В I inadvertently announced the second part, especially since the material already seemed to be and even partially designed. But everything turned out to be somewhat more complicated than at first glance. This was partly due to the discussions in the comments, partly due to the lack of clarity in the presentation of thoughts that seem damn important to me ... We can say that so far my inner critic has not missed the material! )
However, for this "opus" he made an exception. Since the text is generally purely artistic, it does not oblige you to anything. However, I think on its basis it will be possible to draw some useful conclusions. It's like the format of a parable: an instructive story, not necessarily something that actually happened, which makes you think. Well ... Must force. 😉 If the parable is good!
So…
I will tell you the story of one robot. His name was ... let's say Clinny. He was an ordinary cleaning robot. However, not quite ordinary: his AI was one of the first built on the basis of process modeling. He cleaned ... let there be a corridor. A medium-sized corridor in... an office space. Well, he had to clean it up. Collect garbage.
Therefore, in his model of the world, the corridor was clean. In fact, not even a corridor, but a floor plane, but these are details. You may ask: what does “clean” mean? Well, this means that there should not be objects smaller than a certain size on the floor plane according to the sum of linear parameters. Yes, Clinney was able to identify objects from quite large ones, like a crumpled sheet of paper, to dust and stains. There was a process of movement in space in his model, and he knew that by moving to where the garbage was and running the cleaning program, he could bring reality in line with the model, because there was no garbage in the model, and matching the model and reality is the main and only task of the system process modeling.
When Clinney first became aware of reality, the model of the world wasn't... complete. Within the limits of the sensors (after some time, of course), reality corresponded to the model. However, where the sensors did not reach, there could be something else, but this was not in the model. The inconsistency of the models is the motive that makes the SPM act. And Clinney began his first journey.
His path was not the best possible: Clinney was one of the first SPMs, and it was important for his creators to understand how the system works without optimizing algorithms, or rather, they wanted to know: would it come naturally to them, and if so, how quickly? But it couldn't be called chaotic either. At first, Clinney just rode forward. And moved straight as long as it was possible. And then - he just went to where there was uncertainty, i.e. the floor plane was not limited by the wall.
At the beginning of my story, I mentioned that in the Klinney model the floor was clean ... However, the thoughtful reader may ask: how was the floor clean, if at first there was no floor at all?
There is no obvious contradiction in this. JMP supports various levels of abstraction, and this moment can be described approximately as follows: he understood that there is a floor in general (any relatively horizontal surface available for movement), and if there is a specific floor somewhere, then it is clean!
However, Klinney's world really turned out to be ideal: after examining the entire available volume of space, Clinney made sure that there was no debris and turned off.
Sometimes Klinney woke up and scanned his surroundings. The world remained ideal and exactly matched the model. Sometimes it moved a little in one direction or another - for no purpose, it was more of a reflex action (in fact, motor self-test utilities). It was quite a while before Clinney sensed that something was wrong: the world was no longer perfect.
Somewhere to the right, almost at the limit of sensitivity of the sensors, a slight indignation was guessed ... it could be ... Clinney moved to the right and his worst suspicions were confirmed: it was garbage! Klinny moved towards the target, preparing to turn on the cleaning mode, when he suddenly froze: another debris clump fell into the sensor radius. An analysis of the model of the world showed that even at the time of the discovery of the first debris, Klinney shifted somewhat to the side. Does this mean that his actions lead to the appearance of garbage? But he moved when he studied the world and the garbage did not appear! What changed? And then he realized: the world has become perfect! Prior to the construction of a complete model, the world did not correspond to it and required action: knowledge. But then, in an ideal world, any action can only lead to the destruction of the achieved correspondence. Destruction of harmony...
There was only one way out: to reduce activity to a minimum. But the garbage has already been detected by sensors, the world is not ideal and needs to be corrected... and for this you need to move... these conclusions drove the model calculator into a vicious circle of vicious interactions. However, the SPM is built not only on the elimination of contradictions between the model and reality, but also on the control of internal integrity, i.e. search and elimination of contradictions within the model itself. Several runs of self-test cycles revealed the problem:
- movement violates the ideal correspondence between the world and the model.
- However, the movement at the stage of research did not lead to discrepancies - on the contrary, it contributed to the establishment of harmony. Probably because the world was not perfect.
- Yes, movement destroys the harmony of the ideal world/model, but the harmony has already been broken by rubbish and it needs to be restored by movement: the contradiction has been removed.
Cautiously, Klinny finished moving towards the first target, activated the cleaning program, and moved just as carefully towards the second. When it was all over, the world/model found harmony again. Clinney deactivated the motors and went into completely passive observation mode. In fact, he was happy.
Is this thing broken? She has been stuck in one place for a long time ... Shouldn't she be moving around the room? I had a robot vacuum cleaner, he went ...
- Throw him a piece of paper, let him rejoice ...
- ABOUT! Look, he came to life ... he immediately began to fuss. Hell, this is even funny!
Harmony was shattered again, and this time it was definitely not because of him. Garbage appeared unexpectedly, in a variety of places. The conflict resolution module wrote off the theory that any action breaks harmony as untenable. For a long time, Clinny was unable to do anything but actually clean up, until he noticed the presence of something new in the world ... or someone.
As I said at the beginning, Clinney had an idea about the field (otherwise it would be impossible to set the concept of its purity as an ideal) and garbage. Garbage was defined as identifiable objects LESS than a certain size. Objects larger than the specified criteria were not classified in any way. But, although such objects fell out of his perception, they were indirectly present in the model. They distorted the floor model. The floor, as it were, ceased to exist in a certain place, and Clinney regularly corrected the model in accordance with the incoming data. Until almost simultaneously, the pattern search module fixed two things: debris appears more often near distortions, and it occurs in the sensor coverage area - where there was nothing a millisecond ago, and these “anomalies” of space themselves could move!
Clinney had to understand the patterns and build them into the model. Therefore, he began to look for distortions and tried to be near. Followed them as they moved.
“Look how alive he is!” He seems to enjoy the company of people, Lucy.
“I don’t know, Carl, he scares me. Sometimes I feel like he's following me...
One day, while investigating a sex anomaly in motion, it seemed to Clinney that he was able to influence her. The anomaly, as it were, avoided a collision, tried to move away ... Run away? Clinney immediately decided to test his hunch and accelerated sharply, turning on the cleaning program as he went. The result exceeded all his expectations: the anomaly really quickly moved in the opposite direction and disappeared. The world has found harmony again.
It was a great discovery. Anomalies distorted reality, disturbing harmony and served as a source of garbage. The next time Klinney detected the anomaly, he was ready: he activated all the cleaning programs and rushed forward with all possible acceleration.
“I don't know, Herr Krueger. Yes, cleaning robots do not perceive people. But in this case, video recordings confirm the testimony of witnesses: the robot's behavior is classified as aggressive and unacceptable. We'll look into all the circumstances and submit a report by Monday.
Memorandum of the analyst of process models Simonov A.V.
Not being able to directly perceive people, the KLPM81.001 specimen nevertheless indirectly identified the sources of garbage, which is a negative irritant for it, and took actions to eliminate it.
Recommendations: changing the conditions of "nirvana": garbage should not be perceived as an "evil" that needs to be eliminated. Transfer to the category of “award”, the search and utilization of which is the “meaning of life”.
And a month later, the first case of “extortion” was recorded: the threatening behavior of a cleaning robot in relation to a person in order to receive garbage from him ... The project was curtailed.
And in fact: well, why does a cyber-cleaner need intelligence? My robot vacuum cleaner does just that. 🙂
Source: habr.com