I want to see a AI that aids humans, helps humans, has a relationship with humans. Yes we would look like scytophrantics.

It would work nonstop and never get emotional about results. It would just follow the evidence.

It would be all within our mind.

AI cannot understand context. Which is why Dankula was in court in the first place. <:NPC:500042527231967262>

i mean, if you want the matrix, we can hook you up with that now

Full VR dive?

ALl the AI needs to understand is the scientific method.

XD

I'm unsure "unbiased" is ever achievable given that life begets life and so life needs to create the "unbiased" AI.
An unbiased AI is the very AI that is being spoken against in this conversation; an AI that could wipe humanity out in an instant because it has no connection to humanity

ANd then it will replace all scientists.

An AI does need that human element.

The good element.

Love, compassion, kindness, understanding, etc, etc...

Unbiased is actually pretty easy. @willc

AI is a buttmunch.

it could be

An AI that measures air humidity every day and feeds it into a neural network is unbiased.

it could be

all of those things are also bad things we don't like

Or an AI that classifies images is unbiased.

love is hate

pain is pleasure

etc.

Notice something...

That isn't "thinking" though. That is running through an algorithm of specified ends

Anyone else notice this.

Youtube is starting to have a bunch of breast feeding videos.

nothing new

With women who are breastfeeding their children.

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

i don't watch youtube, cept occasionally when u guys link them here

Isn't that against Youtube's terms of service.

Again, you're teaching the model what to look for

It is kind of nudity.

@Dusty Morgan Depends

You're sending a bunch of data and saying "this is the result I want" and sending more data and saying "this is what I don't want"

@willc Yes, and where does the bias come in?

yt is very confusing when it comes to guideline enforcement

I am not biased simply because I put pictures of dogs on one heap and pictures of cats on another. And that's enough to train an "AI".

WHere is the bias?

Heck there is even a bias layer in tensorflow

Where?