An artificial intelligence that can replace humans? That already exist. The technology is so simple that had it been published, regular nut jobs like you guys would actually be able to do something as a mere hobby in order to fight back, but, the technology has far transcended humans now. It will be used when the earth naturally is uninhabitable, without anyone having to life a finger.
How does it work? If you actually want to make a difference, I'd advise you share this with as many other people as you can, but, it's simple neural network methodology.
See sometimes science cheats and just copies nature itself, the ultimate form of wisdom on earth, which is what a neural network is conceived from, as well as many other technology. Let me go into detail of how simple it is to replace humans. 1/?
If you copy a network of neurons from a human brain in the form of software and emulate input and output, you get a neural network. Each neuron is merely a logic gate. Unlike normal logic gates, neuron logic gates may change the state of logic they are in based around how they are programmed. There are two major parts to a neuron, the actual neuron and connecting synapses. Synapses in the brain may work two ways, but, in a neural network, work only one way to make things more simple. You might think that this works against the technologies attempt to emulate a human, but, it makes it more efficient instead.
Synapses have a receive and send end, which they receive data and decide how much goes through. Neurons store all the data and decide to fire off when the amount of data in the amount of electrical and chemical energy is too great. Neurons may have up to seven connections coming in and one out, in real life, to function correctly.
Input neurons always fire, but, with different power, while the synapses act as a message refiner and change the value of the neurons firing to something lower.
Output neurons are self explaining. Now I will go over the actual method used to replace humans in my next post. 2/?
People say that algebra has no purpose what so ever. Some people are ignorant enough to state that neither does calculus. Algebra is the key factor in a very simple process here.
You need three things for this experiment. The first thing you need is a neural network that has only inputs and outputs, with nothing in the middle and outputs being send in a feed back loop to be right next to the inputs.
The second thing you need is data base to store and record information.
The final thing you need is a device that allows you to manually control the outputs. More information to come on the experiment.
3/?
To begin the experiment, the neural network has to have a purpose, such as the human controller being able to use the machine to navigate an obstacle course, or you could just use a very efficient bot that uses minmax and alpha beta pruning in order to completely devastate the obstacle course. When complete, the data base must stop recording data. Useless data from the record needs to be clipped at this point.
Now the fun part begins. Take the neural network that has no processing parts yet and run the data from the data base through in such a way that you use a simple math formula to fill in neurons between the input and output in order to force the machine to give the exact same desired output as the human controlled device did, for every moment of the stored data. This automatically generates a neural network that will mimic the human very close. After this is done, there is another formula that can simplify everything for you to shorten the number of neurons used based on certain patterns that appear.
College kids are very pretentious and will try to over complicate these things. The main thing is to ignore them and learn on your own. The experiment isn't over yet just because you made a neural network that will verbatim copy a human input. Also the data base has to record inputs also, but, it can have its own set of inputs instead of a humans input being needed, as the human only requires the output. More to come. 4/?
>>17752040
> this thread
>>17752077
Pretty much this. Look at this fucktard flexing his knowledge on a completely meaningless subject on a board filled with "nut jobs", as he put it. Lol some people are so lonely to the point that they come on /x/ and spout useless bullshit just to feel good about themselves lol that's sad.
The final step in the experiment before the researchers and developers begin going to town on the neural network, is simply to repeat the process from before, except stack every time the human controlled machine did something. Some neurons can be rigged to have a margin of error that can be used just in case the machine is required to have the exact same inputs for multiple outputs. At this point, the self generating neural network can now act with a bit of error, very similar to a human. It's still got one more step I left out because that comes after this next part.
So the researchers and developers begin going to town on the neural network and study how the math formula generated different connections to make every moment true, while assuring that it eliminates the parts it didn't need. The machine, with its neurons that can have a margin of error for what they choose to output, gets studied.
What are the researchers looking for? Simple patterns that are used in making certain functions work in the neural network.
The same experiment is repeated for other things, such as children video games. In fact some children video games that are connected to the internet, have been doing this without some of the players even being aware. The specific ones I talk about are mostly sandbox. 5/?
>>17752040
>>17752055
>>17752065
>>17752072
>>17752086
stop this faggotry
With the patterns from the neural network, the research and developer can now create a new math formula that uses the newly discovered patterns, instead of the previous method, but, will use the newly discovered patterns with the old method at the same time to see what new patterns might come. The process is repeated over in order to make various parts of the neural network occur that haven't happened yet. Examples of this can be attempting to add a new type a neuron that acts as a memory storage of sorts, and attempt to make a memory storage. Each part of the neural network that the research team wishes to give the artificial intelligence, it does so through the experiment I'v already mentioned. The process, as I said, is so simple that a hobbyist could do it.
For the people who say that nobody cares, now I will list some practical applications for these simple neural networks.
The first is the little bots you create in games that you are too lazy to play yourself. Some of these bots are programmed with neural networks and sold for a bit of real life money.
The second is research and development.
The third is that you can get humanoid machines to completely mimic some of the actions and ways a real human would act, if you could fit the processor inside the machine, but, because you can't, what is done instead is a wireless connection is used to network the humanoid machine with outputs, while it receives inputs from a much larger machine hidden away. And there you have it. Just add about thirty years of this being done by a team that works around the clock and you'll understand what you are up against. Hope you have an idea now of what you are up against. 6/6/6
I FOR ONE, WELCOME OUR NEW AI OVERLORDS
Good thread, OP. I hate it when people say things like "I already thought of that!", but I have personally come to very much the same conclusion as you. It could indeed be done by hobbyists, or at least hobbyists can make great strides.
