Is our feelings kind of an intelligence? Can we make computers solve problems by making them feel happy, sad, mad, ...?
Actually, the base already exists i think starting from parser and compilers and tokens, we just need to express those emotions in terms of varieties of databases
Thanks. Would you please give me a more detailed description about it?
Thank you, Kees. I think some interesting results for solving problems may emerge if we design a model of feelings (based on any assumption) in computers. We may provoke the "feelings of computers" and see how they would respond.
Yeah, I think such an agent also needs to have a rich source of information, some degrees of freedom and a number of predefined rules in order to be able to perform the proper actions. Actually, I'm pondering on a mechanism of learning other than RL, I mean the agent shouldn't respond only to our rewards, but instead it should act in a way to satisfy its internal feelings too.
Has anyone taken the time to pull together a logic flow diagram of this or maybe a preliminary program? Or am I jumping to "start" too early? Until we put this in a program, it is only a debate.
When I see the term "set of rules" it translates to the programming term "conditions". When these "conditions are met, something happens." With that in mind, I propose that we start with some really preliminary stuff. This will all boil down to a bunch of - "IF (condition something exists)", "Then (sub-routine)" - or - "While (condition something exists)", "Do (sub-routine)" - condition statements. For example:
Condition: If: Hazard exists in this robot's area - It then falls to the programmer to define Hazard. The programmer can examine several things to define and later refine situations that can be appreciated by our robot. It could be defined as being given a command from some person. "You are in danger." or "There is a fire."
This could be defined as when a specific temperature is reached or when a smoke detector (like a standard house hold fire alarm) senses smoke. Or it could be scanning for flames and detect a temperature that is "out of the normal temperature range."
The programmed response will be to first decide the "fight or flight?" response.
If fight response,
then the robot will need to choose how to fight. If "Fire exists", is it electrical, chemical or general? Then "get appropriate fire extinguishing equipment." When the robot arrives at the fire, it will "while" fire exists "do" fire extinguish sub-routine.
Then there is the simple hazard, If trip hazard, Then the robot simply needs to move the obstacle.
If flight is required, then there is a program designed to help him exit the room, building or area. "While" in danger "do" escape sub-routine.
It might sound easy, but there are a lot of branches in this process.
This is very time consuming to do properly.
How extensive do you wish to get? Who wants to put "pen to paper" or start typing?
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence