Now about "don't know how to program" this is where it gets interesting. Being a "developer of value" implies having a developer mindset at the very least, which implies existence of at least some minimal knowledge and at least minimal effort for learning something new.

The most interesting phenomenon for me today is how there are all these vibecoded products popping up, and they have very obvious glaring bugs and issues that are NOT hard to fix at all. All you need is to simply use your own product and pay attention to what is broken, and you need at least surface level understanding of what you do. Which is, again, not hard to learn. Just read a book or two.

With AI bros, asking them to read a book is too much. They refuse to apply even minimal effort into developing something. They have no future vision of what they do, their strategy is dump the turd and move on. Sub-Saharan. They would create giga-complex AI setups with hundreds of agents burning billions of tokens per hour in hopes "AI will figure it out" instead of learning something themselves and telling their agent what to do.
They have quasi-religious relationship with AI where they insist that AI will wipe your ass for you. But AI is a probability engine, it needs to be pointed to the correct approach for probabilities to work in your favor.

Which is, again, quite interesting, and it creates a lot of questions with no definitive answers. Where do these people come from? Did AI naturally attract hordes of mouth-breathers who are too braindead to complete a single small project? Is the entire pro-AI field just them? Is it AI being bad, or is it about the retards who operate the machines? Is it AI making them braindead or are they natural? Do properly working AI assisted projects even exist? Most importantly, do there exist people who successfully synthesize AI knowledge with the human mind, or is the AI slop doomed to be forever unusable slop that does nothing but create informational noise and destruction of value?