Message from @Kealor
Discord ID: 641236773170708482
not the source
thats is a massive assertion
and that's the disagreement
my fucking god this _name in winders kills me
most see the brain as being like a computer, each portion having specific hardware
The machine just does not do searches with normal spaces and blaNAMEbla formats
rather than the brain being like a memory register
winders is dumb this way
where the hardware is the same, but the parts of memory which handle certain functions are because of a running kernel
that is an empirically defined system though
there is no way to tell these apart empirically
bridging that with the constructed system of computational processing is likely what will lead to the development of ai, rather than sohpisticated illusions
agreed
singularity
just because it hasnt been bridged yet doesnt mean a bridge is impossible as you seem to be claiming
why can't NN be the source and a result of intelligence?
That's literally a description of an evolutionary lööp
which is one of the key reasons i think they are greatly exaggerating
because a NN doesn't self-seed nor can it handle selection of optimzation criteria
or atleast by virtue of them being different, to assert that they fundamentally cannot be merged requires more than just the assertion
how does it not self-seed?
There are NNs that create other NNs and evaluate them
Neural networking is a phenomena relatively detatched from its namesake
Constantly improving NNs that do a better version of self is literally what Amazon spends billions on
because these are two seperate processes and the input schema is always pre-defined
the brain doesn't require you to know what you are looking for when you first encounter an instance
btw Amazon throws out algorythms that are starting to make too many decisions
lol
you create a representation in it's mind and until you associate that with a definition, you only have it's identity
Imagine a poor algyrythm being "I am self-sentien"
Amazon: delet
machines are capeable of the same behaviour
behaviour?
they just generally lack the abstraction to connect data points that usually follows and is used to improve the algorithms used
present NNs can do that, but its limited to its own scope
again, the result of abstraction is not the mechanism which allows abstraction
it simply isn't capable of knowledge representation in a native context
why not? interesting philosophy is predicated on the understanding that circular loops exist
because O^n
Thou shalt not make a machine in the likeness of a man's mind
for each additional layer, the propagation time of the network will increase
the only away around this is a convention based arrangment of storage elements which can self-organize and use the same pathways to access multiple iterations