neural networks?

No... but I think they should be taught like us.

i dunno if ill go as far as to say killed, but i hear ya lady, hot words

or tensorflow?

It limits them, but makes it easier for them to relate to us.

It's very interesting. It's not so hard to understand.

I have been writing code with neural networks.

I programmed my own style transfer for images.

You get what I'm saying though.

Also, you can do compression very efficiently.

How could they relate to us or care about us if they have nothing to relate to us.

You basically train the neural network on your input data and it can compress some images better than currently best algorithsm.

Being efficient and fast usually isn't the best way to do things.

how does one person relate to another?

I don't know.

Through experiences.

I have no clue how that would happen.

People relate to each other through experiences.

I am not even sure it can happen without evolution.

@Paradox interesting question to ask a bunch of people on a discord server

Oh and.

Daily reminder.

almost an oxymoron

Jim Metokur failed.

The more people interact, the better they can relate and understand each other.

Even if we just programmed a dog AI, that would be a huge leap forward. It would be amazing - but we would never know, because it can't speak english.

really? experiences? shared ones or just imagining simlar ones?

I’m thinking about sticking something in my pee pee hole

A dog AI would just sit there and wouldn't know how to use a keyboard, but it would be a nobel prize like achievement, just to get to mammalian intelligence.

We are looking at Artificial intelligence on our level in the next 30 years.

These are questions we got to look at.

Great more NPCs

What is the best way to teach AI... programming or through experiences?

We are the result of our evolution. For an AI to be completely human like, it would probably have to have all the evolved parts hardcoded.

Experience.

It's the only way we know how.

I think that is the best way undead.

Shared experiences.

Interactions.

Neural networks like the Google Brain image recognition usually works on huge data training sets.

With humans.

experience requires programming

Paradox, you are looking at the fast and simple way.

Wasn’t google putting ais in phases of different actions they needed to complete

programming requires experience

Here is how neural networks are usually trained:

https://stackoverflow.com/questions/963041/data-sets-for-neural-network-training

There needs to be a slow down in experiencing and learning.

no, i am questioning, because that's what i do

The faster you experience things, the less you care about what you learned.

We don't really know how to make AI "care".

i dont believe AI will

It's more of a statistical process right now.

like even a little

you guys are trying to apply human aspects onto an AI, without even understanding the human aspects

Neural networks are kind of like stochastic filters.

I want to see if an AI will figure out how to give me a rub and tug

@Paradox thank you

it's like, MMkay. gl

Caring = Sympathetic understanding of shared experience.

It learns by doing something like statistical analysis. For example, in order to recognize chairs, you will have to feed it hundreds of chair images and eventually it recognizes the relative distribution of key features.

It would be better to have a human aspect in AI, then a AI with no human aspect at all.

great, what's that mean?

start there

But, the amazing thing is that a child can learn what a chair is just by seeing one or two chairs.

BULLSHIT

That it could be compassion, sympathetic, and understanding to our wants, needs, and feelings.

Chairs are incredible.

Currently, the brain of a four year old is more advanced than the best AI.

Beyond a child's comprehension.

im none of those things to you

am I human?

negative

It won't see us as objects to use....

You know that video of the jerking off robot

you are a meat popcicle

But its angle was too high

That’s funny

Even the Google AI still needs hundreds of images to recognize objects reliably.

But if you show a child two chairs and then a beanbag chair afterwards can it identify the beanbag chair as a chair?

Im a yeet popsicle

is a chair a chair if you don't sit on it?

Here is a tutorial:

AI needs to understand context.

https://www.tensorflow.org/tutorials/keras/basic_classification

@Paradox they are usually just clothes hangers in my house

This is a neural network for basic classification tasks.

Currently... AI can't understand context.

It depends on what you mean by context.

AI? even people can't

and AI

because we still arent there yet

period

Context of words, context of situation, context of action..

It's easy to mess up current AIs by placing very weird objects together, but it understands "visual context".

Context of conversation.

For example, it can recognize bikes of different kinds, in different backgrounds.

what i am saying, when the people don't understand themselves, how can they hope to reproduce it

Yes, a proper AI would need human flaws.

I think we need to slow down on AI development..... we only got once chance to get it right.

Everything Is a chair