statistical functions, etc. in the

[rifattryo.ut11]

The article "Turing Test" mentioned a simple test case. Similar ideas can be extended to different roles and models. In this case, a similar leap is needed, not the progress of the previous one, but to make the current intelligence have another leap. This is actually the same as what Zhu Xiaohu mentioned about the core of seeing when it will come out. If this part can be upgraded, the model's own subscription income will increase and the potential impact space will also be larger. Even if it jumps to a higher level, if there is no rise of new applications, referring to the business model imagined in the past, it will still be a dead end.



The second disaster: Can I iv applications be released? If japan cell number I is not used to make applications that have not been made in the past, but is entangled in the past model, there is no way out. Is it not the case that I can be called based on the model? The previous wave of I made solutions with unbalanced income and expenditure. Will the income and expenditure be balanced if it is replaced by a large model model? Is it because the technology is not good that the solution cannot support the I company's business closed loop? If it is not simply a change of a set of technical terms, why can the old model suddenly work?If not, then it won’t work.



From the end point of view, the super app has made the entrance very tight. What new problem can you solve by making an assistant based on I? If you simply aggregate other people’s content, Doubao can also aggregate Douyin. Why can you aggregate it? Which of the existing I, e-commerce, and payment can be subverted by I? So it’s impossible. This has led to divisions and everyone has chosen sides, but in fact, it’s useless to choose. No matter how you look at it, it’s a dead end. In fact, it’s not like this.