The black box problem
When we see the black box, we see only the core of the thing. And the thing that is in the box remains a mystery before we open it.
The black box problem is the thing that we know that there are bites in the box. We know that there is the possibility that there is a DVD. Or any other merchandise in the box. But we don't know what kind of merchandise or item is in that black box. The thing is that. We know what kind of merchandise we order or what should put in the box. There is the possibility that the deliverer has made mistake and sent the wrong box to the wrong place.
We know what should be in the box. But only the physical observation confirms the information. The fact is that we are just assuming the thing that should be in the box. We cannot be sure and confirm the information before we see what is in the box.
We can predict that there is a certain thing. But we don't know what exactly is in the box. The "black box" testing means that only the functions of the computer program are tested. So there can be many mistakes and errors in the code.
When we are thinking of the working day it's like a black box. Many things can happen during our day. But we don't know exactly what is happening when we are beginning our day.
The useless information is the problem.
When we are thinking about the information that we need when we are ordering the things from the net we need to know the color or we need to know the material. But there is one thing that is ever imagined. That thing is what kind of surface that material has.
What does the material or merchandise feel when we are taking it in hand? And of course, nobody mentioned how large the box is, where the sender packs the merchandise. The information that we get is what the merchandise looks like and what it should feel like?
But that kind of data would not help us when we go to a shop and buy a box of coffee. That is an example of how specific the data we use is. The same data matrix that is used for ordering merchandise from the internet. Is useless when we are visiting the shop.
And both of the data matrixes are useless when we are walking on the streets. We don't know anything about the price of the coffee packet. And the color of the skirt while we are crossing the zebra crossing. That is an example of how the information that we use is specific.
The problem with information is that most information that we get during our lifetime is useless in everyday life. We never imagine how large databases we need. When we are making robots we must program every single action to that robot.
When we are making a program for the robot that can walk on the streets and open the door the programmer must determine every action that is necessary for those actions. When the robot opens the door it must know how to turn the knob? And then, the robot should know should it push or pull the door. But here we are always, forgetting one thing. That is the power that the robot should use when it opens the door. If the robot uses too much power it will break the locks.
We want to make robots for making our life easier. And those machines are planned to replace humans in simple, dangerous, and boring missions. But when we are thinking about the possibility to make a robot that helps us in everyday life. Those kinds of systems are quite complicated to make.
Like hat cleans our house and goes shopping for us. We are facing one thing. Those kinds of systems are quite hard to make. But when we are thinking about things like robots that are serving us in the future making those systems act safely is problematic. The paradox is this: we can make the artificial intelligence that can find exoplanets near other stars. And we can make a drone that can flight independently in the military area and avoid enemy fire. But we have problems making the robot that goes shop for us.
The day is like a black box.
When we are making robots that are serving us in everyday actions. We are facing the so-called "black box" problem. That means there is a lot of things that can happen when a robot visits shopping. There can be a car that is parked to the route of the robot. And a robot must evade that object. There is the possibility that there is a zebra crossing where is no light. And in that case, the robot must estimate the distance to cars independently.
Those things are the most usual things that are happening in everyday life. But then the robot can see the seizures and it should know how to act in that case. There is a lot of things that are happening and what could be happening in everyday life. And the greatest challenge for robots and AI makers is to create robots that can operate in normal life. Is harder to make a robot that can visit to shop for us, and clean our house than some kind of automatic killing machine.
If that killing machine operates in a conflict zone it must know what the enemy vehicle or an armed person looks like and if the IFF signal of those targets will be red. That robot opens the fire. But when the robot must clean the house. It must follow many things like is there enough garbage bags and make reports of that thing.
https://thoughtsaboutsuperpositions.blogspot.com/
