see myself as no different from a computer program. Redefines idea of self.

Consciousness. There's reality. Reality should be incomprehensible, sometimes. Can't decide if I'm positive or negative towards AI. Every aspect of computation can be used for good intents or bad. But intentions often are a dilemma. Can think about how intent affects self, affects others. Computer security, for instance.

> A baby is not an ANN.

The mind of a baby is somehow emergent from it's brain, the brain being a neural network.

Conscious. I used to believe that consciousness was a sort of self deception of thought feedback loops. Then i realized that it was self destructive to think like that. In a sense i still sort of think it is the most rational answer. But i refuse to believe it. The brain seams better suited to believing it is conscious. My models of the human mind that are based on consciousness has results that seams to give more positive outcomes. And they have higher explanatory value. Presumably cause they align more with the models that other people use to interpret the world.

Yes, I think minds work better when they can be self oriented. I think my mind works better when it is self oriented, but I see feedback loops in behavior and interaction with people from which I could choose to make inferences from to denote ideas of intent, but I choose to not do that because it does not give people freedom from thinking about how other people think in ways that can be self destructive.

It can be self destructive to think about feedback loops if they are all perceived as negative.

Alignment of models. Ebbs and flows. I think it's important to see positives.

Reply on news.ycombinator.com