How would AI change our world?

Game Theory is based on human thought though. I mean we are talking about a machine that could learn everything there is to learn in the world by simply downloading it into its hard drive. One with an IQ far superior to a human beings. So I'm not sure if ethics is something that an AI would have a use for.

 

Well since this AI is part of the world to be perfectly optimal in conjunction with other sentient beings it must interact with its surrondings. These actions may be deemed ethical and game theory would still play a central role in finding the perfect action. Game theory isn't based purely off human thought now anyway.

 

I just had a vision. PREACHERBOT SAYS HALLELUJA!

When did we determine the 'true AI' would be smarter? I think it's more logical to look at earlier AI's which would be about as smart as dogs or so.


I think a lot of the applications depend on the first few AI's. Most important:
- Do they show emotions?
- Can they talk?
- Are they about as smart as humans?
- Are they as strong and mobile as humans?
If the information to all is yes, their creation will either be banned, or they will be developed as adoptable 'babies'. If the answer to question 1 and 4 is no, I can see a lot of applications in government administration. No more corruptable low-tier officials. if the answers to 2 and 4 are yes and the answer to 3 is 'not as smart, but close' they would probably be used in the military as cannon fodder. If 2 and 3 are yes, say hello to automatic chaffeurs. if they have 1 and 4, and have programming to pretend the answer to 1 is yes, say hello to hookerbot.

The story is of course completely different when we start talking about superhuman AI, but these would probably be develop out of earlier weaker AI's and will, at least for a long time, be concidered better alternatives to these humanoid AI's.

 

game theory isn't based on human thought. it's actually based on math. it's built into the human brain, but it just happens everywhere there's competition. you can even study game theory in evolution of competing species.

 

But what would the AI be competing with?

 

Consider: If an AI formulates a goal, then it would reason that maintaining it's own existence is necessary in order to achieve that goal. Survival is prerequisite.

In that need to survive, the AI must secure a power supply in order to keep operational, and also be able to defend itself against critical damage. As a consequence of this, and depending on its competence, it may end up achieving world domination as a means to ensure that no army exists that can attack it, and that no human being is able to deny it electricity.

It would be competing with all other cognitive agents, as do you.

But competing is perhaps not the best word. It would be more accurate to say the AI is in "games" with all other cognitive agents. Games are situations where each party has choice, and one can either hurt or help the other. Most life strategies revolve around repositioning yourself so that other cognitive agents are less able to hurt you, and better able to help you.

In the above example, the AI would be in a game with the people who work at the power station. The engineers can choose whether or not to switch off the power grid, or leave it switched on, which would either hurt or help the AI computer.

It would be in a game with the local police, who could choose to shoot it on sight or protect it like they would any other citizen. Their choice hurts or helps the AI computer.

Understand that you are in those two same games. You pay money to your energy provider, and keep them rolling in the clover, while they give you an electricity supply. You are currently helping each other. But either party could at any time withdraw from that contract, and hurt the other. Depending on how much your survival depends on electricity, and depending on how much you value your life, you might find that possibility of having your power switched off at any time an unacceptable risk, or not.