Beating OpenAI

Is it possible to create a general-purpose Seed-AI before Google?

Google employs all the best AI researchers and constantly searching for new talents. They will surely do breakthrough before you, retard, even think about it. So just curl-up in the corner and die.

>>2
have they employed mentifex yet?

>>2
Not that fast. I'm confident they're missing key crucial information on biological mechanisms of thought and will not reach above neural mimicry of specialized functions.
That information is however easy to deduce if you know the true function of the brain and its place in the mind. Its outside of scientific paradigms.

>>4
Google has fucking self driving cars already.

>>4

they're missing key crucial information on biological mechanisms of thought and will not reach above neural mimicry of specialized functions.

They can fill it by attaching a camera to infant to sample all its interactions with outside world, as it is being developed. Then will be just a matter of fitting a curve (i.e. neural network).

>>6
Basically they need to see, hear and feel all that infant sees, hears and feels. As soon as they sample that, they will have good enough dataset to bootstrap AI. And it will be the end.

the private sector never makes massive research breakthroughs. the R&D cost for strong AI would be in the trillions. no company even Alphabet could afford it. the most any private sector company will get to is shitty self driving cars and automated burger flipping.

neural networks are very limited in terms of real AI. something totally new will have to be created, which means academia + DARPA research.

I wouldn't be surprised if the Pentagon/DARPA is already working hard toward this. it's literally the new Manhattan Project. whoever gets to strong AI first will own all the resources in the solar system.

>>7
Sampling feels doesn't sound very ethical, and might not be the same as the AI having the feel apperatus. It seems more like teaching a parrot to squawk happy or sad

>>9
Thoughts and emotion are intertwined. Motivation and goals are emotional.

https://www.sciencealert.com/scientists-simulate-artificial-life-in-quantum-algorithm-for-first-time
Using the IBM QX4 quantum computer, the researchers coded units of quantum life made up of two qubits (those basic building blocks of quantum physics): one to represent the genotype (the genetic code passed between generations) and one to represent the phenotype (the outward manifestation of that code or the "body").

These units were then programmed to reproduce, mutate, evolve and die, in part using entanglement – just as any real living being would.

Random changes were introduced via rotations of the quantum state to simulate mutation, for example.

>>8
>neural networks are very limited in terms of real AI. something totally new will have to be created, which means academia + DARPA research.

Currently ANNs recognize almost everything animals can recognize, the remaining problem is integrating their output. But I think the main problem would be the amount of connections real neuron has (up to 10000) vs artificial neuron (bare minimum). The more connections you have the more precise is the answer. So with 10000 connections you will get 99% speech recognition, while with 100 connections you will get say 50%. The difference is huge, as between the system will perform well vs it will bump into a corner or go nuts other way, missing some edge cases.

Can ANNs recognize jewish plots to import violent homophobes into white countries and only into white countries?