Sup Veeky Forums

sup Veeky Forums

ITT: we share our ideas, big and small, theoretical and concrete, how to achieve artificial superintelligence.

My suggestion would be to build something like opencog, but on blockchain so that people will be motivated to help, since no single rich person can just instantiate the awesome open source agi on his server farm and control the god all by himself.

Besides that, I guess improving machine learning technology and applying it to some other crucial problems wouldn't be wrong. Although I have no idea what those "crucial problems" would be - anyone?


inb4 hur lel singularity is not 100% proven lets not even try in the 0.5+ century we have left fallacy

before you can simulate something, you first need to know what it is and how it works.

op, please define 'intelligence' and 'consciousness', then explain in detail how they work.

I don't want to simulate anything. Consciousness is also not the issue.

What I want (and you should, too) is essentially a "super science machine" that can improve on a level as meta as possible until we get magic life extension.

Maybe someone in the life sciences can help and tell us what currently the most tedious processes are, so we can think about how to automate those (or have a machine automate them)

first we need to migrate all binar logic to ternary

them we can start building a real computed brain

wont go into detail here tho as most of you understand jackshit anyway

Is converting binary logic to ternary even possible?

I GET IT

HE'S "BOGGED DOWN" AT WORK.

10/10 VISUAL PUN :D:D:D xD LMAOROFL

IMMA PRINT DAT SHIT OUT AND HANG IT UP AT MY RECEPTIONIST'S DESK AS A GOOF

GREAT THREAD OP

go back to your friends at

This seems an appropriate thread to ask a question

I am seriously considering researching this but perhaps someone already knows


When you break an egg into a pan, and you get a bit of shell in there, it's almost impossible to get it out - you try your finger, a knife, a spoon or anything you can't get it.

But - you use part of the shell to get it out, it comes out no problem.

How the hell?

Is it like a spider not getting stuck by its own web?

yes just google

Meta cognition is what will take AI from computers "programmed to think" to "programs that think deeper than human beings."

>create a sandbox environment that AI can interact with
>create primary-AI with goal of producing another sub-AI that will populate the sandbox with NUMEROUS unique meta-AIs
>only meta-AIs are allowed to interact with the sandbox
>sub-AI is only allowed to interact/change meta-AI
>primary-AI is only allowed to interact/change the sub-AI
>after enough iterations remove primary-AI and replace it with the sub-AI, it is the new primary
>change perimeters of the sandbox
>repeat from start

!!!!!

What is happening in the OP's picture?

What kind of sandbox do you propose to sprout intelligence? You'd need high amount of complexity, similar to universe, and we're still far away from being able to simulate what took 14 billion years in some sensible timeframe.

Also, humans have various levels of metaconcsiousness.

babadook

HOLY

I would start out very simple at first with very simple clear cut goals for the meta-AI so success can be gauged easily. Then gradually introduce more levels of complexity. When the complexity becomes too much to simulate in a computer give it a robot body and instruct it to interact with the real world as if it's the sandbox. When this happens, instead of meta-AI interacting with the new "sandbox" directly they would each control parts of the robot, or rather the robot becomes the new "sandbox" for them to interact with.

1. Read more machine learning literature
2. Write more code

>implying artificial sensory systems are even coming close to ours

>inb4 hur lel god is not 100% proven lets not even try in the .5+ century we have left fallacy

Eggshell = scoop

lolol wtf OP

It plays on the fallacy most teenagers who stumble upon science make - i.e. measure everything against what is proven, not what might be a good idea to try even if one might fail.

try what, singularity ?

Hmm, think the best way is to get a man and woman together, and make them do it

>artificial superintelligence
First prove to me that such a thing is possibly.

that picture is absolutely amazing!

you comsider that artificial?

it's all yours my friend

You need to know how the mind works in the first place in order to do this. If you try to take a simulated evolutionary approach, (which is hardly better than cycling through all possible Turing machines until you land on the magic intelligent machine) you wouldn't be able to understand or control anything you get as a result.

Here's how it can be done:
1. Figure out how the mind works.
2. Implement it.

>sight
>taste
>touch
>pressure
>itch
>thermoception
>sound
>smell
>proprioception
>tension sensors
>nociception
>equilibrioception
>stretch receptors
>chemoreceptors
>thirst
>hunger
>magnetoception
>time

I'd say more than have of those machines have us beat already. One quarter of that they have more accurate detailed senses but lack the brain power to interpret the sensory data. The other quarter they may never beat us at.

>>artificial superintelligence
>First prove to me that such a thing is possibly.

First prove to me you are not a super intelligent AI indistinguishable from humans. You can't, checkmate atheists.

and the other two quarters?
your post is a mess hombre

"prove to me that jumping from the burning plane will not kill me"

Default is you die mate. Might as well try to achieve godhood.

Also, what kind of engineer thinks like this? You try, and if and when you built it, it's possibility proven.

>prove to me that jumping from the burning plane will not kill me

Proof by counter example :
Flight Sergeant Nicholas Stephen Alkemade

>grow 10 brains in a vat
>mash them together into 1 brain

BOOM, superrhuman

You laugh, but this would actually be the most efficient way to do it.

Our brains evolved over millions of years. What hope do we have of reverse engineering them in a tiny fraction of the time?

Let's not reinvent the wheel. Intelligence already exists in us. And everything we know about it so far suggests that manipulating it really could be this counterintuitively simple.

The only thing standing in our way is our squeamishness.