[kokhanserhii]

I think adding the word "self-improvement" would be sufficient, and today the emphasis will be less on "improvement" and more on "self-improvement," because today humans are better at improving "smart chats" and robots through "theories, trial, and error" than systems are at improving themselves. A significant step forward would be the use of the phrase "architectural self-improvement." Of course, there are different approaches and different understandings of "improvement" here. I don't advocate predictions of any kind of "explosive self-improvement" in artificial intelligence systems. It's obvious. Just look at how people improve them... As improvements progress, phrases like "we need $10 billion" and "we need a trillion dollars" periodically appear. Some people mistakenly believe that self-improvement will be free. Yes, it will be free, but the scale of improvement will be limited. Nowadays, huge sums of money are spent on experiments for improvement, not for the sake of the reasonably expected improvements themselves, but for the sake of the race for leadership in the global market.

[kokhanserhii]

It's important to understand that intelligent chats possess a vast amount of human scientific knowledge and are well-versed in areas that are important and promising for people, in vast quantities and volumes. Perhaps, in light of this, our understanding of self-improvement should be slightly adjusted?