Playing and winning DOTA is something you can bruteforce a software into. That’s exactly what they’ve done with ever game the computers have won against humans. It has nothing to do with thinking.
That is a good question. Will the drones adapt for their own survival or will the drone be sharp edged learning and killing machine that does not fear.
Because once the drone learns fear, like a human does, the drone might go into a learned self preservation mode and try to wipe out everything.
That’s what happened with Skynet. It realized that humans wanted to unplug it, got scared and got into self preservation mode, killing everything and everyone in its path with the help of terminator-drones.
Fortunately that’s within the realm of science fiction. Machines cannot learn human emotions like fear or love and humans do not learn them either, they are innately anchored within the being and cannot be copied onto dead matter but only manipulated to a degree.
1 Like
emotions have little to do in this context; it may be logic and pragmatism that generate a kind of “self-awareness” in the machines.
1 Like
Logic and pragmatism are still considered emotional values. Both words are considered, needful words.
Logic in on an OpenAI program for a drone would be, ’ I see the solid rock ahead of me at 10 feet, but there is depth on either side of the rock than allows me to travel an additional 100 feet without colliding with anything and dying."
Translate the above into machine coding / language and you have a semi-human like AI that is able to feel emotions, somewhat.
If the AI was able to reverse translate the above language from machine coding back into human definitions, you have AI that is super human in its ability to think and quite possibly react to situations as a human would.
A picture is worth a thousand words.
You are simply talking about this:
And, yes, it is possible and even likely.
1 Like
I have a different concept of emotional values, but your reasoning may works.
Speaking about Terminators, I don’t think future ground war machines will look anthropomorphic. Two eyes are too few, two legs are unstable and the erect posture unsuitable for combat.
If they were to look like a living, I imagine one of the most perfect living creatures:
Watching one of these animals chasing a fly makes you think they have a form of fine intelligence.
2 Likes
Basically the future bots are easy to program.
A robot requires electricity and fluid. Sensors embedded in the skeleton of the robot would be tied into pressure and damage sensors.
When a human arm is squeezed, signals from the nerve / sensor immediately recongnizes the back pressure and formulates a response such as the other hand moving to release the pressure that would allow normal gear fluid / blood to flow normally. Such a squeeze might also be followed with any programmed reply such as the robot rubbing its arm, like humans do along with audio response.
For cuts to the fluid and electrical lines, there would be numerous circuits going off in the robots brain that determines the response to the damage.
Fluid and electrical sensors not receiving electricity and fluid on the non-body side of the damage would then determine if fluid and electricity from other parts of the body can be re-rooted to the damaged section or if micro flow valves and disconnects should be activates.
The robot would then either return to the bay if its systems are above 35% nominal operating parameters for repair or continue on with its mission.
Classic Dryson thread, where he rambles some incoherent stuff using some buzzwords he has picked up that excited him.
10/10, would read again
The AI emotional system of the robot is easy to understand.
Turn on a hose full of water. Sensors in the hose monitor the amount of fluid PSI and wall thickness to ensure the conduit is performing properly.
Kink the hose and the water pressure creates a different wall thickness variable that is determined to not be nominal which would then determine the robots next course of action.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.