With the rise of ChatGPT type of machine learning technology, it opens the door for post Shannon communications, namely, the effectiveness communications. The purpose of communications has been generalized into transmission of the intend-intelligent, rather than information bits in the form of the text or sensory data. The goal of the effectiveness communications is to transmit the intend generated from machine learning such that, at the receive-end, to generate the desired generative intelligence and associated actions to achieve the goal of communications. In this talk, we resent a framework of machine-to-machine communications based on the generated pre-trained model, for example in the usage scenario of cloud-driven robots, we also present a new distributed wireless network computing architecture for 6G to support the “data-driven” AI or even AGI services and applications in 2030-2050 time-frame.