/sci/ - Science & Math

also apologies, since I'm using geoplexami which no one else knows exists and is a completely unique creation.

>>9938481
Dude what are you on?
>> abstractions
if we don't have some way to learn all these abstractions we have to code them all in.

>>9938594
DOMAIN RESTRICTION AGI
DOMAIN BEING ARTIFICIAL INTELLIGENCE
RECURSIVITY
ABSTRACTION LAYERING NESTING OUTERPRODUCT AND CONTRACTION ON THOUGHT VECTORS

WAKE UP

parallelism runs into brick walls
need divergent uniqueness for optimal problem solving
singularity is a foolish notion when it will be exponential numbers of DR-AGI

Need context -> Restrict Domain
Optimal domain choice? Video Games? LOL
Artificial intelligence is domain
Two types, static pattern matching, majority of human brain, with some dynamic capability -> consciousness.
It has to plug into a lot of the static systems to work though or we lose context. Can just replace the static with very simple domain

Processing power not needed.

so many experiments possible
>behavior
>drives
Seeing it alternate strategies from developing larger structures versus optimizing, fuzzy vs detailed etc

I'm going far ahead!

>>9938651
>>9938654
>>9938666
>>9938661
If you don't define what the hell you are talking about I'm going to urge you to see a psychiatrist. So why don't you define what you mean by a domain and thought vectors?

>>9938688
lol @ brainlet

Just brainstorming my project out. Domain is like, "Image recognition." - "Go" - "Chess" - "Drawing in paint". Thought Vectors is braindead to understand too. In the static case a bunch of branches mix together and at the local point of max complexity you get the Outerproduct of all the vectors aka the thought vector. It's localized, and has many abstraction layers.

I've only been working on AI for a few hours though so you can check back in a week when I'm finished with the problem.

>>9938481
>triangle/pyramid shapes
posts Greek columns

>>9938693
Schizo/brainlet/retard/larper

>>9938697
>not able to intuitively progress on AI
kek, bet you don't even realize how dynamic structures reduce the structure size by being able to combine instead of just exploding up 2^n for increasing thought vectors.

Shit's efficient.

>>9938699
Schizo/brainlet/retard/larper

>>9938693
ok I see what you mean by domains now, and it confirms my suspicions that you are in fact a schizo/brainlet/retard. AGI is not domain restricted! General implies not being limited to one domain.

>>9938779
Damn you don't get it, brainlet

a general intelligence can operate in a restricted domain and be AGI. Testing, experimentation, etc

>>9939139
Schizo/brainlet/retard/larper

>>9938481
AGI is only 4 triangles. Are you brainlet?

>>9939221
How many triangles do you think it is?

>>9939221
Make the simplest bitvector constructor and you'll understand brainlet.

2^n = 2 double every step, looks like a triangle

>>9939234
Not sure yet. I dance across the problem. The mostly static part is easy, dynamic part is a little more complicated I think.

It's a very interesting problem though. I'm surprised how fun and pretty it is. The interesting thing is how the different structures work together. I wonder if it can be sped up with more complex basic units than a neuron that store more complex states or not.

It must be the trick to consciousness that it can mix all that stuff together. Imagination is really cool though reminds me of how games are programmed in some respects.

That's probably the most surprising thing recently is that AI and Games have a lot in common and are similar objects.

>>9939234

4 like I said before. First of all our AGI has to have the ability to get things done and make changes to itself and the world (so power). Secondly it has to have some ethics, a set of laws to follow, it has to make decisions based on the law (wisdom). Third, it has to be capable of mastering anything we throw at it and the desire to survive in the world (courage). Lastly us humans have to realize an AGI is just as real as you and me. It's nothing physical the AGI has. Its other entities giving it a soul basically.

When we combine all these "triangles" our AGI is completed and can rule the world.

>>9939269
>>9939359
I wasn't asking you, schizo

Seek a mental health professional.

>>9939378
>being so close minded you can't read between the lines
Imagine if I posted shit that was understandable Think why?.

>>9939382
Shut up schizo

>>9939382
You are having a manic episode. Get help.

>>9938481
You can create an intelligence but without the fundamental inputs of this world it will be useless. If you can imbue a robot with fear, you have succesfully made it evolve on its own.

>>9939415
General Directive: Survive.

>>9939410
lol, you've never danced across algorithms man. It's not a manic episode or schizo. You guys are the ones with really fucked up brains and thought processes.

You can understand it totally simply. Imagine a simple branching tree with binary choices each level. To create a thought vector out of it, each point has an incoming source based on level. The connected line with all signals on = the thought vector.

Decode the above into your shitty language of math, which I refuse to use as it's arbitrary and there are better choices for AI/algos.

>>9939455
>Math...which I refuse to use.

HERESY OF THE HIGHEST ORDER.

>>9939461
I like math but it can really mess you up. People rely on it too much. It's an amazing tool though but why go with the herd? Much more interesting to develop something unique based on entirely different foundation: symmetry, duals, geometry etc.

It's still related though. Not hating on math but if you approach everything like the rut math puts you in it's not a good thing.

>>9939467
>Geometry is not a branch of mathematics.


Get out. GET OUT.

REEEEEEEEEEEEEEEEEEEE

>>9939467
>symmetry, duals, geometry
>entirely different from math
wat

>It's an amazing tool though but why go with the herd?
if following universal rules is "going with the herd", why don't you swim against the stream in other areas of life and just jump down a building to see what will happen?

>>9939473
kek, it is, doesn't mean you can't have a different thing that uses geometry, dual is too, symmetry is too, etc

You can also have writing with "three people go across the street" and say it's math because it has 3 in it right?

>>9939478
>>9939478
>You can call it math.

A NUMBER STATEMENT CONTAINING JUST ONE NUMBER ISN'T MATH.

SYMMETRY IS A PROPERTY OF GEOMOETRY.

DUALS ARE EXTRAPOLATIONS FROM EXISTING GEOMETRIC ENTITIES.


AAAAAAAAAAAAAAAAAAAAAAAAAAH

>>9939494
scizo "math" posting detected

>tfw solving AGI in the though vector space

>>9939514
REEEEEEEEEEEEEEEEEEEEEEEEEEEEEE

>>9939514
How did you rate my thread so far?

>>9939538
10. 10/100.

Fuck and here I was trying to cram the Monster group into a Hodge theatre.