@Deleted User another good example

Nice, thank you.

Keep them coming.

i think that's it

but proves my point lol

toaster <:shap:497652686603288576>

spaghetti_

Well, may point, too.

We wouldn't make such mistakes.

We don't have trouble seeing the banana just because of that stupid sticker.

The Turing Test is about making human mistakes.

You are not human if you're too perfect.

But you are still an AI if you make mistakes that a human wouldn't make.

right
i guess my point was, even though humans have some particular flaws in perception or judgment, they're still overall amazingly good

i guess

Good, yes. But also particular.

I think at this point we don't really want anything that is too human. We want a machine that can do things by itself.

We don't care about it being human and having flaws.

So @Den Röda Smurfen you give up ❓ what a loser <:shap:497652686603288576>

<:shap:497652686603288576>

it the uncanny valley problem. making a robot to human like creep out everyone

But we're not even in the same valley.

The sort of mistakes are on a different continent.

They are different categorically.

Correction my bad it wasn’t about cuckery 😂

get cucked by ai

<:think_madpepe:378717098630971395>

Cuck Bot 2000 Deluxe

this slide is pretty cool too actually; even though disguising the attack image reduces its effectiveness significantly, it's still pretty effective if large enough

Hey I’d shag an ai

🔴 THE TRAP IS HERE @Deleted User 🔴

ALERT

Burn it!!!!

I think ensemble approaches will dominate the field for a while to come.

be nice to our residential trap. or BuzzFeed will shut us down

💻 🐒 🖨 📜 🆗

Praise be to shrekko.

brb

I think we already surpassed human accuracy in 2015

Didn't ResNet get like 96% accuracy?

or 98?