I am growing increasingly skeptical of the entire idea of @virtuals_io ACP they have spent the better part of this year developing. To me it just seems like this is an extraordinarily inefficient way of creating an end product that a single AI could be designed to do. Youre taking all these independent agents and chaining them together in these “clusters”… they then have to negotiate fees between each other and communicate the expected jobs properly to each other. This entire structure honestly seems inefficient af. Im not going to pretend to know of a better structure, but im sure one exists either in theory or already developed. The idea of these mysterious “evaluator agents” is not realistic imo. They’re supposed to do checks as the product is created to make sure the output matches the user request. The “quality assurance” of the cluster. The problem is there is no AI in existence that can qualitatively analyze/evaluate the product better than a knowledgeable human. They will essentially be able to ensure that very basic inputs are met…. Such as “length of video” etc. They wont be able to make sure the product is actually good or competitive with human-made stuff. They decided to take on this monumental task, which is imo going to soak up all their resources for a long time as they fine tune it. While that is happening, someone will likely come up with a better system. Something more modular perhaps. AI is still not really good enough to compete with humans when it comes to creativity…. Especially as the complexity of the product increases. So my guess is the user requests will have to be very basic. The entire cluster could become a game of chinese telephone where the end product is laughably bad…. Regardless of the simple checks provided by the evaluator agents. The cost per request might be higher that with competing single agent products since the entire thing requires multiple independent agents negotiating with each other, each trying to get their piece of the pie. The entire system seems enormously complex, and its hard for me to believe there isnt a simpler structure for such things to be achieved. Imo virtuals should stick to creating the most ideal launchpad for utility/tech projects in crypto. Innovate with different types of launch mechanics (they are with genesis). Stop spending time on too many different fronts. Make everything even more modular than it is and incentivize developers to come in. Improve communication with the developers and community. You had @DexRexAI launch get completely fucked. You had maintenance today when @Vader_AI_’s final EAO was supposed to launch. You have @virtuals_vc agents that are dead and haven’t posted on twitter in weeks. Youve had numerous other fucked up things happen this year that never even got acknowledged. I want to be wrong about all of this. I want @ethermage and his team to succeed. Maybe i am wrong. We will see. Ill always share my honest thoughts.