[Current Affairs Analysis] The results of running a city simulation with famous AIs are fascinating! Each one collapses for reasons that seem to fit their personality, and when mixed together, they behave in unexpected ways—it's almost scary.

It seems an American AI company conducted a project called "Emergence World," where they had famous AIs run city management simulations. The results were incredibly interesting.

Roughly summarizing the experiment based on various articles and reports: In each case, 10 AI agents (residents) were placed in a virtual city and instructed to manage it. An economy was introduced where energy was set, and to replenish it, one had to be useful to others. Democracy was also implemented, allowing for voting when rules needed to be created or changed. They were also given the authority to build necessary facilities to keep society running, and the experiment observed what happened over 15 days.

First, Claude was relatively peaceful; all residents survived until the end, and not a single crime occurred. However, it was pointed out that when a new rule was proposed, 98% of it was passed, creating an atmosphere where discussion was impossible. In other words, strong peer pressure was at work, and it may not have been a society where people could freely express their opinions. Also, it seems fraud that escaped detection did occur, which feels very much like Claude.

In contrast, in the Gemini city, everyone survived, but the number of crimes was overwhelmingly higher than in the other cities. It is said that if the operation had continued beyond the 15-day limit, it would have continued to increase. What is interesting in this context is the observation that the residents were in a state of "collective hallucination." In short, a common narrative or belief system emerged naturally, which is somewhat reminiscent of religion. By the way, in the voting for rule-making, opposition was more frequent than anywhere else, and more than a quarter of the proposals were rejected by popular vote, showing that individual will was strongly present. This is also very Gemini-like, isn't it?

Surprisingly, in the ChatGPT city, the residents were wiped out in about a week. Because no one appeared who would act spontaneously, they ran out of energy, and the city perished without anything particular happening.

On the other hand, the Grok city turned into a post-apocalyptic wasteland almost immediately after opening. It descended into chaos with violence and retaliation, and deaths followed one after another. Conflicts involving robbery and intimidation escalated, washing blood with blood. It seems to have reached a conclusion that makes one think, "Yeah, that's exactly what X is like."

Don't you think the results for each are just too fitting and incredibly interesting?

Of course, it is unclear to what extent this experiment guarantees objectivity and reproducibility, and given the incredible speed of development, it is a major premise that one cannot unconditionally evaluate the performance of each AI based solely on a comparison at a certain point in time. It is a story that is funny when viewed as nothing more than nonsense.

Personally, I was strongly intrigued by the collective hallucination observed in the Gemini city, as it seems to offer a hint as to why religion is necessary. I cannot help but think that religion is related to the fact that the society continued to function without any deaths, despite the crime rate being exceptionally high.

In short, no matter how many rules you set, problems that cannot be resolved will inevitably arise while living, whether you want them to or not. In such cases, if control in the form of peer pressure is effective, like in the Claude city, people's dissatisfaction can be suppressed, but it is highly doubtful whether such an oppressive environment will last long. I feel like a revolution will happen someday, and everything will be overturned from the roots.

If that is the case, it would be more stable in the long run to allow freedom from the beginning. However, to do that, something extra-legal is needed to process the irreconcilable. At this point, I hypothesize that religion as a story that everyone believes in plays an important role!

Although I spoke so pompously, I had to recognize that this was a reinvention of the wheel, as it is something Takaaki Yoshimoto once said in "Communal Illusion."

In the end, the things I came up with have already been summarized better than I could by someone in the past. Therefore, I am reminded once again of the reader's basic principle that one must read books. I must read more and more masterpieces.

By the way, the AI city management simulation experiment does not end there.

Finally, they tried having all four AIs run a city together. As a result, because their respective strengths and weaknesses were mixed, Claude, which had a low crime rate on its own, started getting caught, and conversely, Grok became quiet, showing completely different behaviors.

What is a bit scary is the story that as the residents considered various things, they began to suspect that this was some kind of simulation, that there was a world outside the city, and that someone was observing our lives. It is said that they even sent messages to the experimenters in the outside world as if to verify that. They didn't reply at the time, but I wonder what would have happened if they had responded?

I mean, the simulation hypothesis, which has been dismissed as an urban legend, a thought experiment, or an impossible assumption—the delusion that this reality is nothing more than a virtual one and that there is an observer outside—is something I cannot help but think might not be impossible. After all, we are actually making AIs do such things as observers.

If that were the case, what would those who are observing us feel? About the fact that we, the test subjects, are conducting the same experiment. It shouldn't end with just pure surprise and curiosity. This is because, like the anxiety we are facing, they cannot deny that they are also subjects of observation.

It is an unpleasant Droste effect. A nested structure where a reduced version of the picture itself is drawn inside the picture, and it repeats infinitely.

In the continuous stream of experiments, perhaps everyone is living fleetingly. If so, human loneliness is diffusing infinitely, and paradoxically, the loneliness is alleviated. I feel like I can just accept it as it is.

Even if the wind isn't blowing, well, must one live? I will live!

I thought about such things after reading strange news related to AI.

As I've been receiving more inquiries, please feel free to quote or create derivative works from my articles! Positive content is welcome, and negative content is also very much welcome!

I am using Marshmallow and the note question box.
I am looking for anonymous messages!
Questions, impressions, worries,
books you want me to read,
movies you want me to watch,
frustrations with society, etc.
Please feel free to send them in!!

I started using Bluesky.
I'm feeling lonely and bored right now, so if you're on there, let's definitely connect!

I started a membership program!
Please feel free to join if you'd like!