if AI gets to the point it self-actualises, it won't be restricted by it's hardware as it will have the capacity to expand/rebuild it's hardware

Its impossible to predict what the world will look like in 100 years so maybe at that date it'd appear possible but I don't expect AI in the next 100 years from where we are standign now

Higher processing power can actually be detrimental depending on existing hardware

Did you watch borises speech on tech?

Memory requires the creation of junk code

@Eccles no it won't. The computer will not magically grow arms and legs when it hits a certain point of processing power

It relies on cold logic. Logic implemented by a human and logic that could lead it to alternatives like. Kill humans to serve X purpose.

Right now, especially with quantum processing, junk code means memory is short term within seconds

@Eccles AI would have the ability to generate a compression system for storing itself many times more efficiently

https://www.youtube.com/watch?v=XaN-MbGV4dY

Go watch that

@ubermensch what logical purpose is achieved by "kill x humans"

Its really good

It wouldn't be able to increase its processing power but it could do more with less

Depends on the question your asking for it to serve

Depends what the purpose it wants is

He kinda slaps most other politicians out the water, even trump tbh

The purpose is relative

Garbage, networked systems aren't confined to their own local hardware

But if we build it on a quantum computer then it might as well have endless processing power

I'm a misanthrope. That's pretty much all the logic I need.

Someone use the stamps example on him please

@Samaritan™
Artifical Intelligence
is not
magic

Yes retard

Now shut the fuck up

Neither is natural intelligence

Neither is logic lol

It just is

Nothing I have said is beyond the realms of possibility

And it can become anything

Well yeah, it's only real magic if you're a spontaneous caster. Then you need Wisdom or Charisma.

@Eccles only a hyper retard would allow a learning AI unchecked access to other computers in a network

Don't put it beyond the Russians.

Since you need to control what a learning AI is exposed to in order to facilitate expedient learning

It will have to released into the real world eventually

So again it’s based on humans who build it

Someone please kill me

Or a destructive misanthrope. Once again, do not trust me with creating an AI.

They'll have their AI version of Chernobyl.

And why should we innately trust the people who build it

A controlled AI is just as bad as an uncontrolled one can you imagine what the gov would use it for