A bridge between physical and computer world

[fredreload]

We all know about artificial neural network and genetic algorithm, but what good is it if it can only be tested on the computer world? What if you build a robot, and have the robot runs these algorithms to protect you at all cost in the real world? What if you build a microwave turret and have the microwave turret runs these algorithms to protect you? What if you ask the robot to make you immortal, to make you a god. The possibility seems infinite.

Edited December 6, 2017 by fredreload

[Endy0816]

They have before.

https://www.damninteresting.com/on-the-origin-of-circuits/

You'd have to give them alot of time and be ready for a ton of failures. In some cases the definition of 'success' can also differ between man and machine.

[fredreload]

57 minutes ago, Endy0816 said:

They have before.

https://www.damninteresting.com/on-the-origin-of-circuits/

You'd have to give them alot of time and be ready for a ton of failures. In some cases the definition of 'success' can also differ between man and machine.

This is more what I am looking for and yes I am ready to train the algorithm based on real world physics. The idea about artificial intelligence is that it does not possess a consciousness and that is a good thing, because no consciousness should be trained with such a task. If you let the evolution algorithm runs its course, two things would happen. It would develop a consciousness, or it would dominate our world in a bad way (turned evil).

So the ideal AI, would not develop a consciousness and is there to serve human

[Strange]

As that video shows, genetic algorithms can't be utilised for real-time responses; they take hundreds or thousands of generations to come up with an effective algorithm.

However, both neutral networks and genetic algorithms are used in robotics, so not quite sure what the question is ...

[fredreload]

1 hour ago, Endy0816 said:

They have before.

https://www.damninteresting.com/on-the-origin-of-circuits/

You'd have to give them alot of time and be ready for a ton of failures. In some cases the definition of 'success' can also differ between man and machine.

You got me, my main focus is still on modifying gene expression with electromagnetic radiation. Imagine this directed microwave that would alter your cell's transcription factors to phosphorylate and induce a partial reprogramming to renew itself.

1. You will need to find the correct wavelength of electromagnetic radiation that would induce phosphorylation.

2. You will need to get molecular detail to locate the transcription factors needed for phosphorylation

3. The transcription factors will make the cell's gene expression to change to a previous year (age 30->age 29). So you got up to 2600 transcription factors in the human genome according to Wikipedia, and you play around with all combinations until the gene expression successfully shifted to the previous state. If all that the four Yamanaka factors are required for a successfully shift to a previous state, awesome. But based on SALK's research, the mouse still dies after testing it for serveral times. I am not sure if they mean that the mouse with a genetic defect on premature aging is no longer alive or that the mouse without the defect is still alive as of now. Either way it is mentioned that a successful way to trigger the transcription factors inside the cell is required.

So here it is, the electromagnetic radiation should be possible to activate the transcription factors, or the alternative would be cell signaling molecules. It is to be debated

11 minutes ago, Strange said:

As that video shows, genetic algorithms can't be utilised for real-time responses; they take hundreds or thousands of generations to come up with an effective algorithm.

However, both neutral networks and genetic algorithms are used in robotics, so not quite sure what the question is ...

Right hmm, I thought about it too. You are training the machine based on moving instances. You see the part where they throw boxes at the machine with weight until the machine falls? Imagine the boxes to be moving cars and that each time the machine gets hit by a car and knocked down it resets itself, eventually you'll get a machine that cannot be knocked down by cars or that it can successfully evade the cars. Or you can train the robot to protect the person from cars.

Now if you set the robot to update its own circuitry, so that it makes itself smarter, faster, and better. This is the realm I dare not venture to. You cannot set the computer simulation to evolve like humans do, because we have different parameters, gravity, and space time. However a self adapting machine in the real world knows no bounds, think of the horror movie screamer.

Edited December 6, 2017 by fredreload

[Strange]

17 minutes ago, fredreload said:

This is the realm I dare not venture to.

Never mind. Other people will do it for you! But you could miss out on all the fun! I get the impression you are quite young - and you are certainly enthusiastic. There are going to be some exciting developments in engineering if you get involved (although might have to learn to focus a bit more... instead of jumping all over the place with your ideas).

 

[fredreload]

34 minutes ago, Strange said:

Never mind. Other people will do it for you! But you could miss out on all the fun! I get the impression you are quite young - and you are certainly enthusiastic. There are going to be some exciting developments in engineering if you get involved (although might have to learn to focus a bit more... instead of jumping all over the place with your ideas).

 

Ya that is true, it might needs some kind of validation, you might need to program it with the seven deadly sins.

You are right, my thoughts is going everywhere. The electromagnetic radiation put me in stasis

Edited December 6, 2017 by fredreload